{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 31803,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 33.329856872558594,
      "learning_rate": 4.184100418410042e-08,
      "loss": 2.2297,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 31.511091232299805,
      "learning_rate": 8.368200836820084e-08,
      "loss": 2.2343,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 34.114253997802734,
      "learning_rate": 1.2552301255230126e-07,
      "loss": 2.2307,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 36.17031478881836,
      "learning_rate": 1.6736401673640168e-07,
      "loss": 2.2117,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 31.016225814819336,
      "learning_rate": 2.092050209205021e-07,
      "loss": 2.2382,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 61.1959114074707,
      "learning_rate": 2.5104602510460253e-07,
      "loss": 2.2331,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 29.927688598632812,
      "learning_rate": 2.9288702928870297e-07,
      "loss": 2.1966,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 31.241943359375,
      "learning_rate": 3.3472803347280335e-07,
      "loss": 2.2997,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 36.80055236816406,
      "learning_rate": 3.765690376569038e-07,
      "loss": 2.2102,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 31.098207473754883,
      "learning_rate": 4.184100418410042e-07,
      "loss": 2.2303,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 32.90201950073242,
      "learning_rate": 4.6025104602510467e-07,
      "loss": 2.1998,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 29.618532180786133,
      "learning_rate": 5.020920502092051e-07,
      "loss": 2.1713,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 31.947437286376953,
      "learning_rate": 5.439330543933055e-07,
      "loss": 2.2329,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 32.617088317871094,
      "learning_rate": 5.857740585774059e-07,
      "loss": 2.2253,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 29.32814598083496,
      "learning_rate": 6.276150627615063e-07,
      "loss": 2.1654,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 24.30267333984375,
      "learning_rate": 6.694560669456067e-07,
      "loss": 2.167,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 23.019336700439453,
      "learning_rate": 7.112970711297073e-07,
      "loss": 2.1873,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 23.711467742919922,
      "learning_rate": 7.531380753138076e-07,
      "loss": 2.1332,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 23.006345748901367,
      "learning_rate": 7.94979079497908e-07,
      "loss": 2.1772,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 22.298721313476562,
      "learning_rate": 8.368200836820084e-07,
      "loss": 2.1336,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.456756591796875,
      "learning_rate": 8.786610878661088e-07,
      "loss": 1.9715,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.23736572265625,
      "learning_rate": 9.205020920502093e-07,
      "loss": 2.004,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 15.685251235961914,
      "learning_rate": 9.623430962343098e-07,
      "loss": 2.0285,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.311368942260742,
      "learning_rate": 1.0041841004184101e-06,
      "loss": 2.0019,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.19590187072754,
      "learning_rate": 1.0460251046025104e-06,
      "loss": 1.9712,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.604945182800293,
      "learning_rate": 1.087866108786611e-06,
      "loss": 2.0914,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.54394245147705,
      "learning_rate": 1.1297071129707113e-06,
      "loss": 1.9831,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.704313278198242,
      "learning_rate": 1.1715481171548119e-06,
      "loss": 1.9213,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.45096492767334,
      "learning_rate": 1.2133891213389122e-06,
      "loss": 1.824,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.20331859588623,
      "learning_rate": 1.2552301255230125e-06,
      "loss": 1.8047,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.640190124511719,
      "learning_rate": 1.297071129707113e-06,
      "loss": 1.7843,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.381649971008301,
      "learning_rate": 1.3389121338912134e-06,
      "loss": 0.9798,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.196405410766602,
      "learning_rate": 1.380753138075314e-06,
      "loss": 1.8094,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.818713665008545,
      "learning_rate": 1.4225941422594145e-06,
      "loss": 1.802,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.449642658233643,
      "learning_rate": 1.4644351464435146e-06,
      "loss": 1.8196,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.172857284545898,
      "learning_rate": 1.5062761506276152e-06,
      "loss": 1.7752,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.943195343017578,
      "learning_rate": 1.5481171548117155e-06,
      "loss": 1.7778,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.04423713684082,
      "learning_rate": 1.589958158995816e-06,
      "loss": 1.7417,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.931229114532471,
      "learning_rate": 1.6317991631799166e-06,
      "loss": 1.7252,
      "step": 39
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.2164626121521,
      "learning_rate": 1.6736401673640167e-06,
      "loss": 1.6952,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.581117630004883,
      "learning_rate": 1.7154811715481173e-06,
      "loss": 1.7383,
      "step": 41
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.623687744140625,
      "learning_rate": 1.7573221757322176e-06,
      "loss": 1.7125,
      "step": 42
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.528269290924072,
      "learning_rate": 1.7991631799163181e-06,
      "loss": 1.6656,
      "step": 43
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.202449798583984,
      "learning_rate": 1.8410041841004187e-06,
      "loss": 1.6861,
      "step": 44
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.643449068069458,
      "learning_rate": 1.8828451882845188e-06,
      "loss": 1.6975,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.403696060180664,
      "learning_rate": 1.9246861924686196e-06,
      "loss": 1.6924,
      "step": 46
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.579756736755371,
      "learning_rate": 1.96652719665272e-06,
      "loss": 1.6011,
      "step": 47
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.658903121948242,
      "learning_rate": 2.0083682008368202e-06,
      "loss": 1.6048,
      "step": 48
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.801294803619385,
      "learning_rate": 2.0502092050209206e-06,
      "loss": 1.6287,
      "step": 49
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5773539543151855,
      "learning_rate": 2.092050209205021e-06,
      "loss": 1.6386,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.93416690826416,
      "learning_rate": 2.1338912133891217e-06,
      "loss": 1.5994,
      "step": 51
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.359320640563965,
      "learning_rate": 2.175732217573222e-06,
      "loss": 1.564,
      "step": 52
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.9275460243225098,
      "learning_rate": 2.2175732217573223e-06,
      "loss": 1.5753,
      "step": 53
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.917264699935913,
      "learning_rate": 2.2594142259414227e-06,
      "loss": 1.5623,
      "step": 54
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.964813709259033,
      "learning_rate": 2.301255230125523e-06,
      "loss": 1.5214,
      "step": 55
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.04970645904541,
      "learning_rate": 2.3430962343096237e-06,
      "loss": 1.538,
      "step": 56
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.968022346496582,
      "learning_rate": 2.384937238493724e-06,
      "loss": 1.5859,
      "step": 57
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.084798812866211,
      "learning_rate": 2.4267782426778244e-06,
      "loss": 1.5063,
      "step": 58
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.508085250854492,
      "learning_rate": 2.4686192468619247e-06,
      "loss": 1.5548,
      "step": 59
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.676292419433594,
      "learning_rate": 2.510460251046025e-06,
      "loss": 1.4858,
      "step": 60
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.180440902709961,
      "learning_rate": 2.552301255230126e-06,
      "loss": 1.5659,
      "step": 61
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.0393025875091553,
      "learning_rate": 2.594142259414226e-06,
      "loss": 1.4459,
      "step": 62
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.495028972625732,
      "learning_rate": 2.635983263598327e-06,
      "loss": 1.5195,
      "step": 63
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.585282802581787,
      "learning_rate": 2.677824267782427e-06,
      "loss": 1.4703,
      "step": 64
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.742009162902832,
      "learning_rate": 2.719665271966527e-06,
      "loss": 1.4765,
      "step": 65
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.533092737197876,
      "learning_rate": 2.761506276150628e-06,
      "loss": 1.4872,
      "step": 66
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.3026609420776367,
      "learning_rate": 2.8033472803347283e-06,
      "loss": 1.4599,
      "step": 67
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.53637409210205,
      "learning_rate": 2.845188284518829e-06,
      "loss": 1.4627,
      "step": 68
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.103423118591309,
      "learning_rate": 2.887029288702929e-06,
      "loss": 1.4094,
      "step": 69
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.585911750793457,
      "learning_rate": 2.9288702928870293e-06,
      "loss": 1.4248,
      "step": 70
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.2304863929748535,
      "learning_rate": 2.97071129707113e-06,
      "loss": 1.4076,
      "step": 71
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.949634552001953,
      "learning_rate": 3.0125523012552303e-06,
      "loss": 1.4397,
      "step": 72
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.470211029052734,
      "learning_rate": 3.054393305439331e-06,
      "loss": 1.4211,
      "step": 73
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.9078080654144287,
      "learning_rate": 3.096234309623431e-06,
      "loss": 1.4589,
      "step": 74
    },
    {
      "epoch": 0.0,
      "grad_norm": 19.75151824951172,
      "learning_rate": 3.1380753138075313e-06,
      "loss": 1.3876,
      "step": 75
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.0558550357818604,
      "learning_rate": 3.179916317991632e-06,
      "loss": 1.4,
      "step": 76
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.677576541900635,
      "learning_rate": 3.2217573221757324e-06,
      "loss": 1.3917,
      "step": 77
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.998560428619385,
      "learning_rate": 3.263598326359833e-06,
      "loss": 1.3889,
      "step": 78
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.31503963470459,
      "learning_rate": 3.305439330543933e-06,
      "loss": 1.3574,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.28566575050354,
      "learning_rate": 3.3472803347280334e-06,
      "loss": 1.3694,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.405963897705078,
      "learning_rate": 3.389121338912134e-06,
      "loss": 1.3812,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.052747964859009,
      "learning_rate": 3.4309623430962345e-06,
      "loss": 1.3375,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.058832168579102,
      "learning_rate": 3.4728033472803353e-06,
      "loss": 1.337,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0704598426818848,
      "learning_rate": 3.514644351464435e-06,
      "loss": 1.366,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.829763174057007,
      "learning_rate": 3.5564853556485355e-06,
      "loss": 1.3604,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.326866865158081,
      "learning_rate": 3.5983263598326363e-06,
      "loss": 1.2915,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1212856769561768,
      "learning_rate": 3.6401673640167366e-06,
      "loss": 1.3248,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.041646957397461,
      "learning_rate": 3.6820083682008374e-06,
      "loss": 1.3413,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3949811458587646,
      "learning_rate": 3.7238493723849373e-06,
      "loss": 0.9819,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.6485109329223633,
      "learning_rate": 3.7656903765690376e-06,
      "loss": 1.3448,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.169336318969727,
      "learning_rate": 3.8075313807531384e-06,
      "loss": 1.3388,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.138458728790283,
      "learning_rate": 3.849372384937239e-06,
      "loss": 1.3119,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.635039329528809,
      "learning_rate": 3.8912133891213395e-06,
      "loss": 1.3189,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.414746046066284,
      "learning_rate": 3.93305439330544e-06,
      "loss": 1.2973,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.9652888774871826,
      "learning_rate": 3.97489539748954e-06,
      "loss": 1.3165,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.839945077896118,
      "learning_rate": 4.0167364016736405e-06,
      "loss": 1.317,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.80806565284729,
      "learning_rate": 4.058577405857741e-06,
      "loss": 1.2839,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0614490509033203,
      "learning_rate": 4.100418410041841e-06,
      "loss": 1.3339,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.4989848136901855,
      "learning_rate": 4.142259414225942e-06,
      "loss": 1.2821,
      "step": 99
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.6587111949920654,
      "learning_rate": 4.184100418410042e-06,
      "loss": 1.2692,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5256028175354004,
      "learning_rate": 4.225941422594142e-06,
      "loss": 1.336,
      "step": 101
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.138798475265503,
      "learning_rate": 4.267782426778243e-06,
      "loss": 1.3186,
      "step": 102
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4118640422821045,
      "learning_rate": 4.309623430962344e-06,
      "loss": 1.2789,
      "step": 103
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.340080261230469,
      "learning_rate": 4.351464435146444e-06,
      "loss": 1.2421,
      "step": 104
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.8364245891571045,
      "learning_rate": 4.393305439330544e-06,
      "loss": 1.3068,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.212522983551025,
      "learning_rate": 4.435146443514645e-06,
      "loss": 1.2907,
      "step": 106
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6237246990203857,
      "learning_rate": 4.476987447698745e-06,
      "loss": 1.3338,
      "step": 107
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.77582049369812,
      "learning_rate": 4.518828451882845e-06,
      "loss": 1.2948,
      "step": 108
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.335937261581421,
      "learning_rate": 4.5606694560669465e-06,
      "loss": 1.238,
      "step": 109
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.350453853607178,
      "learning_rate": 4.602510460251046e-06,
      "loss": 1.2424,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.058448553085327,
      "learning_rate": 4.644351464435146e-06,
      "loss": 1.2829,
      "step": 111
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.074069499969482,
      "learning_rate": 4.6861924686192475e-06,
      "loss": 1.2208,
      "step": 112
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.18929386138916,
      "learning_rate": 4.728033472803348e-06,
      "loss": 1.2727,
      "step": 113
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.333191871643066,
      "learning_rate": 4.769874476987448e-06,
      "loss": 1.2182,
      "step": 114
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.443491220474243,
      "learning_rate": 4.8117154811715485e-06,
      "loss": 1.2079,
      "step": 115
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.217432022094727,
      "learning_rate": 4.853556485355649e-06,
      "loss": 1.2323,
      "step": 116
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.847862482070923,
      "learning_rate": 4.895397489539749e-06,
      "loss": 1.2343,
      "step": 117
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.392519474029541,
      "learning_rate": 4.9372384937238495e-06,
      "loss": 1.2189,
      "step": 118
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.7109591960906982,
      "learning_rate": 4.979079497907951e-06,
      "loss": 1.2289,
      "step": 119
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0395877361297607,
      "learning_rate": 5.02092050209205e-06,
      "loss": 1.2101,
      "step": 120
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.242081165313721,
      "learning_rate": 5.062761506276151e-06,
      "loss": 1.2341,
      "step": 121
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.337377071380615,
      "learning_rate": 5.104602510460252e-06,
      "loss": 1.2211,
      "step": 122
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.005611896514893,
      "learning_rate": 5.146443514644351e-06,
      "loss": 1.2256,
      "step": 123
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.108304977416992,
      "learning_rate": 5.188284518828452e-06,
      "loss": 1.1958,
      "step": 124
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.426984786987305,
      "learning_rate": 5.230125523012553e-06,
      "loss": 1.2024,
      "step": 125
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.062593460083008,
      "learning_rate": 5.271966527196654e-06,
      "loss": 1.2266,
      "step": 126
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.9947845935821533,
      "learning_rate": 5.313807531380753e-06,
      "loss": 1.2012,
      "step": 127
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.604424238204956,
      "learning_rate": 5.355648535564854e-06,
      "loss": 1.2156,
      "step": 128
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.744723081588745,
      "learning_rate": 5.397489539748955e-06,
      "loss": 1.1692,
      "step": 129
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.472827196121216,
      "learning_rate": 5.439330543933054e-06,
      "loss": 1.2521,
      "step": 130
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.85957670211792,
      "learning_rate": 5.4811715481171555e-06,
      "loss": 1.2216,
      "step": 131
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.257476806640625,
      "learning_rate": 5.523012552301256e-06,
      "loss": 1.1928,
      "step": 132
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2176401615142822,
      "learning_rate": 5.564853556485355e-06,
      "loss": 1.1792,
      "step": 133
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.935103893280029,
      "learning_rate": 5.6066945606694565e-06,
      "loss": 1.1879,
      "step": 134
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.3367156982421875,
      "learning_rate": 5.648535564853557e-06,
      "loss": 1.1783,
      "step": 135
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.02912712097168,
      "learning_rate": 5.690376569037658e-06,
      "loss": 1.2101,
      "step": 136
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0572192668914795,
      "learning_rate": 5.7322175732217575e-06,
      "loss": 1.1878,
      "step": 137
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.631429433822632,
      "learning_rate": 5.774058577405858e-06,
      "loss": 1.191,
      "step": 138
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.007631778717041,
      "learning_rate": 5.815899581589959e-06,
      "loss": 1.1819,
      "step": 139
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.287233352661133,
      "learning_rate": 5.8577405857740585e-06,
      "loss": 1.1639,
      "step": 140
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.682846784591675,
      "learning_rate": 5.89958158995816e-06,
      "loss": 1.147,
      "step": 141
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.5044989585876465,
      "learning_rate": 5.94142259414226e-06,
      "loss": 1.2127,
      "step": 142
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7548468112945557,
      "learning_rate": 5.9832635983263595e-06,
      "loss": 1.2133,
      "step": 143
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.72615385055542,
      "learning_rate": 6.025104602510461e-06,
      "loss": 1.1479,
      "step": 144
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.613144874572754,
      "learning_rate": 6.066945606694561e-06,
      "loss": 1.1472,
      "step": 145
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.090764999389648,
      "learning_rate": 6.108786610878662e-06,
      "loss": 1.1371,
      "step": 146
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.5506157875061035,
      "learning_rate": 6.150627615062762e-06,
      "loss": 1.1825,
      "step": 147
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.018399715423584,
      "learning_rate": 6.192468619246862e-06,
      "loss": 1.1353,
      "step": 148
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.420623302459717,
      "learning_rate": 6.234309623430963e-06,
      "loss": 1.187,
      "step": 149
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0519158840179443,
      "learning_rate": 6.276150627615063e-06,
      "loss": 1.1792,
      "step": 150
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.577958106994629,
      "learning_rate": 6.317991631799164e-06,
      "loss": 1.1853,
      "step": 151
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6326589584350586,
      "learning_rate": 6.359832635983264e-06,
      "loss": 1.1997,
      "step": 152
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.038844108581543,
      "learning_rate": 6.401673640167364e-06,
      "loss": 1.1602,
      "step": 153
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7656564712524414,
      "learning_rate": 6.443514644351465e-06,
      "loss": 1.1567,
      "step": 154
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.306248903274536,
      "learning_rate": 6.485355648535565e-06,
      "loss": 1.116,
      "step": 155
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.624532461166382,
      "learning_rate": 6.527196652719666e-06,
      "loss": 1.1689,
      "step": 156
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.435182571411133,
      "learning_rate": 6.569037656903766e-06,
      "loss": 1.1523,
      "step": 157
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.64935827255249,
      "learning_rate": 6.610878661087866e-06,
      "loss": 1.1247,
      "step": 158
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5091290473937988,
      "learning_rate": 6.652719665271967e-06,
      "loss": 0.9198,
      "step": 159
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0998950004577637,
      "learning_rate": 6.694560669456067e-06,
      "loss": 1.1519,
      "step": 160
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.986159086227417,
      "learning_rate": 6.736401673640168e-06,
      "loss": 1.203,
      "step": 161
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8588573932647705,
      "learning_rate": 6.778242677824268e-06,
      "loss": 1.1076,
      "step": 162
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.6413509845733643,
      "learning_rate": 6.820083682008368e-06,
      "loss": 1.1527,
      "step": 163
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.6915688514709473,
      "learning_rate": 6.861924686192469e-06,
      "loss": 1.1201,
      "step": 164
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.759608507156372,
      "learning_rate": 6.903765690376569e-06,
      "loss": 0.9362,
      "step": 165
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8459343910217285,
      "learning_rate": 6.9456066945606706e-06,
      "loss": 1.1125,
      "step": 166
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.053657054901123,
      "learning_rate": 6.98744769874477e-06,
      "loss": 1.1186,
      "step": 167
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.7361278533935547,
      "learning_rate": 7.02928870292887e-06,
      "loss": 1.1432,
      "step": 168
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.999213695526123,
      "learning_rate": 7.0711297071129716e-06,
      "loss": 1.1109,
      "step": 169
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.113543510437012,
      "learning_rate": 7.112970711297071e-06,
      "loss": 1.1791,
      "step": 170
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.7114107608795166,
      "learning_rate": 7.154811715481172e-06,
      "loss": 1.1001,
      "step": 171
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4712624549865723,
      "learning_rate": 7.1966527196652726e-06,
      "loss": 1.1114,
      "step": 172
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.684647798538208,
      "learning_rate": 7.238493723849372e-06,
      "loss": 1.0726,
      "step": 173
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7734146118164062,
      "learning_rate": 7.280334728033473e-06,
      "loss": 1.13,
      "step": 174
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4748003482818604,
      "learning_rate": 7.3221757322175736e-06,
      "loss": 1.0707,
      "step": 175
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.184093475341797,
      "learning_rate": 7.364016736401675e-06,
      "loss": 0.9493,
      "step": 176
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.976660251617432,
      "learning_rate": 7.405857740585774e-06,
      "loss": 1.1513,
      "step": 177
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.485933303833008,
      "learning_rate": 7.4476987447698746e-06,
      "loss": 1.0969,
      "step": 178
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5350301265716553,
      "learning_rate": 7.489539748953976e-06,
      "loss": 1.0949,
      "step": 179
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2504935264587402,
      "learning_rate": 7.531380753138075e-06,
      "loss": 1.1515,
      "step": 180
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.001731872558594,
      "learning_rate": 7.573221757322176e-06,
      "loss": 1.0837,
      "step": 181
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.908815860748291,
      "learning_rate": 7.615062761506277e-06,
      "loss": 1.1098,
      "step": 182
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.1650795936584473,
      "learning_rate": 7.656903765690377e-06,
      "loss": 1.0953,
      "step": 183
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2260961532592773,
      "learning_rate": 7.698744769874478e-06,
      "loss": 1.06,
      "step": 184
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.615259885787964,
      "learning_rate": 7.740585774058578e-06,
      "loss": 1.0675,
      "step": 185
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.357839345932007,
      "learning_rate": 7.782426778242679e-06,
      "loss": 1.1096,
      "step": 186
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2356116771698,
      "learning_rate": 7.824267782426778e-06,
      "loss": 1.1145,
      "step": 187
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5119173526763916,
      "learning_rate": 7.86610878661088e-06,
      "loss": 1.0802,
      "step": 188
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.93316125869751,
      "learning_rate": 7.907949790794979e-06,
      "loss": 1.1123,
      "step": 189
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0717971324920654,
      "learning_rate": 7.94979079497908e-06,
      "loss": 1.1025,
      "step": 190
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8000950813293457,
      "learning_rate": 7.991631799163181e-06,
      "loss": 1.0927,
      "step": 191
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.2984209060668945,
      "learning_rate": 8.033472803347281e-06,
      "loss": 1.0446,
      "step": 192
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.743616819381714,
      "learning_rate": 8.075313807531382e-06,
      "loss": 1.0755,
      "step": 193
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.8698389530181885,
      "learning_rate": 8.117154811715482e-06,
      "loss": 1.104,
      "step": 194
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.841881513595581,
      "learning_rate": 8.158995815899581e-06,
      "loss": 1.0971,
      "step": 195
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.94660758972168,
      "learning_rate": 8.200836820083682e-06,
      "loss": 1.0675,
      "step": 196
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.665743589401245,
      "learning_rate": 8.242677824267783e-06,
      "loss": 1.0827,
      "step": 197
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.9379169940948486,
      "learning_rate": 8.284518828451885e-06,
      "loss": 1.0946,
      "step": 198
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.415921926498413,
      "learning_rate": 8.326359832635984e-06,
      "loss": 1.0642,
      "step": 199
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.1108481884002686,
      "learning_rate": 8.368200836820084e-06,
      "loss": 1.0709,
      "step": 200
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2757818698883057,
      "learning_rate": 8.410041841004185e-06,
      "loss": 1.0931,
      "step": 201
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.382082939147949,
      "learning_rate": 8.451882845188284e-06,
      "loss": 1.0051,
      "step": 202
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.362422466278076,
      "learning_rate": 8.493723849372385e-06,
      "loss": 1.0486,
      "step": 203
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.3258774280548096,
      "learning_rate": 8.535564853556487e-06,
      "loss": 1.1125,
      "step": 204
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.334695816040039,
      "learning_rate": 8.577405857740586e-06,
      "loss": 1.0742,
      "step": 205
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.9175634384155273,
      "learning_rate": 8.619246861924687e-06,
      "loss": 1.0803,
      "step": 206
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.96854305267334,
      "learning_rate": 8.661087866108787e-06,
      "loss": 1.0632,
      "step": 207
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7646377086639404,
      "learning_rate": 8.702928870292888e-06,
      "loss": 1.1031,
      "step": 208
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2340638637542725,
      "learning_rate": 8.744769874476987e-06,
      "loss": 1.0507,
      "step": 209
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.701727867126465,
      "learning_rate": 8.786610878661089e-06,
      "loss": 1.0428,
      "step": 210
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.139406681060791,
      "learning_rate": 8.82845188284519e-06,
      "loss": 1.0574,
      "step": 211
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6874516010284424,
      "learning_rate": 8.87029288702929e-06,
      "loss": 1.0631,
      "step": 212
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.087262153625488,
      "learning_rate": 8.91213389121339e-06,
      "loss": 1.0296,
      "step": 213
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.5974628925323486,
      "learning_rate": 8.95397489539749e-06,
      "loss": 1.0404,
      "step": 214
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6592392921447754,
      "learning_rate": 8.995815899581591e-06,
      "loss": 1.0513,
      "step": 215
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.1334409713745117,
      "learning_rate": 9.03765690376569e-06,
      "loss": 1.0716,
      "step": 216
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4860122203826904,
      "learning_rate": 9.079497907949792e-06,
      "loss": 1.014,
      "step": 217
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.296205043792725,
      "learning_rate": 9.121338912133893e-06,
      "loss": 1.0297,
      "step": 218
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.654794454574585,
      "learning_rate": 9.163179916317992e-06,
      "loss": 1.0634,
      "step": 219
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2576541900634766,
      "learning_rate": 9.205020920502092e-06,
      "loss": 1.0396,
      "step": 220
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4390573501586914,
      "learning_rate": 9.246861924686193e-06,
      "loss": 1.0597,
      "step": 221
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.8543221950531006,
      "learning_rate": 9.288702928870293e-06,
      "loss": 1.0366,
      "step": 222
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5076091289520264,
      "learning_rate": 9.330543933054394e-06,
      "loss": 1.0253,
      "step": 223
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.5822222232818604,
      "learning_rate": 9.372384937238495e-06,
      "loss": 0.9996,
      "step": 224
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.672956705093384,
      "learning_rate": 9.414225941422594e-06,
      "loss": 1.0231,
      "step": 225
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.24424409866333,
      "learning_rate": 9.456066945606696e-06,
      "loss": 1.0719,
      "step": 226
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.323415756225586,
      "learning_rate": 9.497907949790795e-06,
      "loss": 1.045,
      "step": 227
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.418738842010498,
      "learning_rate": 9.539748953974896e-06,
      "loss": 1.0107,
      "step": 228
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.42337703704834,
      "learning_rate": 9.581589958158996e-06,
      "loss": 1.0573,
      "step": 229
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6549816131591797,
      "learning_rate": 9.623430962343097e-06,
      "loss": 0.9875,
      "step": 230
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.384407043457031,
      "learning_rate": 9.665271966527198e-06,
      "loss": 1.0216,
      "step": 231
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0144898891448975,
      "learning_rate": 9.707112970711298e-06,
      "loss": 1.0487,
      "step": 232
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.0239481925964355,
      "learning_rate": 9.748953974895399e-06,
      "loss": 1.0502,
      "step": 233
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.159718036651611,
      "learning_rate": 9.790794979079498e-06,
      "loss": 1.0492,
      "step": 234
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.897611379623413,
      "learning_rate": 9.8326359832636e-06,
      "loss": 1.0246,
      "step": 235
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.402046203613281,
      "learning_rate": 9.874476987447699e-06,
      "loss": 0.997,
      "step": 236
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8004987239837646,
      "learning_rate": 9.9163179916318e-06,
      "loss": 1.0296,
      "step": 237
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8607256412506104,
      "learning_rate": 9.958158995815901e-06,
      "loss": 1.0288,
      "step": 238
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.476633071899414,
      "learning_rate": 1e-05,
      "loss": 0.9884,
      "step": 239
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2999916076660156,
      "learning_rate": 1.00418410041841e-05,
      "loss": 1.0288,
      "step": 240
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.8279783725738525,
      "learning_rate": 1.0083682008368201e-05,
      "loss": 1.0086,
      "step": 241
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.8002545833587646,
      "learning_rate": 1.0125523012552303e-05,
      "loss": 1.0,
      "step": 242
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.657150983810425,
      "learning_rate": 1.0167364016736402e-05,
      "loss": 0.9929,
      "step": 243
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.898716688156128,
      "learning_rate": 1.0209205020920503e-05,
      "loss": 1.0459,
      "step": 244
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1322028636932373,
      "learning_rate": 1.0251046025104603e-05,
      "loss": 0.9222,
      "step": 245
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0539956092834473,
      "learning_rate": 1.0292887029288702e-05,
      "loss": 1.0236,
      "step": 246
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.1758508682250977,
      "learning_rate": 1.0334728033472805e-05,
      "loss": 1.0462,
      "step": 247
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0720245838165283,
      "learning_rate": 1.0376569037656905e-05,
      "loss": 1.0436,
      "step": 248
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.047100782394409,
      "learning_rate": 1.0418410041841006e-05,
      "loss": 1.0237,
      "step": 249
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7972495555877686,
      "learning_rate": 1.0460251046025105e-05,
      "loss": 1.025,
      "step": 250
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.9470434188842773,
      "learning_rate": 1.0502092050209205e-05,
      "loss": 1.02,
      "step": 251
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.8121418952941895,
      "learning_rate": 1.0543933054393308e-05,
      "loss": 1.0121,
      "step": 252
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.094548225402832,
      "learning_rate": 1.0585774058577407e-05,
      "loss": 1.0302,
      "step": 253
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.908203125,
      "learning_rate": 1.0627615062761507e-05,
      "loss": 1.028,
      "step": 254
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.114875316619873,
      "learning_rate": 1.0669456066945608e-05,
      "loss": 1.0239,
      "step": 255
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.599640130996704,
      "learning_rate": 1.0711297071129707e-05,
      "loss": 1.0062,
      "step": 256
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.608515739440918,
      "learning_rate": 1.0753138075313809e-05,
      "loss": 0.9745,
      "step": 257
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.420987606048584,
      "learning_rate": 1.079497907949791e-05,
      "loss": 0.9881,
      "step": 258
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.48219895362854,
      "learning_rate": 1.083682008368201e-05,
      "loss": 1.0243,
      "step": 259
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.877729654312134,
      "learning_rate": 1.0878661087866109e-05,
      "loss": 1.0284,
      "step": 260
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.433801174163818,
      "learning_rate": 1.092050209205021e-05,
      "loss": 0.9994,
      "step": 261
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.003145694732666,
      "learning_rate": 1.0962343096234311e-05,
      "loss": 1.0238,
      "step": 262
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7207024097442627,
      "learning_rate": 1.100418410041841e-05,
      "loss": 1.0785,
      "step": 263
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.653914213180542,
      "learning_rate": 1.1046025104602512e-05,
      "loss": 1.042,
      "step": 264
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.619702696800232,
      "learning_rate": 1.1087866108786611e-05,
      "loss": 0.9344,
      "step": 265
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.6602845191955566,
      "learning_rate": 1.112970711297071e-05,
      "loss": 1.0215,
      "step": 266
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.267752647399902,
      "learning_rate": 1.1171548117154814e-05,
      "loss": 0.9771,
      "step": 267
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.064958095550537,
      "learning_rate": 1.1213389121338913e-05,
      "loss": 1.0331,
      "step": 268
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.713857412338257,
      "learning_rate": 1.1255230125523014e-05,
      "loss": 1.0044,
      "step": 269
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0844781398773193,
      "learning_rate": 1.1297071129707114e-05,
      "loss": 0.9625,
      "step": 270
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.873843193054199,
      "learning_rate": 1.1338912133891213e-05,
      "loss": 0.9928,
      "step": 271
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.597048282623291,
      "learning_rate": 1.1380753138075316e-05,
      "loss": 1.0272,
      "step": 272
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7630624771118164,
      "learning_rate": 1.1422594142259416e-05,
      "loss": 0.984,
      "step": 273
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.71529483795166,
      "learning_rate": 1.1464435146443515e-05,
      "loss": 0.9802,
      "step": 274
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.8670408725738525,
      "learning_rate": 1.1506276150627616e-05,
      "loss": 0.9993,
      "step": 275
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.139448165893555,
      "learning_rate": 1.1548117154811716e-05,
      "loss": 0.9772,
      "step": 276
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0115442276000977,
      "learning_rate": 1.1589958158995817e-05,
      "loss": 1.0405,
      "step": 277
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7485203742980957,
      "learning_rate": 1.1631799163179918e-05,
      "loss": 1.0361,
      "step": 278
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.929211378097534,
      "learning_rate": 1.1673640167364018e-05,
      "loss": 1.0062,
      "step": 279
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.549943447113037,
      "learning_rate": 1.1715481171548117e-05,
      "loss": 0.9775,
      "step": 280
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.904848098754883,
      "learning_rate": 1.1757322175732218e-05,
      "loss": 0.9912,
      "step": 281
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6954424381256104,
      "learning_rate": 1.179916317991632e-05,
      "loss": 0.9437,
      "step": 282
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2467567920684814,
      "learning_rate": 1.184100418410042e-05,
      "loss": 1.0053,
      "step": 283
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6815707683563232,
      "learning_rate": 1.188284518828452e-05,
      "loss": 0.9743,
      "step": 284
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.362985849380493,
      "learning_rate": 1.192468619246862e-05,
      "loss": 0.959,
      "step": 285
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6890785694122314,
      "learning_rate": 1.1966527196652719e-05,
      "loss": 0.9593,
      "step": 286
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0175716876983643,
      "learning_rate": 1.2008368200836822e-05,
      "loss": 1.0173,
      "step": 287
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.473397731781006,
      "learning_rate": 1.2050209205020921e-05,
      "loss": 0.9963,
      "step": 288
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7308058738708496,
      "learning_rate": 1.2092050209205023e-05,
      "loss": 1.0184,
      "step": 289
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.831840753555298,
      "learning_rate": 1.2133891213389122e-05,
      "loss": 0.9423,
      "step": 290
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.2684295177459717,
      "learning_rate": 1.2175732217573222e-05,
      "loss": 0.9651,
      "step": 291
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0615792274475098,
      "learning_rate": 1.2217573221757324e-05,
      "loss": 0.9817,
      "step": 292
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.702976226806641,
      "learning_rate": 1.2259414225941424e-05,
      "loss": 0.9741,
      "step": 293
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.223897933959961,
      "learning_rate": 1.2301255230125523e-05,
      "loss": 0.9571,
      "step": 294
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.9992401599884033,
      "learning_rate": 1.2343096234309625e-05,
      "loss": 1.001,
      "step": 295
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.8689093589782715,
      "learning_rate": 1.2384937238493724e-05,
      "loss": 0.9527,
      "step": 296
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4085733890533447,
      "learning_rate": 1.2426778242677825e-05,
      "loss": 0.9611,
      "step": 297
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.405859470367432,
      "learning_rate": 1.2468619246861926e-05,
      "loss": 1.0264,
      "step": 298
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.3627395629882812,
      "learning_rate": 1.2510460251046026e-05,
      "loss": 0.9693,
      "step": 299
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.585907459259033,
      "learning_rate": 1.2552301255230125e-05,
      "loss": 1.0052,
      "step": 300
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.820676565170288,
      "learning_rate": 1.2594142259414227e-05,
      "loss": 0.9943,
      "step": 301
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.101377010345459,
      "learning_rate": 1.2635983263598328e-05,
      "loss": 0.9385,
      "step": 302
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.316331148147583,
      "learning_rate": 1.2677824267782429e-05,
      "loss": 0.9413,
      "step": 303
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5300638675689697,
      "learning_rate": 1.2719665271966528e-05,
      "loss": 1.0035,
      "step": 304
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6184823513031006,
      "learning_rate": 1.2761506276150628e-05,
      "loss": 0.9383,
      "step": 305
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.784156084060669,
      "learning_rate": 1.2803347280334727e-05,
      "loss": 0.9318,
      "step": 306
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.775285482406616,
      "learning_rate": 1.284518828451883e-05,
      "loss": 1.0176,
      "step": 307
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.417123556137085,
      "learning_rate": 1.288702928870293e-05,
      "loss": 0.9418,
      "step": 308
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.596451282501221,
      "learning_rate": 1.2928870292887031e-05,
      "loss": 1.0316,
      "step": 309
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0444324016571045,
      "learning_rate": 1.297071129707113e-05,
      "loss": 0.943,
      "step": 310
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1935083866119385,
      "learning_rate": 1.301255230125523e-05,
      "loss": 0.9231,
      "step": 311
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.248492956161499,
      "learning_rate": 1.3054393305439333e-05,
      "loss": 0.9579,
      "step": 312
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.273103713989258,
      "learning_rate": 1.3096234309623432e-05,
      "loss": 0.9861,
      "step": 313
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7242002487182617,
      "learning_rate": 1.3138075313807532e-05,
      "loss": 0.8872,
      "step": 314
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.9838356971740723,
      "learning_rate": 1.3179916317991633e-05,
      "loss": 0.9384,
      "step": 315
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.3858208656311035,
      "learning_rate": 1.3221757322175732e-05,
      "loss": 0.9566,
      "step": 316
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.7024989128112793,
      "learning_rate": 1.3263598326359835e-05,
      "loss": 0.9864,
      "step": 317
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.849533796310425,
      "learning_rate": 1.3305439330543935e-05,
      "loss": 0.943,
      "step": 318
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4558379650115967,
      "learning_rate": 1.3347280334728034e-05,
      "loss": 0.9531,
      "step": 319
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.6852450370788574,
      "learning_rate": 1.3389121338912134e-05,
      "loss": 0.9289,
      "step": 320
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.2742323875427246,
      "learning_rate": 1.3430962343096235e-05,
      "loss": 0.9422,
      "step": 321
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6482200622558594,
      "learning_rate": 1.3472803347280336e-05,
      "loss": 0.9857,
      "step": 322
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.1265716552734375,
      "learning_rate": 1.3514644351464437e-05,
      "loss": 0.9732,
      "step": 323
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7221920490264893,
      "learning_rate": 1.3556485355648537e-05,
      "loss": 0.9479,
      "step": 324
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.048187494277954,
      "learning_rate": 1.3598326359832636e-05,
      "loss": 0.9361,
      "step": 325
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.965358257293701,
      "learning_rate": 1.3640167364016736e-05,
      "loss": 0.9595,
      "step": 326
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.5422990322113037,
      "learning_rate": 1.3682008368200839e-05,
      "loss": 0.9665,
      "step": 327
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.38057804107666,
      "learning_rate": 1.3723849372384938e-05,
      "loss": 0.9693,
      "step": 328
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.398742437362671,
      "learning_rate": 1.376569037656904e-05,
      "loss": 0.9425,
      "step": 329
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.8455963134765625,
      "learning_rate": 1.3807531380753139e-05,
      "loss": 0.9164,
      "step": 330
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.653040647506714,
      "learning_rate": 1.3849372384937238e-05,
      "loss": 0.9354,
      "step": 331
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.1199381351470947,
      "learning_rate": 1.3891213389121341e-05,
      "loss": 0.9131,
      "step": 332
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.367114305496216,
      "learning_rate": 1.393305439330544e-05,
      "loss": 0.963,
      "step": 333
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7795064449310303,
      "learning_rate": 1.397489539748954e-05,
      "loss": 0.9279,
      "step": 334
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.087001323699951,
      "learning_rate": 1.4016736401673641e-05,
      "loss": 0.9618,
      "step": 335
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0728001594543457,
      "learning_rate": 1.405857740585774e-05,
      "loss": 0.9192,
      "step": 336
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7811949253082275,
      "learning_rate": 1.4100418410041844e-05,
      "loss": 0.9618,
      "step": 337
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.4157092571258545,
      "learning_rate": 1.4142259414225943e-05,
      "loss": 0.9753,
      "step": 338
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.1570897102355957,
      "learning_rate": 1.4184100418410043e-05,
      "loss": 0.9559,
      "step": 339
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.5689759254455566,
      "learning_rate": 1.4225941422594142e-05,
      "loss": 0.9088,
      "step": 340
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.4650776386260986,
      "learning_rate": 1.4267782426778243e-05,
      "loss": 0.8997,
      "step": 341
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.597356081008911,
      "learning_rate": 1.4309623430962344e-05,
      "loss": 0.9186,
      "step": 342
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.642559766769409,
      "learning_rate": 1.4351464435146446e-05,
      "loss": 0.9456,
      "step": 343
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.481351375579834,
      "learning_rate": 1.4393305439330545e-05,
      "loss": 0.9383,
      "step": 344
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.8062381744384766,
      "learning_rate": 1.4435146443514645e-05,
      "loss": 0.9236,
      "step": 345
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.194732904434204,
      "learning_rate": 1.4476987447698744e-05,
      "loss": 0.9059,
      "step": 346
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.565412998199463,
      "learning_rate": 1.4518828451882847e-05,
      "loss": 0.9391,
      "step": 347
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.161421775817871,
      "learning_rate": 1.4560669456066946e-05,
      "loss": 0.9199,
      "step": 348
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.82495641708374,
      "learning_rate": 1.4602510460251048e-05,
      "loss": 0.9396,
      "step": 349
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.4770851135253906,
      "learning_rate": 1.4644351464435147e-05,
      "loss": 0.9202,
      "step": 350
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.210707902908325,
      "learning_rate": 1.4686192468619247e-05,
      "loss": 0.952,
      "step": 351
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.1530308723449707,
      "learning_rate": 1.472803347280335e-05,
      "loss": 0.9492,
      "step": 352
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.942448616027832,
      "learning_rate": 1.4769874476987449e-05,
      "loss": 0.9293,
      "step": 353
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.586683511734009,
      "learning_rate": 1.4811715481171548e-05,
      "loss": 0.9418,
      "step": 354
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.832998514175415,
      "learning_rate": 1.485355648535565e-05,
      "loss": 0.9427,
      "step": 355
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.2683541774749756,
      "learning_rate": 1.4895397489539749e-05,
      "loss": 0.9165,
      "step": 356
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.8520824909210205,
      "learning_rate": 1.4937238493723852e-05,
      "loss": 0.9783,
      "step": 357
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.662947654724121,
      "learning_rate": 1.4979079497907951e-05,
      "loss": 0.9405,
      "step": 358
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.6258203983306885,
      "learning_rate": 1.5020920502092051e-05,
      "loss": 0.9354,
      "step": 359
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0246031284332275,
      "learning_rate": 1.506276150627615e-05,
      "loss": 0.9168,
      "step": 360
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.2681007385253906,
      "learning_rate": 1.5104602510460253e-05,
      "loss": 0.9065,
      "step": 361
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4554424285888672,
      "learning_rate": 1.5146443514644353e-05,
      "loss": 0.9507,
      "step": 362
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.512722492218018,
      "learning_rate": 1.5188284518828454e-05,
      "loss": 0.9419,
      "step": 363
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.348745584487915,
      "learning_rate": 1.5230125523012553e-05,
      "loss": 0.9097,
      "step": 364
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0051469802856445,
      "learning_rate": 1.5271966527196653e-05,
      "loss": 0.9568,
      "step": 365
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.507467746734619,
      "learning_rate": 1.5313807531380754e-05,
      "loss": 0.9033,
      "step": 366
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.1319987773895264,
      "learning_rate": 1.5355648535564855e-05,
      "loss": 0.9322,
      "step": 367
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.8861143589019775,
      "learning_rate": 1.5397489539748957e-05,
      "loss": 0.9412,
      "step": 368
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.73506498336792,
      "learning_rate": 1.5439330543933054e-05,
      "loss": 0.9101,
      "step": 369
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.764439344406128,
      "learning_rate": 1.5481171548117155e-05,
      "loss": 0.9437,
      "step": 370
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.647302150726318,
      "learning_rate": 1.5523012552301257e-05,
      "loss": 0.9405,
      "step": 371
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.965416431427002,
      "learning_rate": 1.5564853556485358e-05,
      "loss": 0.9422,
      "step": 372
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6019716262817383,
      "learning_rate": 1.560669456066946e-05,
      "loss": 0.8824,
      "step": 373
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.4240036010742188,
      "learning_rate": 1.5648535564853557e-05,
      "loss": 0.8943,
      "step": 374
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.830317735671997,
      "learning_rate": 1.5690376569037658e-05,
      "loss": 0.9455,
      "step": 375
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.8910844326019287,
      "learning_rate": 1.573221757322176e-05,
      "loss": 0.8586,
      "step": 376
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.81917142868042,
      "learning_rate": 1.577405857740586e-05,
      "loss": 0.8601,
      "step": 377
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.703667402267456,
      "learning_rate": 1.5815899581589958e-05,
      "loss": 0.9324,
      "step": 378
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.832730531692505,
      "learning_rate": 1.585774058577406e-05,
      "loss": 0.9142,
      "step": 379
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4839322566986084,
      "learning_rate": 1.589958158995816e-05,
      "loss": 0.9256,
      "step": 380
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5092241764068604,
      "learning_rate": 1.594142259414226e-05,
      "loss": 0.914,
      "step": 381
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0198748111724854,
      "learning_rate": 1.5983263598326363e-05,
      "loss": 0.9263,
      "step": 382
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.1292171478271484,
      "learning_rate": 1.602510460251046e-05,
      "loss": 0.8807,
      "step": 383
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.566610336303711,
      "learning_rate": 1.6066945606694562e-05,
      "loss": 0.9169,
      "step": 384
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.833841323852539,
      "learning_rate": 1.6108786610878663e-05,
      "loss": 0.9384,
      "step": 385
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.737791061401367,
      "learning_rate": 1.6150627615062764e-05,
      "loss": 0.8923,
      "step": 386
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.718545913696289,
      "learning_rate": 1.6192468619246865e-05,
      "loss": 0.9315,
      "step": 387
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3747074604034424,
      "learning_rate": 1.6234309623430963e-05,
      "loss": 0.9764,
      "step": 388
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5686962604522705,
      "learning_rate": 1.6276150627615064e-05,
      "loss": 0.8811,
      "step": 389
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.66809344291687,
      "learning_rate": 1.6317991631799162e-05,
      "loss": 0.9154,
      "step": 390
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.43936824798584,
      "learning_rate": 1.6359832635983267e-05,
      "loss": 0.9416,
      "step": 391
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.121681213378906,
      "learning_rate": 1.6401673640167365e-05,
      "loss": 0.9141,
      "step": 392
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7611823081970215,
      "learning_rate": 1.6443514644351466e-05,
      "loss": 0.9885,
      "step": 393
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.1847736835479736,
      "learning_rate": 1.6485355648535567e-05,
      "loss": 0.9279,
      "step": 394
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4531033039093018,
      "learning_rate": 1.6527196652719665e-05,
      "loss": 0.9362,
      "step": 395
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.054617404937744,
      "learning_rate": 1.656903765690377e-05,
      "loss": 0.8712,
      "step": 396
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2602791786193848,
      "learning_rate": 1.6610878661087867e-05,
      "loss": 0.8953,
      "step": 397
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.226452350616455,
      "learning_rate": 1.6652719665271968e-05,
      "loss": 0.8992,
      "step": 398
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4858908653259277,
      "learning_rate": 1.669456066945607e-05,
      "loss": 0.8942,
      "step": 399
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6008460521698,
      "learning_rate": 1.6736401673640167e-05,
      "loss": 0.902,
      "step": 400
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.7626450061798096,
      "learning_rate": 1.6778242677824272e-05,
      "loss": 0.8627,
      "step": 401
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.438953161239624,
      "learning_rate": 1.682008368200837e-05,
      "loss": 0.9105,
      "step": 402
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.49332332611084,
      "learning_rate": 1.686192468619247e-05,
      "loss": 0.8704,
      "step": 403
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.8084230422973633,
      "learning_rate": 1.690376569037657e-05,
      "loss": 0.8828,
      "step": 404
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.3207550048828125,
      "learning_rate": 1.694560669456067e-05,
      "loss": 0.9049,
      "step": 405
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.8829598426818848,
      "learning_rate": 1.698744769874477e-05,
      "loss": 0.9634,
      "step": 406
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.232316017150879,
      "learning_rate": 1.7029288702928872e-05,
      "loss": 0.8883,
      "step": 407
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.670722246170044,
      "learning_rate": 1.7071129707112973e-05,
      "loss": 0.9056,
      "step": 408
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.736259698867798,
      "learning_rate": 1.711297071129707e-05,
      "loss": 0.8599,
      "step": 409
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6173176765441895,
      "learning_rate": 1.7154811715481172e-05,
      "loss": 0.8712,
      "step": 410
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.8067994117736816,
      "learning_rate": 1.7196652719665273e-05,
      "loss": 0.8793,
      "step": 411
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.1362688541412354,
      "learning_rate": 1.7238493723849375e-05,
      "loss": 0.8941,
      "step": 412
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.714433193206787,
      "learning_rate": 1.7280334728033476e-05,
      "loss": 0.8948,
      "step": 413
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.19277811050415,
      "learning_rate": 1.7322175732217574e-05,
      "loss": 0.8962,
      "step": 414
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.818874359130859,
      "learning_rate": 1.7364016736401675e-05,
      "loss": 0.9605,
      "step": 415
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6511871814727783,
      "learning_rate": 1.7405857740585776e-05,
      "loss": 0.9117,
      "step": 416
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.3402724266052246,
      "learning_rate": 1.7447698744769877e-05,
      "loss": 0.8995,
      "step": 417
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.674389362335205,
      "learning_rate": 1.7489539748953975e-05,
      "loss": 0.8683,
      "step": 418
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.8967771530151367,
      "learning_rate": 1.7531380753138076e-05,
      "loss": 0.9553,
      "step": 419
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.434424638748169,
      "learning_rate": 1.7573221757322177e-05,
      "loss": 0.8885,
      "step": 420
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.017524242401123,
      "learning_rate": 1.761506276150628e-05,
      "loss": 0.8649,
      "step": 421
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5265140533447266,
      "learning_rate": 1.765690376569038e-05,
      "loss": 0.9166,
      "step": 422
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4230334758758545,
      "learning_rate": 1.7698744769874477e-05,
      "loss": 0.8876,
      "step": 423
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.329719066619873,
      "learning_rate": 1.774058577405858e-05,
      "loss": 0.871,
      "step": 424
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.159644365310669,
      "learning_rate": 1.778242677824268e-05,
      "loss": 0.929,
      "step": 425
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6853408813476562,
      "learning_rate": 1.782426778242678e-05,
      "loss": 0.8829,
      "step": 426
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.235889196395874,
      "learning_rate": 1.7866108786610882e-05,
      "loss": 0.8731,
      "step": 427
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.3459174633026123,
      "learning_rate": 1.790794979079498e-05,
      "loss": 0.9028,
      "step": 428
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.1981000900268555,
      "learning_rate": 1.794979079497908e-05,
      "loss": 0.8608,
      "step": 429
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.9937095642089844,
      "learning_rate": 1.7991631799163182e-05,
      "loss": 0.9249,
      "step": 430
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4853508472442627,
      "learning_rate": 1.8033472803347283e-05,
      "loss": 0.8671,
      "step": 431
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0963492393493652,
      "learning_rate": 1.807531380753138e-05,
      "loss": 0.8862,
      "step": 432
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2027416229248047,
      "learning_rate": 1.8117154811715482e-05,
      "loss": 0.8758,
      "step": 433
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4639370441436768,
      "learning_rate": 1.8158995815899584e-05,
      "loss": 0.8757,
      "step": 434
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6342031955718994,
      "learning_rate": 1.820083682008368e-05,
      "loss": 0.85,
      "step": 435
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2917673587799072,
      "learning_rate": 1.8242677824267786e-05,
      "loss": 0.8719,
      "step": 436
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.338665723800659,
      "learning_rate": 1.8284518828451884e-05,
      "loss": 0.8617,
      "step": 437
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.955965280532837,
      "learning_rate": 1.8326359832635985e-05,
      "loss": 0.8525,
      "step": 438
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3260996341705322,
      "learning_rate": 1.8368200836820086e-05,
      "loss": 0.9126,
      "step": 439
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.740125894546509,
      "learning_rate": 1.8410041841004184e-05,
      "loss": 0.8562,
      "step": 440
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6774723529815674,
      "learning_rate": 1.845188284518829e-05,
      "loss": 0.8917,
      "step": 441
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.385040283203125,
      "learning_rate": 1.8493723849372386e-05,
      "loss": 0.8759,
      "step": 442
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.561842203140259,
      "learning_rate": 1.8535564853556487e-05,
      "loss": 0.9434,
      "step": 443
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.7378838062286377,
      "learning_rate": 1.8577405857740585e-05,
      "loss": 0.8968,
      "step": 444
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.840512752532959,
      "learning_rate": 1.8619246861924686e-05,
      "loss": 0.8892,
      "step": 445
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.526768445968628,
      "learning_rate": 1.8661087866108788e-05,
      "loss": 0.888,
      "step": 446
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6893978118896484,
      "learning_rate": 1.870292887029289e-05,
      "loss": 0.9082,
      "step": 447
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.636446475982666,
      "learning_rate": 1.874476987447699e-05,
      "loss": 0.8834,
      "step": 448
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.538410186767578,
      "learning_rate": 1.8786610878661088e-05,
      "loss": 0.8757,
      "step": 449
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.034524917602539,
      "learning_rate": 1.882845188284519e-05,
      "loss": 0.8628,
      "step": 450
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.436457395553589,
      "learning_rate": 1.887029288702929e-05,
      "loss": 0.8732,
      "step": 451
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.8758139610290527,
      "learning_rate": 1.891213389121339e-05,
      "loss": 0.8956,
      "step": 452
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5114362239837646,
      "learning_rate": 1.8953974895397492e-05,
      "loss": 0.8704,
      "step": 453
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3144125938415527,
      "learning_rate": 1.899581589958159e-05,
      "loss": 0.8753,
      "step": 454
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.735821008682251,
      "learning_rate": 1.903765690376569e-05,
      "loss": 0.8617,
      "step": 455
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.304569721221924,
      "learning_rate": 1.9079497907949793e-05,
      "loss": 0.8491,
      "step": 456
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.491661548614502,
      "learning_rate": 1.9121338912133894e-05,
      "loss": 0.8833,
      "step": 457
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6937601566314697,
      "learning_rate": 1.916317991631799e-05,
      "loss": 0.8856,
      "step": 458
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.589310646057129,
      "learning_rate": 1.9205020920502093e-05,
      "loss": 0.8911,
      "step": 459
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4138336181640625,
      "learning_rate": 1.9246861924686194e-05,
      "loss": 0.8759,
      "step": 460
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.206303834915161,
      "learning_rate": 1.9288702928870295e-05,
      "loss": 0.8636,
      "step": 461
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.372328758239746,
      "learning_rate": 1.9330543933054396e-05,
      "loss": 0.8592,
      "step": 462
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.36767315864563,
      "learning_rate": 1.9372384937238494e-05,
      "loss": 0.835,
      "step": 463
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.649428367614746,
      "learning_rate": 1.9414225941422595e-05,
      "loss": 0.865,
      "step": 464
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.0828614234924316,
      "learning_rate": 1.9456066945606696e-05,
      "loss": 0.8722,
      "step": 465
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.360872507095337,
      "learning_rate": 1.9497907949790798e-05,
      "loss": 0.8505,
      "step": 466
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.28739857673645,
      "learning_rate": 1.95397489539749e-05,
      "loss": 0.8449,
      "step": 467
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5911624431610107,
      "learning_rate": 1.9581589958158997e-05,
      "loss": 0.9279,
      "step": 468
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2286853790283203,
      "learning_rate": 1.9623430962343098e-05,
      "loss": 0.8432,
      "step": 469
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.508402109146118,
      "learning_rate": 1.96652719665272e-05,
      "loss": 0.8475,
      "step": 470
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1167960166931152,
      "learning_rate": 1.97071129707113e-05,
      "loss": 0.82,
      "step": 471
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6248416900634766,
      "learning_rate": 1.9748953974895398e-05,
      "loss": 0.8586,
      "step": 472
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6742031574249268,
      "learning_rate": 1.97907949790795e-05,
      "loss": 0.8486,
      "step": 473
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.397233486175537,
      "learning_rate": 1.98326359832636e-05,
      "loss": 1.0055,
      "step": 474
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.1938393115997314,
      "learning_rate": 1.9874476987447698e-05,
      "loss": 0.8891,
      "step": 475
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.340144634246826,
      "learning_rate": 1.9916317991631803e-05,
      "loss": 0.8726,
      "step": 476
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.878605842590332,
      "learning_rate": 1.99581589958159e-05,
      "loss": 0.8546,
      "step": 477
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.779021739959717,
      "learning_rate": 2e-05,
      "loss": 0.9348,
      "step": 478
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.524148464202881,
      "learning_rate": 1.999999979254112e-05,
      "loss": 0.8795,
      "step": 479
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.8975718021392822,
      "learning_rate": 1.9999999170164486e-05,
      "loss": 0.8968,
      "step": 480
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6998679637908936,
      "learning_rate": 1.9999998132870124e-05,
      "loss": 0.9208,
      "step": 481
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5320427417755127,
      "learning_rate": 1.999999668065808e-05,
      "loss": 0.8571,
      "step": 482
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4207239151000977,
      "learning_rate": 1.999999481352841e-05,
      "loss": 0.8434,
      "step": 483
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.40512752532959,
      "learning_rate": 1.999999253148119e-05,
      "loss": 0.8591,
      "step": 484
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.256650447845459,
      "learning_rate": 1.999998983451652e-05,
      "loss": 0.8539,
      "step": 485
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.379611015319824,
      "learning_rate": 1.999998672263451e-05,
      "loss": 0.8547,
      "step": 486
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5022308826446533,
      "learning_rate": 1.999998319583529e-05,
      "loss": 0.8319,
      "step": 487
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5116937160491943,
      "learning_rate": 1.9999979254119002e-05,
      "loss": 0.8694,
      "step": 488
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1019020080566406,
      "learning_rate": 1.9999974897485818e-05,
      "loss": 0.8543,
      "step": 489
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.293107748031616,
      "learning_rate": 1.999997012593591e-05,
      "loss": 0.8684,
      "step": 490
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.495701789855957,
      "learning_rate": 1.999996493946948e-05,
      "loss": 0.8373,
      "step": 491
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.303110122680664,
      "learning_rate": 1.9999959338086743e-05,
      "loss": 0.8477,
      "step": 492
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.180703639984131,
      "learning_rate": 1.9999953321787928e-05,
      "loss": 0.8672,
      "step": 493
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2083683013916016,
      "learning_rate": 1.999994689057329e-05,
      "loss": 0.8621,
      "step": 494
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.397160530090332,
      "learning_rate": 1.9999940044443095e-05,
      "loss": 0.8354,
      "step": 495
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.6382687091827393,
      "learning_rate": 1.9999932783397627e-05,
      "loss": 0.9858,
      "step": 496
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.8865694999694824,
      "learning_rate": 1.9999925107437184e-05,
      "loss": 0.9308,
      "step": 497
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.0634593963623047,
      "learning_rate": 1.9999917016562084e-05,
      "loss": 0.8719,
      "step": 498
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2721989154815674,
      "learning_rate": 1.999990851077267e-05,
      "loss": 0.8873,
      "step": 499
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5183863639831543,
      "learning_rate": 1.9999899590069286e-05,
      "loss": 0.8448,
      "step": 500
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0145986080169678,
      "learning_rate": 1.9999890254452306e-05,
      "loss": 0.8324,
      "step": 501
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.4634180068969727,
      "learning_rate": 1.9999880503922123e-05,
      "loss": 0.8675,
      "step": 502
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.8005926609039307,
      "learning_rate": 1.999987033847913e-05,
      "loss": 0.8797,
      "step": 503
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9956698417663574,
      "learning_rate": 1.9999859758123757e-05,
      "loss": 0.8521,
      "step": 504
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.733058452606201,
      "learning_rate": 1.9999848762856442e-05,
      "loss": 0.8547,
      "step": 505
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.295274019241333,
      "learning_rate": 1.999983735267764e-05,
      "loss": 0.8546,
      "step": 506
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.195747137069702,
      "learning_rate": 1.9999825527587823e-05,
      "loss": 0.8705,
      "step": 507
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.496394634246826,
      "learning_rate": 1.9999813287587483e-05,
      "loss": 0.8275,
      "step": 508
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3572804927825928,
      "learning_rate": 1.9999800632677133e-05,
      "loss": 0.8204,
      "step": 509
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.8487443923950195,
      "learning_rate": 1.9999787562857285e-05,
      "loss": 0.8769,
      "step": 510
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.46423077583313,
      "learning_rate": 1.9999774078128496e-05,
      "loss": 0.8567,
      "step": 511
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6214702129364014,
      "learning_rate": 1.9999760178491318e-05,
      "loss": 0.8988,
      "step": 512
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.58048677444458,
      "learning_rate": 1.9999745863946326e-05,
      "loss": 0.877,
      "step": 513
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.3730318546295166,
      "learning_rate": 1.9999731134494116e-05,
      "loss": 0.9796,
      "step": 514
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.1458799839019775,
      "learning_rate": 1.9999715990135303e-05,
      "loss": 0.8515,
      "step": 515
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.345498561859131,
      "learning_rate": 1.999970043087051e-05,
      "loss": 0.8721,
      "step": 516
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.266444683074951,
      "learning_rate": 1.9999684456700382e-05,
      "loss": 0.8651,
      "step": 517
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4046294689178467,
      "learning_rate": 1.999966806762559e-05,
      "loss": 0.845,
      "step": 518
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0653622150421143,
      "learning_rate": 1.9999651263646808e-05,
      "loss": 0.7943,
      "step": 519
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.36204195022583,
      "learning_rate": 1.999963404476473e-05,
      "loss": 0.7964,
      "step": 520
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0180015563964844,
      "learning_rate": 1.9999616410980076e-05,
      "loss": 0.9112,
      "step": 521
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6189467906951904,
      "learning_rate": 1.9999598362293575e-05,
      "loss": 0.853,
      "step": 522
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.380749464035034,
      "learning_rate": 1.9999579898705975e-05,
      "loss": 0.8679,
      "step": 523
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0601160526275635,
      "learning_rate": 1.9999561020218047e-05,
      "loss": 0.8549,
      "step": 524
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.546802282333374,
      "learning_rate": 1.9999541726830572e-05,
      "loss": 0.8624,
      "step": 525
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4818639755249023,
      "learning_rate": 1.9999522018544347e-05,
      "loss": 0.8465,
      "step": 526
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3844802379608154,
      "learning_rate": 1.9999501895360192e-05,
      "loss": 0.8734,
      "step": 527
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1644415855407715,
      "learning_rate": 1.9999481357278943e-05,
      "loss": 0.8701,
      "step": 528
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3936679363250732,
      "learning_rate": 1.999946040430145e-05,
      "loss": 0.8374,
      "step": 529
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.7713468074798584,
      "learning_rate": 1.9999439036428587e-05,
      "loss": 0.8037,
      "step": 530
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2969837188720703,
      "learning_rate": 1.9999417253661235e-05,
      "loss": 0.8656,
      "step": 531
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9742426872253418,
      "learning_rate": 1.9999395056000303e-05,
      "loss": 0.8557,
      "step": 532
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0437381267547607,
      "learning_rate": 1.9999372443446705e-05,
      "loss": 0.831,
      "step": 533
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3008339405059814,
      "learning_rate": 1.9999349416001385e-05,
      "loss": 0.8664,
      "step": 534
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.063551664352417,
      "learning_rate": 1.99993259736653e-05,
      "loss": 0.8762,
      "step": 535
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1993088722229004,
      "learning_rate": 1.9999302116439416e-05,
      "loss": 0.7955,
      "step": 536
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.997389554977417,
      "learning_rate": 1.9999277844324726e-05,
      "loss": 0.8162,
      "step": 537
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.102219820022583,
      "learning_rate": 1.999925315732224e-05,
      "loss": 0.8062,
      "step": 538
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4425761699676514,
      "learning_rate": 1.9999228055432977e-05,
      "loss": 0.8806,
      "step": 539
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0836493968963623,
      "learning_rate": 1.9999202538657987e-05,
      "loss": 0.8499,
      "step": 540
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.275494337081909,
      "learning_rate": 1.999917660699832e-05,
      "loss": 0.7975,
      "step": 541
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3620364665985107,
      "learning_rate": 1.9999150260455052e-05,
      "loss": 0.8387,
      "step": 542
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8936179876327515,
      "learning_rate": 1.9999123499029284e-05,
      "loss": 0.8452,
      "step": 543
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0183825492858887,
      "learning_rate": 1.9999096322722116e-05,
      "loss": 0.8207,
      "step": 544
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3548049926757812,
      "learning_rate": 1.9999068731534684e-05,
      "loss": 0.8863,
      "step": 545
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2385616302490234,
      "learning_rate": 1.999904072546813e-05,
      "loss": 0.8366,
      "step": 546
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.359286069869995,
      "learning_rate": 1.9999012304523617e-05,
      "loss": 0.8247,
      "step": 547
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.013007640838623,
      "learning_rate": 1.9998983468702323e-05,
      "loss": 0.8381,
      "step": 548
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9902924299240112,
      "learning_rate": 1.9998954218005444e-05,
      "loss": 0.8439,
      "step": 549
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0028774738311768,
      "learning_rate": 1.9998924552434192e-05,
      "loss": 0.8365,
      "step": 550
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5395703315734863,
      "learning_rate": 1.9998894471989804e-05,
      "loss": 0.7702,
      "step": 551
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.164900064468384,
      "learning_rate": 1.9998863976673523e-05,
      "loss": 0.7993,
      "step": 552
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6024935245513916,
      "learning_rate": 1.9998833066486617e-05,
      "loss": 0.8267,
      "step": 553
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1585423946380615,
      "learning_rate": 1.9998801741430363e-05,
      "loss": 0.8035,
      "step": 554
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7938300371170044,
      "learning_rate": 1.999877000150607e-05,
      "loss": 0.8434,
      "step": 555
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.306025505065918,
      "learning_rate": 1.999873784671505e-05,
      "loss": 0.834,
      "step": 556
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0556046962738037,
      "learning_rate": 1.999870527705863e-05,
      "loss": 0.8164,
      "step": 557
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5304393768310547,
      "learning_rate": 1.9998672292538175e-05,
      "loss": 0.8324,
      "step": 558
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3627688884735107,
      "learning_rate": 1.9998638893155046e-05,
      "loss": 0.8422,
      "step": 559
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2391576766967773,
      "learning_rate": 1.999860507891063e-05,
      "loss": 0.7796,
      "step": 560
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.197885036468506,
      "learning_rate": 1.9998570849806326e-05,
      "loss": 0.856,
      "step": 561
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.320101499557495,
      "learning_rate": 1.999853620584356e-05,
      "loss": 0.838,
      "step": 562
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0973286628723145,
      "learning_rate": 1.9998501147023765e-05,
      "loss": 0.7933,
      "step": 563
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0967214107513428,
      "learning_rate": 1.9998465673348404e-05,
      "loss": 0.7912,
      "step": 564
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2318851947784424,
      "learning_rate": 1.999842978481894e-05,
      "loss": 0.8124,
      "step": 565
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0396173000335693,
      "learning_rate": 1.9998393481436863e-05,
      "loss": 0.7842,
      "step": 566
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1071856021881104,
      "learning_rate": 1.9998356763203683e-05,
      "loss": 0.8243,
      "step": 567
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.315293312072754,
      "learning_rate": 1.9998319630120925e-05,
      "loss": 0.8143,
      "step": 568
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.319780111312866,
      "learning_rate": 1.999828208219012e-05,
      "loss": 0.7985,
      "step": 569
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3786582946777344,
      "learning_rate": 1.9998244119412836e-05,
      "loss": 0.8239,
      "step": 570
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1613852977752686,
      "learning_rate": 1.9998205741790647e-05,
      "loss": 0.8315,
      "step": 571
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2382776737213135,
      "learning_rate": 1.999816694932514e-05,
      "loss": 0.8165,
      "step": 572
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1310315132141113,
      "learning_rate": 1.9998127742017926e-05,
      "loss": 0.8219,
      "step": 573
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1730072498321533,
      "learning_rate": 1.9998088119870637e-05,
      "loss": 0.8053,
      "step": 574
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.141110420227051,
      "learning_rate": 1.999804808288491e-05,
      "loss": 0.8102,
      "step": 575
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3077547550201416,
      "learning_rate": 1.9998007631062412e-05,
      "loss": 0.8055,
      "step": 576
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0090291500091553,
      "learning_rate": 1.9997966764404816e-05,
      "loss": 0.8042,
      "step": 577
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0530202388763428,
      "learning_rate": 1.9997925482913824e-05,
      "loss": 0.8067,
      "step": 578
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0001113414764404,
      "learning_rate": 1.9997883786591143e-05,
      "loss": 0.8156,
      "step": 579
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5116782188415527,
      "learning_rate": 1.9997841675438508e-05,
      "loss": 0.8179,
      "step": 580
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.112147092819214,
      "learning_rate": 1.9997799149457663e-05,
      "loss": 0.8273,
      "step": 581
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4030652046203613,
      "learning_rate": 1.999775620865037e-05,
      "loss": 0.853,
      "step": 582
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.337937831878662,
      "learning_rate": 1.9997712853018417e-05,
      "loss": 0.8139,
      "step": 583
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2691612243652344,
      "learning_rate": 1.9997669082563597e-05,
      "loss": 0.8514,
      "step": 584
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3706376552581787,
      "learning_rate": 1.9997624897287733e-05,
      "loss": 0.8253,
      "step": 585
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4782276153564453,
      "learning_rate": 1.999758029719265e-05,
      "loss": 0.7714,
      "step": 586
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9720518589019775,
      "learning_rate": 1.9997535282280206e-05,
      "loss": 0.8531,
      "step": 587
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.099210739135742,
      "learning_rate": 1.9997489852552264e-05,
      "loss": 0.8174,
      "step": 588
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0982089042663574,
      "learning_rate": 1.9997444008010712e-05,
      "loss": 0.8444,
      "step": 589
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.112231731414795,
      "learning_rate": 1.999739774865745e-05,
      "loss": 0.793,
      "step": 590
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.798864483833313,
      "learning_rate": 1.9997351074494398e-05,
      "loss": 0.8087,
      "step": 591
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.023737668991089,
      "learning_rate": 1.9997303985523492e-05,
      "loss": 0.805,
      "step": 592
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.954448938369751,
      "learning_rate": 1.9997256481746684e-05,
      "loss": 0.7671,
      "step": 593
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8242018222808838,
      "learning_rate": 1.999720856316595e-05,
      "loss": 0.8299,
      "step": 594
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.949776291847229,
      "learning_rate": 1.9997160229783277e-05,
      "loss": 0.7736,
      "step": 595
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9198299646377563,
      "learning_rate": 1.999711148160067e-05,
      "loss": 0.8013,
      "step": 596
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4279260635375977,
      "learning_rate": 1.999706231862015e-05,
      "loss": 0.8031,
      "step": 597
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9350900650024414,
      "learning_rate": 1.9997012740843752e-05,
      "loss": 0.7731,
      "step": 598
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1206939220428467,
      "learning_rate": 1.9996962748273545e-05,
      "loss": 0.8107,
      "step": 599
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.08754301071167,
      "learning_rate": 1.9996912340911592e-05,
      "loss": 0.7992,
      "step": 600
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9362516403198242,
      "learning_rate": 1.9996861518759997e-05,
      "loss": 0.7792,
      "step": 601
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0590217113494873,
      "learning_rate": 1.9996810281820854e-05,
      "loss": 0.8233,
      "step": 602
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7981475591659546,
      "learning_rate": 1.99967586300963e-05,
      "loss": 0.787,
      "step": 603
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2393906116485596,
      "learning_rate": 1.999670656358847e-05,
      "loss": 0.9841,
      "step": 604
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5587873458862305,
      "learning_rate": 1.9996654082299532e-05,
      "loss": 0.8232,
      "step": 605
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1794936656951904,
      "learning_rate": 1.9996601186231663e-05,
      "loss": 0.7922,
      "step": 606
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.135974407196045,
      "learning_rate": 1.999654787538705e-05,
      "loss": 0.7784,
      "step": 607
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1100845336914062,
      "learning_rate": 1.999649414976791e-05,
      "loss": 0.7756,
      "step": 608
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.176189422607422,
      "learning_rate": 1.9996440009376475e-05,
      "loss": 0.8572,
      "step": 609
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.107381582260132,
      "learning_rate": 1.9996385454214983e-05,
      "loss": 0.826,
      "step": 610
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.040105104446411,
      "learning_rate": 1.999633048428571e-05,
      "loss": 0.8004,
      "step": 611
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2563729286193848,
      "learning_rate": 1.9996275099590924e-05,
      "loss": 0.7742,
      "step": 612
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9275959730148315,
      "learning_rate": 1.999621930013293e-05,
      "loss": 0.789,
      "step": 613
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0402042865753174,
      "learning_rate": 1.9996163085914042e-05,
      "loss": 0.8173,
      "step": 614
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.804044485092163,
      "learning_rate": 1.9996106456936592e-05,
      "loss": 0.8008,
      "step": 615
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8951005935668945,
      "learning_rate": 1.9996049413202933e-05,
      "loss": 0.7935,
      "step": 616
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0430076122283936,
      "learning_rate": 1.9995991954715424e-05,
      "loss": 0.764,
      "step": 617
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1759181022644043,
      "learning_rate": 1.999593408147646e-05,
      "loss": 0.7892,
      "step": 618
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2285077571868896,
      "learning_rate": 1.999587579348843e-05,
      "loss": 0.7722,
      "step": 619
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3936328887939453,
      "learning_rate": 1.9995817090753765e-05,
      "loss": 0.8032,
      "step": 620
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9215384721755981,
      "learning_rate": 1.9995757973274892e-05,
      "loss": 0.7595,
      "step": 621
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.154735565185547,
      "learning_rate": 1.9995698441054264e-05,
      "loss": 0.8111,
      "step": 622
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.169736623764038,
      "learning_rate": 1.9995638494094355e-05,
      "loss": 0.7689,
      "step": 623
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9220738410949707,
      "learning_rate": 1.999557813239765e-05,
      "loss": 0.8081,
      "step": 624
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0688624382019043,
      "learning_rate": 1.9995517355966655e-05,
      "loss": 0.8855,
      "step": 625
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9695467948913574,
      "learning_rate": 1.9995456164803892e-05,
      "loss": 0.8032,
      "step": 626
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.126220941543579,
      "learning_rate": 1.9995394558911894e-05,
      "loss": 0.795,
      "step": 627
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8981989622116089,
      "learning_rate": 1.9995332538293225e-05,
      "loss": 0.7916,
      "step": 628
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0491344928741455,
      "learning_rate": 1.9995270102950454e-05,
      "loss": 0.7797,
      "step": 629
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8407200574874878,
      "learning_rate": 1.9995207252886176e-05,
      "loss": 0.7813,
      "step": 630
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0169801712036133,
      "learning_rate": 1.9995143988102993e-05,
      "loss": 0.8172,
      "step": 631
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0181992053985596,
      "learning_rate": 1.999508030860353e-05,
      "loss": 0.8045,
      "step": 632
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0644099712371826,
      "learning_rate": 1.999501621439044e-05,
      "loss": 0.8352,
      "step": 633
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.925586223602295,
      "learning_rate": 1.9994951705466366e-05,
      "loss": 0.8096,
      "step": 634
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0262291431427,
      "learning_rate": 1.9994886781833998e-05,
      "loss": 0.7608,
      "step": 635
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8728125095367432,
      "learning_rate": 1.999482144349602e-05,
      "loss": 0.7892,
      "step": 636
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9263396263122559,
      "learning_rate": 1.9994755690455154e-05,
      "loss": 0.759,
      "step": 637
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2309212684631348,
      "learning_rate": 1.9994689522714115e-05,
      "loss": 0.8268,
      "step": 638
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8442708253860474,
      "learning_rate": 1.9994622940275656e-05,
      "loss": 0.7991,
      "step": 639
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.21720290184021,
      "learning_rate": 1.999455594314254e-05,
      "loss": 0.7768,
      "step": 640
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1015193462371826,
      "learning_rate": 1.9994488531317547e-05,
      "loss": 0.7718,
      "step": 641
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.081153154373169,
      "learning_rate": 1.9994420704803472e-05,
      "loss": 0.7644,
      "step": 642
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1697282791137695,
      "learning_rate": 1.999435246360313e-05,
      "loss": 0.8098,
      "step": 643
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1917455196380615,
      "learning_rate": 1.999428380771935e-05,
      "loss": 0.8512,
      "step": 644
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.254530668258667,
      "learning_rate": 1.9994214737154984e-05,
      "loss": 0.8041,
      "step": 645
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5044972896575928,
      "learning_rate": 1.99941452519129e-05,
      "loss": 0.7452,
      "step": 646
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3083622455596924,
      "learning_rate": 1.9994075351995973e-05,
      "loss": 0.7913,
      "step": 647
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.049799919128418,
      "learning_rate": 1.9994005037407112e-05,
      "loss": 0.7625,
      "step": 648
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.037775993347168,
      "learning_rate": 1.999393430814923e-05,
      "loss": 0.7985,
      "step": 649
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.980259895324707,
      "learning_rate": 1.999386316422526e-05,
      "loss": 0.786,
      "step": 650
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8763400316238403,
      "learning_rate": 1.999379160563816e-05,
      "loss": 0.8052,
      "step": 651
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.185363531112671,
      "learning_rate": 1.9993719632390892e-05,
      "loss": 0.7832,
      "step": 652
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1516878604888916,
      "learning_rate": 1.9993647244486448e-05,
      "loss": 0.7564,
      "step": 653
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0363259315490723,
      "learning_rate": 1.999357444192783e-05,
      "loss": 0.7713,
      "step": 654
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0013949871063232,
      "learning_rate": 1.999350122471805e-05,
      "loss": 0.8163,
      "step": 655
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.149301052093506,
      "learning_rate": 1.999342759286016e-05,
      "loss": 0.7584,
      "step": 656
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4736266136169434,
      "learning_rate": 1.999335354635721e-05,
      "loss": 0.7916,
      "step": 657
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0013649463653564,
      "learning_rate": 1.999327908521227e-05,
      "loss": 0.8108,
      "step": 658
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9305516481399536,
      "learning_rate": 1.999320420942843e-05,
      "loss": 0.7913,
      "step": 659
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0176751613616943,
      "learning_rate": 1.9993128919008796e-05,
      "loss": 0.7702,
      "step": 660
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.016265630722046,
      "learning_rate": 1.9993053213956496e-05,
      "loss": 0.7724,
      "step": 661
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9794402122497559,
      "learning_rate": 1.999297709427467e-05,
      "loss": 0.7791,
      "step": 662
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1290931701660156,
      "learning_rate": 1.999290055996647e-05,
      "loss": 0.801,
      "step": 663
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1865546703338623,
      "learning_rate": 1.999282361103508e-05,
      "loss": 0.7689,
      "step": 664
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0027964115142822,
      "learning_rate": 1.9992746247483686e-05,
      "loss": 0.8231,
      "step": 665
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9383741617202759,
      "learning_rate": 1.9992668469315503e-05,
      "loss": 0.7806,
      "step": 666
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1195573806762695,
      "learning_rate": 1.9992590276533755e-05,
      "loss": 0.7728,
      "step": 667
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0251123905181885,
      "learning_rate": 1.999251166914169e-05,
      "loss": 0.8037,
      "step": 668
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8451539278030396,
      "learning_rate": 1.9992432647142566e-05,
      "loss": 0.7696,
      "step": 669
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.3095996379852295,
      "learning_rate": 1.9992353210539665e-05,
      "loss": 0.9888,
      "step": 670
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.744368314743042,
      "learning_rate": 1.9992273359336277e-05,
      "loss": 0.7888,
      "step": 671
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.199143648147583,
      "learning_rate": 1.999219309353572e-05,
      "loss": 0.7491,
      "step": 672
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1591391563415527,
      "learning_rate": 1.9992112413141324e-05,
      "loss": 0.7746,
      "step": 673
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1845650672912598,
      "learning_rate": 1.9992031318156436e-05,
      "loss": 0.7804,
      "step": 674
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.157092332839966,
      "learning_rate": 1.9991949808584422e-05,
      "loss": 0.7894,
      "step": 675
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0103070735931396,
      "learning_rate": 1.9991867884428663e-05,
      "loss": 0.7522,
      "step": 676
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0558512210845947,
      "learning_rate": 1.9991785545692553e-05,
      "loss": 0.7858,
      "step": 677
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2007639408111572,
      "learning_rate": 1.9991702792379518e-05,
      "loss": 0.8196,
      "step": 678
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.056284189224243,
      "learning_rate": 1.9991619624492986e-05,
      "loss": 0.7582,
      "step": 679
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.696709156036377,
      "learning_rate": 1.9991536042036407e-05,
      "loss": 0.9811,
      "step": 680
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.968381643295288,
      "learning_rate": 1.9991452045013254e-05,
      "loss": 0.7712,
      "step": 681
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0897483825683594,
      "learning_rate": 1.9991367633427006e-05,
      "loss": 0.7859,
      "step": 682
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3393664360046387,
      "learning_rate": 1.999128280728117e-05,
      "loss": 0.7974,
      "step": 683
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3943636417388916,
      "learning_rate": 1.9991197566579262e-05,
      "loss": 0.7958,
      "step": 684
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9186569452285767,
      "learning_rate": 1.999111191132482e-05,
      "loss": 0.7552,
      "step": 685
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.436310052871704,
      "learning_rate": 1.99910258415214e-05,
      "loss": 0.8266,
      "step": 686
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.176987648010254,
      "learning_rate": 1.9990939357172573e-05,
      "loss": 0.8068,
      "step": 687
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3496880531311035,
      "learning_rate": 1.9990852458281925e-05,
      "loss": 0.7986,
      "step": 688
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.413487195968628,
      "learning_rate": 1.9990765144853062e-05,
      "loss": 0.7629,
      "step": 689
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9667763710021973,
      "learning_rate": 1.999067741688961e-05,
      "loss": 0.761,
      "step": 690
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2301061153411865,
      "learning_rate": 1.9990589274395203e-05,
      "loss": 0.8019,
      "step": 691
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2603280544281006,
      "learning_rate": 1.9990500717373505e-05,
      "loss": 0.739,
      "step": 692
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9713934659957886,
      "learning_rate": 1.9990411745828188e-05,
      "loss": 0.7502,
      "step": 693
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2420594692230225,
      "learning_rate": 1.9990322359762937e-05,
      "loss": 0.7727,
      "step": 694
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.052481174468994,
      "learning_rate": 1.999023255918147e-05,
      "loss": 0.7827,
      "step": 695
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.115142822265625,
      "learning_rate": 1.999014234408751e-05,
      "loss": 0.73,
      "step": 696
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.038677215576172,
      "learning_rate": 1.99900517144848e-05,
      "loss": 0.7606,
      "step": 697
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.32489013671875,
      "learning_rate": 1.9989960670377098e-05,
      "loss": 0.8053,
      "step": 698
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.082376003265381,
      "learning_rate": 1.9989869211768186e-05,
      "loss": 0.7579,
      "step": 699
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.147339344024658,
      "learning_rate": 1.9989777338661855e-05,
      "loss": 0.7683,
      "step": 700
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.735652208328247,
      "learning_rate": 1.998968505106192e-05,
      "loss": 0.7756,
      "step": 701
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0057079792022705,
      "learning_rate": 1.99895923489722e-05,
      "loss": 0.7757,
      "step": 702
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.7105491161346436,
      "learning_rate": 1.998949923239656e-05,
      "loss": 0.9936,
      "step": 703
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4938786029815674,
      "learning_rate": 1.998940570133885e-05,
      "loss": 0.781,
      "step": 704
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.086275577545166,
      "learning_rate": 1.9989311755802957e-05,
      "loss": 0.7781,
      "step": 705
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.858659267425537,
      "learning_rate": 1.998921739579277e-05,
      "loss": 0.7654,
      "step": 706
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0111379623413086,
      "learning_rate": 1.9989122621312214e-05,
      "loss": 0.756,
      "step": 707
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0361673831939697,
      "learning_rate": 1.998902743236522e-05,
      "loss": 0.8028,
      "step": 708
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8978227376937866,
      "learning_rate": 1.9988931828955732e-05,
      "loss": 0.7671,
      "step": 709
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9091124534606934,
      "learning_rate": 1.9988835811087725e-05,
      "loss": 0.8029,
      "step": 710
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8811371326446533,
      "learning_rate": 1.9988739378765172e-05,
      "loss": 0.7687,
      "step": 711
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8757492303848267,
      "learning_rate": 1.9988642531992085e-05,
      "loss": 0.7855,
      "step": 712
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9476982355117798,
      "learning_rate": 1.9988545270772476e-05,
      "loss": 0.754,
      "step": 713
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9057884216308594,
      "learning_rate": 1.998844759511038e-05,
      "loss": 0.767,
      "step": 714
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0637736320495605,
      "learning_rate": 1.9988349505009853e-05,
      "loss": 0.7627,
      "step": 715
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7731820344924927,
      "learning_rate": 1.9988251000474964e-05,
      "loss": 0.7339,
      "step": 716
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.818691372871399,
      "learning_rate": 1.9988152081509796e-05,
      "loss": 0.8004,
      "step": 717
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9903788566589355,
      "learning_rate": 1.9988052748118464e-05,
      "loss": 0.8234,
      "step": 718
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1666433811187744,
      "learning_rate": 1.998795300030508e-05,
      "loss": 0.7808,
      "step": 719
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4349355697631836,
      "learning_rate": 1.9987852838073786e-05,
      "loss": 1.0367,
      "step": 720
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5124521255493164,
      "learning_rate": 1.9987752261428734e-05,
      "loss": 0.7859,
      "step": 721
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0623035430908203,
      "learning_rate": 1.9987651270374104e-05,
      "loss": 0.7644,
      "step": 722
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9126496315002441,
      "learning_rate": 1.998754986491408e-05,
      "loss": 0.7641,
      "step": 723
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.062608003616333,
      "learning_rate": 1.9987448045052872e-05,
      "loss": 0.7992,
      "step": 724
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4956904649734497,
      "learning_rate": 1.9987345810794707e-05,
      "loss": 1.0007,
      "step": 725
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2324962615966797,
      "learning_rate": 1.9987243162143826e-05,
      "loss": 0.7327,
      "step": 726
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.171229839324951,
      "learning_rate": 1.9987140099104484e-05,
      "loss": 0.7952,
      "step": 727
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9502924680709839,
      "learning_rate": 1.9987036621680962e-05,
      "loss": 0.7505,
      "step": 728
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2288732528686523,
      "learning_rate": 1.998693272987755e-05,
      "loss": 0.7731,
      "step": 729
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9719821214675903,
      "learning_rate": 1.9986828423698562e-05,
      "loss": 0.7457,
      "step": 730
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7816345691680908,
      "learning_rate": 1.9986723703148324e-05,
      "loss": 0.7502,
      "step": 731
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0612733364105225,
      "learning_rate": 1.998661856823118e-05,
      "loss": 0.7966,
      "step": 732
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.929885983467102,
      "learning_rate": 1.9986513018951493e-05,
      "loss": 0.7748,
      "step": 733
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0169677734375,
      "learning_rate": 1.9986407055313645e-05,
      "loss": 0.7593,
      "step": 734
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.988961100578308,
      "learning_rate": 1.9986300677322028e-05,
      "loss": 0.8053,
      "step": 735
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9240283966064453,
      "learning_rate": 1.998619388498106e-05,
      "loss": 0.7384,
      "step": 736
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0279130935668945,
      "learning_rate": 1.9986086678295168e-05,
      "loss": 0.7828,
      "step": 737
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.19435453414917,
      "learning_rate": 1.9985979057268806e-05,
      "loss": 0.7489,
      "step": 738
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1309802532196045,
      "learning_rate": 1.9985871021906432e-05,
      "loss": 1.0294,
      "step": 739
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.261352062225342,
      "learning_rate": 1.9985762572212535e-05,
      "loss": 0.7921,
      "step": 740
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0677757263183594,
      "learning_rate": 1.9985653708191612e-05,
      "loss": 0.8021,
      "step": 741
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8543763160705566,
      "learning_rate": 1.9985544429848178e-05,
      "loss": 0.7643,
      "step": 742
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9770587682724,
      "learning_rate": 1.998543473718677e-05,
      "loss": 0.7909,
      "step": 743
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9641695022583008,
      "learning_rate": 1.998532463021194e-05,
      "loss": 0.7476,
      "step": 744
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.470292806625366,
      "learning_rate": 1.9985214108928255e-05,
      "loss": 0.8064,
      "step": 745
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6274415254592896,
      "learning_rate": 1.99851031733403e-05,
      "loss": 0.9599,
      "step": 746
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4240903854370117,
      "learning_rate": 1.998499182345268e-05,
      "loss": 0.741,
      "step": 747
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.359056234359741,
      "learning_rate": 1.9984880059270013e-05,
      "loss": 0.7837,
      "step": 748
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7584147453308105,
      "learning_rate": 1.998476788079694e-05,
      "loss": 0.7087,
      "step": 749
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0770339965820312,
      "learning_rate": 1.9984655288038106e-05,
      "loss": 0.752,
      "step": 750
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1700425148010254,
      "learning_rate": 1.9984542280998193e-05,
      "loss": 0.7669,
      "step": 751
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8690860271453857,
      "learning_rate": 1.9984428859681888e-05,
      "loss": 0.7746,
      "step": 752
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1850812435150146,
      "learning_rate": 1.9984315024093895e-05,
      "loss": 0.7829,
      "step": 753
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1767563819885254,
      "learning_rate": 1.9984200774238937e-05,
      "loss": 0.7552,
      "step": 754
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.059927225112915,
      "learning_rate": 1.9984086110121753e-05,
      "loss": 0.7445,
      "step": 755
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9048036336898804,
      "learning_rate": 1.9983971031747104e-05,
      "loss": 0.7786,
      "step": 756
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.099579334259033,
      "learning_rate": 1.9983855539119765e-05,
      "loss": 0.7474,
      "step": 757
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9999624490737915,
      "learning_rate": 1.9983739632244524e-05,
      "loss": 0.7952,
      "step": 758
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.814787745475769,
      "learning_rate": 1.9983623311126192e-05,
      "loss": 0.7615,
      "step": 759
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9434716701507568,
      "learning_rate": 1.9983506575769595e-05,
      "loss": 0.7669,
      "step": 760
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1186299324035645,
      "learning_rate": 1.998338942617958e-05,
      "loss": 0.7486,
      "step": 761
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6903932094573975,
      "learning_rate": 1.9983271862361008e-05,
      "loss": 0.7314,
      "step": 762
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8382139205932617,
      "learning_rate": 1.9983153884318748e-05,
      "loss": 0.7781,
      "step": 763
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8477314710617065,
      "learning_rate": 1.9983035492057703e-05,
      "loss": 0.8303,
      "step": 764
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9347554445266724,
      "learning_rate": 1.9982916685582787e-05,
      "loss": 0.7878,
      "step": 765
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.084519147872925,
      "learning_rate": 1.9982797464898922e-05,
      "loss": 0.7721,
      "step": 766
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.840031623840332,
      "learning_rate": 1.9982677830011063e-05,
      "loss": 0.7667,
      "step": 767
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9237340688705444,
      "learning_rate": 1.9982557780924167e-05,
      "loss": 0.7364,
      "step": 768
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0347988605499268,
      "learning_rate": 1.9982437317643218e-05,
      "loss": 0.7972,
      "step": 769
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.020185708999634,
      "learning_rate": 1.998231644017321e-05,
      "loss": 0.7739,
      "step": 770
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1149866580963135,
      "learning_rate": 1.9982195148519166e-05,
      "loss": 0.7527,
      "step": 771
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9139726161956787,
      "learning_rate": 1.9982073442686116e-05,
      "loss": 0.7635,
      "step": 772
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.046670436859131,
      "learning_rate": 1.9981951322679107e-05,
      "loss": 0.7676,
      "step": 773
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0811195373535156,
      "learning_rate": 1.998182878850321e-05,
      "loss": 0.7435,
      "step": 774
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.865787148475647,
      "learning_rate": 1.9981705840163506e-05,
      "loss": 0.763,
      "step": 775
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0059051513671875,
      "learning_rate": 1.9981582477665094e-05,
      "loss": 0.752,
      "step": 776
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9410730600357056,
      "learning_rate": 1.99814587010131e-05,
      "loss": 0.761,
      "step": 777
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7603917121887207,
      "learning_rate": 1.998133451021265e-05,
      "loss": 0.7408,
      "step": 778
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8810168504714966,
      "learning_rate": 1.9981209905268904e-05,
      "loss": 0.7965,
      "step": 779
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8054713010787964,
      "learning_rate": 1.9981084886187035e-05,
      "loss": 0.7314,
      "step": 780
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8546063899993896,
      "learning_rate": 1.9980959452972222e-05,
      "loss": 0.7793,
      "step": 781
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6865665912628174,
      "learning_rate": 1.9980833605629676e-05,
      "loss": 0.7451,
      "step": 782
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.530540704727173,
      "learning_rate": 1.9980707344164614e-05,
      "loss": 0.7979,
      "step": 783
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0339248180389404,
      "learning_rate": 1.9980580668582274e-05,
      "loss": 0.7545,
      "step": 784
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7914544343948364,
      "learning_rate": 1.9980453578887918e-05,
      "loss": 0.7938,
      "step": 785
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9442545175552368,
      "learning_rate": 1.9980326075086815e-05,
      "loss": 0.7646,
      "step": 786
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7685201168060303,
      "learning_rate": 1.9980198157184256e-05,
      "loss": 0.7322,
      "step": 787
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.771077036857605,
      "learning_rate": 1.998006982518555e-05,
      "loss": 0.7405,
      "step": 788
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8570501804351807,
      "learning_rate": 1.997994107909602e-05,
      "loss": 0.8461,
      "step": 789
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8560341596603394,
      "learning_rate": 1.9979811918921004e-05,
      "loss": 0.7673,
      "step": 790
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.802781581878662,
      "learning_rate": 1.9979682344665863e-05,
      "loss": 0.7845,
      "step": 791
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6968846321105957,
      "learning_rate": 1.9979552356335983e-05,
      "loss": 0.7375,
      "step": 792
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9244003295898438,
      "learning_rate": 1.997942195393675e-05,
      "loss": 0.734,
      "step": 793
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.784098744392395,
      "learning_rate": 1.997929113747357e-05,
      "loss": 0.7704,
      "step": 794
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0452747344970703,
      "learning_rate": 1.997915990695188e-05,
      "loss": 0.7173,
      "step": 795
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9236286878585815,
      "learning_rate": 1.997902826237712e-05,
      "loss": 0.7289,
      "step": 796
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8562978506088257,
      "learning_rate": 1.997889620375475e-05,
      "loss": 0.7618,
      "step": 797
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.279607057571411,
      "learning_rate": 1.9978763731090255e-05,
      "loss": 0.7685,
      "step": 798
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7786463499069214,
      "learning_rate": 1.9978630844389125e-05,
      "loss": 0.7286,
      "step": 799
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7791590690612793,
      "learning_rate": 1.997849754365688e-05,
      "loss": 0.7156,
      "step": 800
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9929677248001099,
      "learning_rate": 1.9978363828899046e-05,
      "loss": 0.757,
      "step": 801
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9684098958969116,
      "learning_rate": 1.997822970012117e-05,
      "loss": 0.7413,
      "step": 802
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8288795948028564,
      "learning_rate": 1.997809515732883e-05,
      "loss": 0.7573,
      "step": 803
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8206950426101685,
      "learning_rate": 1.997796020052759e-05,
      "loss": 0.9757,
      "step": 804
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.509913682937622,
      "learning_rate": 1.9977824829723063e-05,
      "loss": 0.7435,
      "step": 805
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.007812023162842,
      "learning_rate": 1.997768904492086e-05,
      "loss": 0.752,
      "step": 806
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.937010407447815,
      "learning_rate": 1.9977552846126617e-05,
      "loss": 0.7479,
      "step": 807
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1029374599456787,
      "learning_rate": 1.9977416233345983e-05,
      "loss": 0.7867,
      "step": 808
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9607226848602295,
      "learning_rate": 1.997727920658463e-05,
      "loss": 0.7111,
      "step": 809
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.943255066871643,
      "learning_rate": 1.9977141765848235e-05,
      "loss": 0.7216,
      "step": 810
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0469107627868652,
      "learning_rate": 1.9977003911142513e-05,
      "loss": 0.7732,
      "step": 811
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9637432098388672,
      "learning_rate": 1.9976865642473175e-05,
      "loss": 0.784,
      "step": 812
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.916869878768921,
      "learning_rate": 1.997672695984596e-05,
      "loss": 0.7322,
      "step": 813
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.976666808128357,
      "learning_rate": 1.9976587863266624e-05,
      "loss": 0.7623,
      "step": 814
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2293312549591064,
      "learning_rate": 1.9976448352740937e-05,
      "loss": 0.7922,
      "step": 815
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.858865737915039,
      "learning_rate": 1.997630842827469e-05,
      "loss": 0.7575,
      "step": 816
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.013551712036133,
      "learning_rate": 1.997616808987368e-05,
      "loss": 0.7646,
      "step": 817
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.782284140586853,
      "learning_rate": 1.9976027337543745e-05,
      "loss": 0.7266,
      "step": 818
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.741034507751465,
      "learning_rate": 1.997588617129071e-05,
      "loss": 1.0842,
      "step": 819
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.8446693420410156,
      "learning_rate": 1.997574459112044e-05,
      "loss": 0.7503,
      "step": 820
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.510265588760376,
      "learning_rate": 1.9975602597038807e-05,
      "loss": 0.7867,
      "step": 821
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3621230125427246,
      "learning_rate": 1.997546018905171e-05,
      "loss": 0.7133,
      "step": 822
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0907812118530273,
      "learning_rate": 1.9975317367165044e-05,
      "loss": 0.7601,
      "step": 823
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2804391384124756,
      "learning_rate": 1.997517413138474e-05,
      "loss": 0.7199,
      "step": 824
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3163959980010986,
      "learning_rate": 1.997503048171675e-05,
      "loss": 0.7539,
      "step": 825
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7089914083480835,
      "learning_rate": 1.9974886418167026e-05,
      "loss": 0.7627,
      "step": 826
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.5331625938415527,
      "learning_rate": 1.9974741940741546e-05,
      "loss": 0.7177,
      "step": 827
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.259533166885376,
      "learning_rate": 1.9974597049446306e-05,
      "loss": 0.7403,
      "step": 828
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0095624923706055,
      "learning_rate": 1.997445174428732e-05,
      "loss": 0.7314,
      "step": 829
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1842703819274902,
      "learning_rate": 1.997430602527061e-05,
      "loss": 0.7319,
      "step": 830
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8830220699310303,
      "learning_rate": 1.9974159892402228e-05,
      "loss": 0.7348,
      "step": 831
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.026411771774292,
      "learning_rate": 1.9974013345688238e-05,
      "loss": 0.7649,
      "step": 832
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0077106952667236,
      "learning_rate": 1.997386638513472e-05,
      "loss": 0.7061,
      "step": 833
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9341241121292114,
      "learning_rate": 1.9973719010747764e-05,
      "loss": 0.8102,
      "step": 834
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9839916229248047,
      "learning_rate": 1.9973571222533497e-05,
      "loss": 0.7157,
      "step": 835
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8255298137664795,
      "learning_rate": 1.9973423020498042e-05,
      "loss": 0.7391,
      "step": 836
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.984593391418457,
      "learning_rate": 1.9973274404647554e-05,
      "loss": 0.6805,
      "step": 837
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.736088514328003,
      "learning_rate": 1.9973125374988193e-05,
      "loss": 0.7467,
      "step": 838
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0024731159210205,
      "learning_rate": 1.9972975931526146e-05,
      "loss": 0.7468,
      "step": 839
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.991682767868042,
      "learning_rate": 1.9972826074267612e-05,
      "loss": 0.7627,
      "step": 840
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.023413896560669,
      "learning_rate": 1.9972675803218814e-05,
      "loss": 0.7029,
      "step": 841
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.682800054550171,
      "learning_rate": 1.9972525118385984e-05,
      "loss": 1.0317,
      "step": 842
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.7925219535827637,
      "learning_rate": 1.997237401977537e-05,
      "loss": 0.7426,
      "step": 843
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.284316062927246,
      "learning_rate": 1.997222250739325e-05,
      "loss": 0.7698,
      "step": 844
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9151700735092163,
      "learning_rate": 1.99720705812459e-05,
      "loss": 0.7356,
      "step": 845
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.263493299484253,
      "learning_rate": 1.9971918241339633e-05,
      "loss": 0.7277,
      "step": 846
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.120354413986206,
      "learning_rate": 1.9971765487680766e-05,
      "loss": 0.6982,
      "step": 847
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.210435390472412,
      "learning_rate": 1.9971612320275632e-05,
      "loss": 0.7339,
      "step": 848
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.240034580230713,
      "learning_rate": 1.9971458739130598e-05,
      "loss": 0.7164,
      "step": 849
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9476386308670044,
      "learning_rate": 1.9971304744252025e-05,
      "loss": 0.7741,
      "step": 850
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.045933246612549,
      "learning_rate": 1.997115033564631e-05,
      "loss": 0.7761,
      "step": 851
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0376675128936768,
      "learning_rate": 1.9970995513319854e-05,
      "loss": 0.7214,
      "step": 852
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9172693490982056,
      "learning_rate": 1.9970840277279086e-05,
      "loss": 0.7729,
      "step": 853
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9922254085540771,
      "learning_rate": 1.9970684627530445e-05,
      "loss": 0.7382,
      "step": 854
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.052238941192627,
      "learning_rate": 1.997052856408039e-05,
      "loss": 0.7114,
      "step": 855
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9048177003860474,
      "learning_rate": 1.9970372086935394e-05,
      "loss": 0.7372,
      "step": 856
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9175186157226562,
      "learning_rate": 1.9970215196101946e-05,
      "loss": 0.7742,
      "step": 857
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9522300958633423,
      "learning_rate": 1.9970057891586565e-05,
      "loss": 0.7273,
      "step": 858
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.783277988433838,
      "learning_rate": 1.9969900173395772e-05,
      "loss": 0.7635,
      "step": 859
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8465049266815186,
      "learning_rate": 1.996974204153611e-05,
      "loss": 0.7172,
      "step": 860
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.776911735534668,
      "learning_rate": 1.996958349601415e-05,
      "loss": 0.7273,
      "step": 861
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7817169427871704,
      "learning_rate": 1.9969424536836456e-05,
      "loss": 0.7707,
      "step": 862
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8869725465774536,
      "learning_rate": 1.9969265164009634e-05,
      "loss": 0.6883,
      "step": 863
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8741741180419922,
      "learning_rate": 1.996910537754029e-05,
      "loss": 0.7662,
      "step": 864
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6656092405319214,
      "learning_rate": 1.9968945177435057e-05,
      "loss": 0.7915,
      "step": 865
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0183839797973633,
      "learning_rate": 1.9968784563700586e-05,
      "loss": 0.7288,
      "step": 866
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8870784044265747,
      "learning_rate": 1.9968623536343532e-05,
      "loss": 0.7698,
      "step": 867
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8492509126663208,
      "learning_rate": 1.9968462095370583e-05,
      "loss": 0.7758,
      "step": 868
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8030670881271362,
      "learning_rate": 1.9968300240788438e-05,
      "loss": 0.7281,
      "step": 869
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9397461414337158,
      "learning_rate": 1.9968137972603804e-05,
      "loss": 0.7042,
      "step": 870
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9238841533660889,
      "learning_rate": 1.9967975290823424e-05,
      "loss": 0.731,
      "step": 871
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.877073884010315,
      "learning_rate": 1.9967812195454047e-05,
      "loss": 0.7439,
      "step": 872
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8078904151916504,
      "learning_rate": 1.9967648686502433e-05,
      "loss": 0.707,
      "step": 873
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9665498733520508,
      "learning_rate": 1.996748476397537e-05,
      "loss": 0.7666,
      "step": 874
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.866026759147644,
      "learning_rate": 1.9967320427879665e-05,
      "loss": 0.7583,
      "step": 875
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7067965269088745,
      "learning_rate": 1.9967155678222125e-05,
      "loss": 0.7639,
      "step": 876
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7440844774246216,
      "learning_rate": 1.9966990515009598e-05,
      "loss": 0.7162,
      "step": 877
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.730533242225647,
      "learning_rate": 1.9966824938248925e-05,
      "loss": 0.7756,
      "step": 878
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.297884702682495,
      "learning_rate": 1.9966658947946987e-05,
      "loss": 1.0012,
      "step": 879
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1380884647369385,
      "learning_rate": 1.9966492544110666e-05,
      "loss": 0.7335,
      "step": 880
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.051866054534912,
      "learning_rate": 1.9966325726746866e-05,
      "loss": 0.8198,
      "step": 881
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7198278903961182,
      "learning_rate": 1.996615849586251e-05,
      "loss": 0.7294,
      "step": 882
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8123185634613037,
      "learning_rate": 1.9965990851464533e-05,
      "loss": 0.7443,
      "step": 883
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8580708503723145,
      "learning_rate": 1.9965822793559898e-05,
      "loss": 0.7463,
      "step": 884
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7711265087127686,
      "learning_rate": 1.9965654322155573e-05,
      "loss": 0.7406,
      "step": 885
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.925531268119812,
      "learning_rate": 1.996548543725855e-05,
      "loss": 0.7148,
      "step": 886
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.098552942276001,
      "learning_rate": 1.9965316138875833e-05,
      "loss": 0.7017,
      "step": 887
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7817466259002686,
      "learning_rate": 1.996514642701445e-05,
      "loss": 0.7326,
      "step": 888
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7596893310546875,
      "learning_rate": 1.9964976301681444e-05,
      "loss": 0.6786,
      "step": 889
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8696367740631104,
      "learning_rate": 1.9964805762883866e-05,
      "loss": 0.684,
      "step": 890
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.888334035873413,
      "learning_rate": 1.99646348106288e-05,
      "loss": 0.717,
      "step": 891
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7222703695297241,
      "learning_rate": 1.9964463444923338e-05,
      "loss": 0.73,
      "step": 892
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9279762506484985,
      "learning_rate": 1.9964291665774588e-05,
      "loss": 0.8108,
      "step": 893
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.864623785018921,
      "learning_rate": 1.996411947318968e-05,
      "loss": 0.731,
      "step": 894
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9253531694412231,
      "learning_rate": 1.9963946867175753e-05,
      "loss": 0.7266,
      "step": 895
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8100961446762085,
      "learning_rate": 1.9963773847739975e-05,
      "loss": 0.684,
      "step": 896
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8156784772872925,
      "learning_rate": 1.9963600414889524e-05,
      "loss": 0.6958,
      "step": 897
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7894192934036255,
      "learning_rate": 1.9963426568631592e-05,
      "loss": 0.7087,
      "step": 898
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.134014368057251,
      "learning_rate": 1.9963252308973392e-05,
      "loss": 0.9779,
      "step": 899
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3793344497680664,
      "learning_rate": 1.9963077635922165e-05,
      "loss": 0.7265,
      "step": 900
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.213989496231079,
      "learning_rate": 1.9962902549485144e-05,
      "loss": 0.9877,
      "step": 901
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.434697151184082,
      "learning_rate": 1.99627270496696e-05,
      "loss": 0.7024,
      "step": 902
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9809563159942627,
      "learning_rate": 1.996255113648282e-05,
      "loss": 0.7253,
      "step": 903
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5885270833969116,
      "learning_rate": 1.9962374809932094e-05,
      "loss": 0.7231,
      "step": 904
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8659696578979492,
      "learning_rate": 1.996219807002474e-05,
      "loss": 0.7249,
      "step": 905
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8096747398376465,
      "learning_rate": 1.9962020916768097e-05,
      "loss": 0.8086,
      "step": 906
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7960237264633179,
      "learning_rate": 1.996184335016951e-05,
      "loss": 0.7227,
      "step": 907
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8209577798843384,
      "learning_rate": 1.9961665370236346e-05,
      "loss": 0.7593,
      "step": 908
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9027109146118164,
      "learning_rate": 1.9961486976975998e-05,
      "loss": 0.7525,
      "step": 909
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8407354354858398,
      "learning_rate": 1.9961308170395856e-05,
      "loss": 0.7027,
      "step": 910
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8711949586868286,
      "learning_rate": 1.9961128950503345e-05,
      "loss": 0.7318,
      "step": 911
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8899303674697876,
      "learning_rate": 1.9960949317305906e-05,
      "loss": 0.7278,
      "step": 912
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9099665880203247,
      "learning_rate": 1.996076927081098e-05,
      "loss": 0.6901,
      "step": 913
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7329903841018677,
      "learning_rate": 1.9960588811026048e-05,
      "loss": 0.7272,
      "step": 914
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7175817489624023,
      "learning_rate": 1.9960407937958595e-05,
      "loss": 0.7513,
      "step": 915
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.962084174156189,
      "learning_rate": 1.9960226651616124e-05,
      "loss": 0.7633,
      "step": 916
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7863850593566895,
      "learning_rate": 1.9960044952006163e-05,
      "loss": 0.7208,
      "step": 917
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7203221321105957,
      "learning_rate": 1.995986283913624e-05,
      "loss": 0.7366,
      "step": 918
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.783231496810913,
      "learning_rate": 1.9959680313013917e-05,
      "loss": 0.755,
      "step": 919
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8982830047607422,
      "learning_rate": 1.995949737364677e-05,
      "loss": 0.7339,
      "step": 920
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7600499391555786,
      "learning_rate": 1.9959314021042388e-05,
      "loss": 0.7496,
      "step": 921
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8402092456817627,
      "learning_rate": 1.9959130255208373e-05,
      "loss": 0.7098,
      "step": 922
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9995843172073364,
      "learning_rate": 1.995894607615236e-05,
      "loss": 0.7314,
      "step": 923
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.468781471252441,
      "learning_rate": 1.995876148388198e-05,
      "loss": 1.108,
      "step": 924
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2994613647460938,
      "learning_rate": 1.99585764784049e-05,
      "loss": 0.7325,
      "step": 925
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0500998497009277,
      "learning_rate": 1.9958391059728793e-05,
      "loss": 0.7781,
      "step": 926
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.10129976272583,
      "learning_rate": 1.995820522786135e-05,
      "loss": 0.7895,
      "step": 927
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.241117238998413,
      "learning_rate": 1.995801898281029e-05,
      "loss": 0.7445,
      "step": 928
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.918009638786316,
      "learning_rate": 1.9957832324583332e-05,
      "loss": 0.7563,
      "step": 929
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1538913249969482,
      "learning_rate": 1.9957645253188224e-05,
      "loss": 0.7204,
      "step": 930
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.080573320388794,
      "learning_rate": 1.9957457768632725e-05,
      "loss": 0.7369,
      "step": 931
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8393813371658325,
      "learning_rate": 1.995726987092462e-05,
      "loss": 0.7459,
      "step": 932
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.30720591545105,
      "learning_rate": 1.99570815600717e-05,
      "loss": 0.7229,
      "step": 933
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9870439767837524,
      "learning_rate": 1.995689283608178e-05,
      "loss": 0.7387,
      "step": 934
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7307212352752686,
      "learning_rate": 1.9956703698962694e-05,
      "loss": 0.7733,
      "step": 935
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8170180320739746,
      "learning_rate": 1.9956514148722285e-05,
      "loss": 0.7649,
      "step": 936
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6472808122634888,
      "learning_rate": 1.995632418536842e-05,
      "loss": 0.7255,
      "step": 937
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.797399640083313,
      "learning_rate": 1.9956133808908977e-05,
      "loss": 0.7252,
      "step": 938
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7415478229522705,
      "learning_rate": 1.995594301935186e-05,
      "loss": 0.7329,
      "step": 939
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6580792665481567,
      "learning_rate": 1.9955751816704986e-05,
      "loss": 0.7272,
      "step": 940
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7706433534622192,
      "learning_rate": 1.9955560200976283e-05,
      "loss": 0.7599,
      "step": 941
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6839679479599,
      "learning_rate": 1.9955368172173705e-05,
      "loss": 0.7474,
      "step": 942
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7939265966415405,
      "learning_rate": 1.9955175730305217e-05,
      "loss": 0.7197,
      "step": 943
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7833211421966553,
      "learning_rate": 1.9954982875378805e-05,
      "loss": 0.7365,
      "step": 944
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.769425630569458,
      "learning_rate": 1.9954789607402475e-05,
      "loss": 0.7212,
      "step": 945
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7674622535705566,
      "learning_rate": 1.995459592638424e-05,
      "loss": 0.75,
      "step": 946
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6868623495101929,
      "learning_rate": 1.995440183233214e-05,
      "loss": 0.7384,
      "step": 947
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.660934567451477,
      "learning_rate": 1.9954207325254223e-05,
      "loss": 0.6872,
      "step": 948
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7372002601623535,
      "learning_rate": 1.9954012405158568e-05,
      "loss": 0.7287,
      "step": 949
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6880159378051758,
      "learning_rate": 1.9953817072053256e-05,
      "loss": 0.7503,
      "step": 950
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.925756573677063,
      "learning_rate": 1.9953621325946395e-05,
      "loss": 0.7543,
      "step": 951
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7335216999053955,
      "learning_rate": 1.9953425166846107e-05,
      "loss": 0.719,
      "step": 952
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.812554955482483,
      "learning_rate": 1.995322859476053e-05,
      "loss": 0.7033,
      "step": 953
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.736259937286377,
      "learning_rate": 1.9953031609697814e-05,
      "loss": 0.7363,
      "step": 954
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6183650493621826,
      "learning_rate": 1.995283421166614e-05,
      "loss": 0.6764,
      "step": 955
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8283016681671143,
      "learning_rate": 1.99526364006737e-05,
      "loss": 0.7501,
      "step": 956
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6869041919708252,
      "learning_rate": 1.9952438176728694e-05,
      "loss": 0.7176,
      "step": 957
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.972854733467102,
      "learning_rate": 1.9952239539839354e-05,
      "loss": 0.7338,
      "step": 958
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7497997283935547,
      "learning_rate": 1.9952040490013914e-05,
      "loss": 0.746,
      "step": 959
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7949885129928589,
      "learning_rate": 1.995184102726064e-05,
      "loss": 1.0095,
      "step": 960
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.483158588409424,
      "learning_rate": 1.9951641151587804e-05,
      "loss": 0.7245,
      "step": 961
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9579095840454102,
      "learning_rate": 1.9951440863003702e-05,
      "loss": 0.7283,
      "step": 962
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8887386322021484,
      "learning_rate": 1.9951240161516643e-05,
      "loss": 0.7407,
      "step": 963
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.943416714668274,
      "learning_rate": 1.9951039047134953e-05,
      "loss": 0.76,
      "step": 964
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.894146203994751,
      "learning_rate": 1.995083751986698e-05,
      "loss": 0.7578,
      "step": 965
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9487531185150146,
      "learning_rate": 1.995063557972108e-05,
      "loss": 0.7606,
      "step": 966
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7991834878921509,
      "learning_rate": 1.9950433226705636e-05,
      "loss": 0.7225,
      "step": 967
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.792020559310913,
      "learning_rate": 1.9950230460829047e-05,
      "loss": 0.7085,
      "step": 968
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9471650123596191,
      "learning_rate": 1.995002728209972e-05,
      "loss": 0.7437,
      "step": 969
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7360413074493408,
      "learning_rate": 1.994982369052609e-05,
      "loss": 0.6888,
      "step": 970
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8491721153259277,
      "learning_rate": 1.9949619686116597e-05,
      "loss": 0.7134,
      "step": 971
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.86418616771698,
      "learning_rate": 1.9949415268879716e-05,
      "loss": 0.6987,
      "step": 972
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8025037050247192,
      "learning_rate": 1.9949210438823924e-05,
      "loss": 0.7509,
      "step": 973
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6739329099655151,
      "learning_rate": 1.9949005195957717e-05,
      "loss": 0.7309,
      "step": 974
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.70960533618927,
      "learning_rate": 1.9948799540289613e-05,
      "loss": 0.7013,
      "step": 975
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.796212077140808,
      "learning_rate": 1.9948593471828145e-05,
      "loss": 0.7005,
      "step": 976
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8372408151626587,
      "learning_rate": 1.9948386990581862e-05,
      "loss": 0.7247,
      "step": 977
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6899114847183228,
      "learning_rate": 1.9948180096559332e-05,
      "loss": 0.7293,
      "step": 978
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8101073503494263,
      "learning_rate": 1.9947972789769142e-05,
      "loss": 0.6961,
      "step": 979
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.025801181793213,
      "learning_rate": 1.994776507021989e-05,
      "loss": 0.7974,
      "step": 980
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9375996589660645,
      "learning_rate": 1.9947556937920197e-05,
      "loss": 0.7134,
      "step": 981
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9151721000671387,
      "learning_rate": 1.9947348392878698e-05,
      "loss": 0.7051,
      "step": 982
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8962702751159668,
      "learning_rate": 1.9947139435104044e-05,
      "loss": 0.7211,
      "step": 983
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9453188180923462,
      "learning_rate": 1.994693006460491e-05,
      "loss": 0.7297,
      "step": 984
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9841904640197754,
      "learning_rate": 1.9946720281389974e-05,
      "loss": 0.7291,
      "step": 985
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.870995283126831,
      "learning_rate": 1.9946510085467952e-05,
      "loss": 0.6761,
      "step": 986
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.823188304901123,
      "learning_rate": 1.9946299476847558e-05,
      "loss": 0.6939,
      "step": 987
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.864479422569275,
      "learning_rate": 1.994608845553753e-05,
      "loss": 0.7548,
      "step": 988
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9139070510864258,
      "learning_rate": 1.9945877021546626e-05,
      "loss": 0.7218,
      "step": 989
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8522453308105469,
      "learning_rate": 1.9945665174883622e-05,
      "loss": 0.7362,
      "step": 990
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7367916107177734,
      "learning_rate": 1.9945452915557302e-05,
      "loss": 0.7151,
      "step": 991
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9385919570922852,
      "learning_rate": 1.9945240243576477e-05,
      "loss": 0.709,
      "step": 992
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.75130033493042,
      "learning_rate": 1.9945027158949964e-05,
      "loss": 0.6887,
      "step": 993
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7045451402664185,
      "learning_rate": 1.994481366168662e-05,
      "loss": 0.7341,
      "step": 994
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8609263896942139,
      "learning_rate": 1.9944599751795283e-05,
      "loss": 0.6722,
      "step": 995
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.887115716934204,
      "learning_rate": 1.9944385429284843e-05,
      "loss": 0.7316,
      "step": 996
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.776916265487671,
      "learning_rate": 1.9944170694164187e-05,
      "loss": 0.7088,
      "step": 997
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0630736351013184,
      "learning_rate": 1.994395554644223e-05,
      "loss": 0.7095,
      "step": 998
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.846164584159851,
      "learning_rate": 1.9943739986127894e-05,
      "loss": 0.7197,
      "step": 999
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.845771312713623,
      "learning_rate": 1.994352401323012e-05,
      "loss": 0.6998,
      "step": 1000
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3738410472869873,
      "learning_rate": 1.9943307627757874e-05,
      "loss": 1.0039,
      "step": 1001
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.715787887573242,
      "learning_rate": 1.9943090829720134e-05,
      "loss": 0.7459,
      "step": 1002
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9836925268173218,
      "learning_rate": 1.9942873619125896e-05,
      "loss": 0.7039,
      "step": 1003
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8893649578094482,
      "learning_rate": 1.994265599598417e-05,
      "loss": 0.7079,
      "step": 1004
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1455912590026855,
      "learning_rate": 1.9942437960303985e-05,
      "loss": 0.7193,
      "step": 1005
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8761147260665894,
      "learning_rate": 1.9942219512094394e-05,
      "loss": 0.6919,
      "step": 1006
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7849973440170288,
      "learning_rate": 1.994200065136445e-05,
      "loss": 0.7342,
      "step": 1007
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.003943920135498,
      "learning_rate": 1.9941781378123244e-05,
      "loss": 0.7481,
      "step": 1008
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9360045194625854,
      "learning_rate": 1.9941561692379868e-05,
      "loss": 0.7058,
      "step": 1009
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.96418035030365,
      "learning_rate": 1.9941341594143438e-05,
      "loss": 0.69,
      "step": 1010
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.008863687515259,
      "learning_rate": 1.9941121083423095e-05,
      "loss": 0.7175,
      "step": 1011
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7190914154052734,
      "learning_rate": 1.9940900160227975e-05,
      "loss": 0.7141,
      "step": 1012
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.784574270248413,
      "learning_rate": 1.994067882456725e-05,
      "loss": 0.7303,
      "step": 1013
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7118414640426636,
      "learning_rate": 1.9940457076450103e-05,
      "loss": 0.7016,
      "step": 1014
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7373729944229126,
      "learning_rate": 1.994023491588574e-05,
      "loss": 0.6932,
      "step": 1015
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8056803941726685,
      "learning_rate": 1.9940012342883367e-05,
      "loss": 0.7367,
      "step": 1016
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7013905048370361,
      "learning_rate": 1.9939789357452235e-05,
      "loss": 0.659,
      "step": 1017
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.835617184638977,
      "learning_rate": 1.9939565959601582e-05,
      "loss": 0.7099,
      "step": 1018
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.711478352546692,
      "learning_rate": 1.9939342149340683e-05,
      "loss": 0.7398,
      "step": 1019
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7028508186340332,
      "learning_rate": 1.993911792667883e-05,
      "loss": 0.6863,
      "step": 1020
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7051948308944702,
      "learning_rate": 1.9938893291625313e-05,
      "loss": 0.708,
      "step": 1021
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7580265998840332,
      "learning_rate": 1.9938668244189463e-05,
      "loss": 0.6789,
      "step": 1022
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7794114351272583,
      "learning_rate": 1.9938442784380614e-05,
      "loss": 0.7723,
      "step": 1023
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8171879053115845,
      "learning_rate": 1.993821691220812e-05,
      "loss": 0.7173,
      "step": 1024
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8328484296798706,
      "learning_rate": 1.9937990627681355e-05,
      "loss": 0.6911,
      "step": 1025
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6428066492080688,
      "learning_rate": 1.993776393080971e-05,
      "loss": 0.6914,
      "step": 1026
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7972228527069092,
      "learning_rate": 1.9937536821602584e-05,
      "loss": 0.7165,
      "step": 1027
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7547940015792847,
      "learning_rate": 1.9937309300069405e-05,
      "loss": 0.6953,
      "step": 1028
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7074506282806396,
      "learning_rate": 1.993708136621961e-05,
      "loss": 0.7609,
      "step": 1029
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7222988605499268,
      "learning_rate": 1.9936853020062667e-05,
      "loss": 0.6878,
      "step": 1030
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7882771492004395,
      "learning_rate": 1.9936624261608037e-05,
      "loss": 0.6609,
      "step": 1031
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.716368317604065,
      "learning_rate": 1.9936395090865216e-05,
      "loss": 0.7059,
      "step": 1032
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8519593477249146,
      "learning_rate": 1.9936165507843714e-05,
      "loss": 0.7196,
      "step": 1033
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7900131940841675,
      "learning_rate": 1.9935935512553056e-05,
      "loss": 0.6845,
      "step": 1034
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9458123445510864,
      "learning_rate": 1.9935705105002788e-05,
      "loss": 0.7356,
      "step": 1035
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6986628770828247,
      "learning_rate": 1.9935474285202466e-05,
      "loss": 0.7075,
      "step": 1036
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7397241592407227,
      "learning_rate": 1.993524305316167e-05,
      "loss": 0.7112,
      "step": 1037
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7655725479125977,
      "learning_rate": 1.993501140888999e-05,
      "loss": 0.7259,
      "step": 1038
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8606666326522827,
      "learning_rate": 1.9934779352397043e-05,
      "loss": 0.7309,
      "step": 1039
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7802685499191284,
      "learning_rate": 1.9934546883692455e-05,
      "loss": 0.7689,
      "step": 1040
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7928411960601807,
      "learning_rate": 1.9934314002785866e-05,
      "loss": 0.6692,
      "step": 1041
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7447739839553833,
      "learning_rate": 1.9934080709686946e-05,
      "loss": 0.6985,
      "step": 1042
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6595523357391357,
      "learning_rate": 1.9933847004405375e-05,
      "loss": 0.7104,
      "step": 1043
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7089827060699463,
      "learning_rate": 1.9933612886950847e-05,
      "loss": 0.7241,
      "step": 1044
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.60512375831604,
      "learning_rate": 1.9933378357333074e-05,
      "loss": 0.6817,
      "step": 1045
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6799592971801758,
      "learning_rate": 1.993314341556179e-05,
      "loss": 0.6684,
      "step": 1046
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6326543092727661,
      "learning_rate": 1.993290806164674e-05,
      "loss": 0.7621,
      "step": 1047
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6577560901641846,
      "learning_rate": 1.9932672295597697e-05,
      "loss": 0.721,
      "step": 1048
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.715214729309082,
      "learning_rate": 1.9932436117424436e-05,
      "loss": 0.7018,
      "step": 1049
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7143826484680176,
      "learning_rate": 1.9932199527136754e-05,
      "loss": 0.6678,
      "step": 1050
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6826939582824707,
      "learning_rate": 1.993196252474448e-05,
      "loss": 0.7134,
      "step": 1051
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6391438245773315,
      "learning_rate": 1.9931725110257434e-05,
      "loss": 0.6567,
      "step": 1052
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5909743309020996,
      "learning_rate": 1.9931487283685474e-05,
      "loss": 0.7247,
      "step": 1053
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5968847274780273,
      "learning_rate": 1.9931249045038467e-05,
      "loss": 0.7125,
      "step": 1054
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7686277627944946,
      "learning_rate": 1.9931010394326295e-05,
      "loss": 0.7056,
      "step": 1055
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7201720476150513,
      "learning_rate": 1.9930771331558863e-05,
      "loss": 0.6812,
      "step": 1056
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7102398872375488,
      "learning_rate": 1.993053185674609e-05,
      "loss": 0.6617,
      "step": 1057
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8178929090499878,
      "learning_rate": 1.9930291969897912e-05,
      "loss": 0.6889,
      "step": 1058
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7271380424499512,
      "learning_rate": 1.993005167102428e-05,
      "loss": 0.7059,
      "step": 1059
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.830517292022705,
      "learning_rate": 1.992981096013517e-05,
      "loss": 0.7391,
      "step": 1060
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7467973232269287,
      "learning_rate": 1.9929569837240567e-05,
      "loss": 0.7054,
      "step": 1061
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.675822377204895,
      "learning_rate": 1.992932830235047e-05,
      "loss": 0.6964,
      "step": 1062
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9146062135696411,
      "learning_rate": 1.9929086355474907e-05,
      "loss": 0.725,
      "step": 1063
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7953640222549438,
      "learning_rate": 1.9928843996623915e-05,
      "loss": 0.7139,
      "step": 1064
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6415828466415405,
      "learning_rate": 1.992860122580755e-05,
      "loss": 0.6778,
      "step": 1065
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.791130781173706,
      "learning_rate": 1.9928358043035886e-05,
      "loss": 0.749,
      "step": 1066
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.709596872329712,
      "learning_rate": 1.992811444831901e-05,
      "loss": 0.7294,
      "step": 1067
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3470585346221924,
      "learning_rate": 1.9927870441667032e-05,
      "loss": 1.0078,
      "step": 1068
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.681222677230835,
      "learning_rate": 1.9927626023090077e-05,
      "loss": 0.7366,
      "step": 1069
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0122649669647217,
      "learning_rate": 1.9927381192598283e-05,
      "loss": 0.7095,
      "step": 1070
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.820919156074524,
      "learning_rate": 1.992713595020181e-05,
      "loss": 0.7178,
      "step": 1071
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.328462600708008,
      "learning_rate": 1.9926890295910835e-05,
      "loss": 0.7275,
      "step": 1072
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.712049126625061,
      "learning_rate": 1.992664422973555e-05,
      "loss": 0.7391,
      "step": 1073
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0660293102264404,
      "learning_rate": 1.9926397751686163e-05,
      "loss": 0.7086,
      "step": 1074
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8089196681976318,
      "learning_rate": 1.9926150861772902e-05,
      "loss": 0.6947,
      "step": 1075
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7342134714126587,
      "learning_rate": 1.992590356000601e-05,
      "loss": 0.7266,
      "step": 1076
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9171006679534912,
      "learning_rate": 1.992565584639575e-05,
      "loss": 0.7142,
      "step": 1077
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7669641971588135,
      "learning_rate": 1.99254077209524e-05,
      "loss": 0.7025,
      "step": 1078
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9420307874679565,
      "learning_rate": 1.9925159183686252e-05,
      "loss": 0.7242,
      "step": 1079
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7250416278839111,
      "learning_rate": 1.9924910234607625e-05,
      "loss": 0.6633,
      "step": 1080
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6269935369491577,
      "learning_rate": 1.992466087372684e-05,
      "loss": 0.7319,
      "step": 1081
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8808445930480957,
      "learning_rate": 1.9924411101054248e-05,
      "loss": 0.708,
      "step": 1082
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6695414781570435,
      "learning_rate": 1.992416091660021e-05,
      "loss": 0.7413,
      "step": 1083
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6806305646896362,
      "learning_rate": 1.992391032037511e-05,
      "loss": 0.6887,
      "step": 1084
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.820913314819336,
      "learning_rate": 1.9923659312389344e-05,
      "loss": 0.7089,
      "step": 1085
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6342387199401855,
      "learning_rate": 1.9923407892653327e-05,
      "loss": 0.6918,
      "step": 1086
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6418116092681885,
      "learning_rate": 1.9923156061177492e-05,
      "loss": 0.6661,
      "step": 1087
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7308080196380615,
      "learning_rate": 1.9922903817972284e-05,
      "loss": 0.7575,
      "step": 1088
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7493923902511597,
      "learning_rate": 1.992265116304817e-05,
      "loss": 0.7209,
      "step": 1089
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6398024559020996,
      "learning_rate": 1.9922398096415638e-05,
      "loss": 0.6875,
      "step": 1090
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8498406410217285,
      "learning_rate": 1.9922144618085184e-05,
      "loss": 0.9983,
      "step": 1091
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2551567554473877,
      "learning_rate": 1.9921890728067325e-05,
      "loss": 0.709,
      "step": 1092
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.24142587184906,
      "learning_rate": 1.99216364263726e-05,
      "loss": 0.991,
      "step": 1093
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7757102251052856,
      "learning_rate": 1.992138171301155e-05,
      "loss": 0.7092,
      "step": 1094
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.813212513923645,
      "learning_rate": 1.9921126587994754e-05,
      "loss": 0.6886,
      "step": 1095
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6961029767990112,
      "learning_rate": 1.9920871051332794e-05,
      "loss": 0.6895,
      "step": 1096
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6703981161117554,
      "learning_rate": 1.992061510303627e-05,
      "loss": 0.6955,
      "step": 1097
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7539644241333008,
      "learning_rate": 1.992035874311581e-05,
      "loss": 0.7271,
      "step": 1098
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7287245988845825,
      "learning_rate": 1.9920101971582043e-05,
      "loss": 0.6927,
      "step": 1099
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.741894245147705,
      "learning_rate": 1.9919844788445622e-05,
      "loss": 0.6742,
      "step": 1100
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8309617042541504,
      "learning_rate": 1.9919587193717227e-05,
      "loss": 0.6882,
      "step": 1101
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7746914625167847,
      "learning_rate": 1.9919329187407537e-05,
      "loss": 0.6991,
      "step": 1102
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.687551736831665,
      "learning_rate": 1.991907076952726e-05,
      "loss": 0.7438,
      "step": 1103
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7001206874847412,
      "learning_rate": 1.9918811940087116e-05,
      "loss": 0.685,
      "step": 1104
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7158223390579224,
      "learning_rate": 1.991855269909785e-05,
      "loss": 0.6854,
      "step": 1105
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.707594633102417,
      "learning_rate": 1.9918293046570217e-05,
      "loss": 0.7409,
      "step": 1106
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5733925104141235,
      "learning_rate": 1.9918032982514982e-05,
      "loss": 0.7506,
      "step": 1107
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.874480962753296,
      "learning_rate": 1.991777250694295e-05,
      "loss": 0.7075,
      "step": 1108
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.811688780784607,
      "learning_rate": 1.9917511619864915e-05,
      "loss": 0.7102,
      "step": 1109
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5921657085418701,
      "learning_rate": 1.991725032129171e-05,
      "loss": 0.7423,
      "step": 1110
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.799657940864563,
      "learning_rate": 1.9916988611234174e-05,
      "loss": 0.6905,
      "step": 1111
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.624748945236206,
      "learning_rate": 1.9916726489703167e-05,
      "loss": 0.6981,
      "step": 1112
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.715537190437317,
      "learning_rate": 1.991646395670956e-05,
      "loss": 0.6932,
      "step": 1113
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7980438470840454,
      "learning_rate": 1.9916201012264255e-05,
      "loss": 0.689,
      "step": 1114
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.686302661895752,
      "learning_rate": 1.991593765637815e-05,
      "loss": 0.6659,
      "step": 1115
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8827182054519653,
      "learning_rate": 1.9915673889062183e-05,
      "loss": 0.6929,
      "step": 1116
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.765854835510254,
      "learning_rate": 1.9915409710327295e-05,
      "loss": 0.7435,
      "step": 1117
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6440256834030151,
      "learning_rate": 1.9915145120184446e-05,
      "loss": 0.7068,
      "step": 1118
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9014767408370972,
      "learning_rate": 1.9914880118644616e-05,
      "loss": 0.7216,
      "step": 1119
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7844364643096924,
      "learning_rate": 1.9914614705718794e-05,
      "loss": 0.7148,
      "step": 1120
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.718906044960022,
      "learning_rate": 1.9914348881418e-05,
      "loss": 0.6868,
      "step": 1121
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7358835935592651,
      "learning_rate": 1.9914082645753264e-05,
      "loss": 0.7102,
      "step": 1122
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.74891197681427,
      "learning_rate": 1.991381599873563e-05,
      "loss": 0.7488,
      "step": 1123
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7943027019500732,
      "learning_rate": 1.9913548940376157e-05,
      "loss": 0.7119,
      "step": 1124
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6350860595703125,
      "learning_rate": 1.9913281470685932e-05,
      "loss": 0.7265,
      "step": 1125
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0054242610931396,
      "learning_rate": 1.991301358967605e-05,
      "loss": 0.7416,
      "step": 1126
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.811090111732483,
      "learning_rate": 1.9912745297357628e-05,
      "loss": 0.7002,
      "step": 1127
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.726580262184143,
      "learning_rate": 1.9912476593741796e-05,
      "loss": 0.6724,
      "step": 1128
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8327456712722778,
      "learning_rate": 1.9912207478839704e-05,
      "loss": 0.6617,
      "step": 1129
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8274072408676147,
      "learning_rate": 1.9911937952662516e-05,
      "loss": 0.7116,
      "step": 1130
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7168505191802979,
      "learning_rate": 1.991166801522142e-05,
      "loss": 0.7111,
      "step": 1131
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7511694431304932,
      "learning_rate": 1.991139766652761e-05,
      "loss": 0.734,
      "step": 1132
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.701981782913208,
      "learning_rate": 1.9911126906592307e-05,
      "loss": 0.7217,
      "step": 1133
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7454859018325806,
      "learning_rate": 1.9910855735426744e-05,
      "loss": 0.7043,
      "step": 1134
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.887972116470337,
      "learning_rate": 1.9910584153042175e-05,
      "loss": 0.6906,
      "step": 1135
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7269489765167236,
      "learning_rate": 1.9910312159449864e-05,
      "loss": 0.7108,
      "step": 1136
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2422611713409424,
      "learning_rate": 1.9910039754661098e-05,
      "loss": 0.6826,
      "step": 1137
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8550527095794678,
      "learning_rate": 1.990976693868718e-05,
      "loss": 0.7067,
      "step": 1138
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.60304856300354,
      "learning_rate": 1.9909493711539433e-05,
      "loss": 0.7023,
      "step": 1139
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7570534944534302,
      "learning_rate": 1.990922007322919e-05,
      "loss": 0.6743,
      "step": 1140
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6513200998306274,
      "learning_rate": 1.9908946023767806e-05,
      "loss": 0.6943,
      "step": 1141
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5280640125274658,
      "learning_rate": 1.9908671563166652e-05,
      "loss": 0.7255,
      "step": 1142
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.708653450012207,
      "learning_rate": 1.990839669143711e-05,
      "loss": 0.6836,
      "step": 1143
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6932837963104248,
      "learning_rate": 1.9908121408590595e-05,
      "loss": 0.6593,
      "step": 1144
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7059054374694824,
      "learning_rate": 1.9907845714638522e-05,
      "loss": 0.7079,
      "step": 1145
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7097814083099365,
      "learning_rate": 1.9907569609592333e-05,
      "loss": 0.6948,
      "step": 1146
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8644039630889893,
      "learning_rate": 1.990729309346348e-05,
      "loss": 0.7116,
      "step": 1147
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9832849502563477,
      "learning_rate": 1.9907016166263445e-05,
      "loss": 0.7464,
      "step": 1148
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7853416204452515,
      "learning_rate": 1.9906738828003706e-05,
      "loss": 0.755,
      "step": 1149
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.674344539642334,
      "learning_rate": 1.990646107869578e-05,
      "loss": 0.704,
      "step": 1150
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.049628496170044,
      "learning_rate": 1.990618291835119e-05,
      "loss": 0.7026,
      "step": 1151
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6979857683181763,
      "learning_rate": 1.990590434698147e-05,
      "loss": 0.6915,
      "step": 1152
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8063737154006958,
      "learning_rate": 1.9905625364598186e-05,
      "loss": 0.6973,
      "step": 1153
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.668709635734558,
      "learning_rate": 1.9905345971212914e-05,
      "loss": 0.6834,
      "step": 1154
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6284273862838745,
      "learning_rate": 1.990506616683724e-05,
      "loss": 0.6601,
      "step": 1155
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.691644549369812,
      "learning_rate": 1.9904785951482777e-05,
      "loss": 0.6804,
      "step": 1156
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.643471360206604,
      "learning_rate": 1.990450532516116e-05,
      "loss": 0.728,
      "step": 1157
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6339305639266968,
      "learning_rate": 1.9904224287884016e-05,
      "loss": 0.6845,
      "step": 1158
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6449110507965088,
      "learning_rate": 1.9903942839663018e-05,
      "loss": 0.67,
      "step": 1159
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.064893960952759,
      "learning_rate": 1.9903660980509837e-05,
      "loss": 0.696,
      "step": 1160
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6665819883346558,
      "learning_rate": 1.9903378710436175e-05,
      "loss": 0.7009,
      "step": 1161
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6420180797576904,
      "learning_rate": 1.9903096029453736e-05,
      "loss": 0.7052,
      "step": 1162
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.598152756690979,
      "learning_rate": 1.9902812937574255e-05,
      "loss": 0.6778,
      "step": 1163
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.688475251197815,
      "learning_rate": 1.990252943480948e-05,
      "loss": 0.7065,
      "step": 1164
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.768371820449829,
      "learning_rate": 1.9902245521171163e-05,
      "loss": 0.657,
      "step": 1165
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7017220258712769,
      "learning_rate": 1.9901961196671096e-05,
      "loss": 0.7226,
      "step": 1166
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8407995700836182,
      "learning_rate": 1.990167646132107e-05,
      "loss": 0.73,
      "step": 1167
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6864280700683594,
      "learning_rate": 1.99013913151329e-05,
      "loss": 0.6928,
      "step": 1168
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6866384744644165,
      "learning_rate": 1.9901105758118417e-05,
      "loss": 0.6762,
      "step": 1169
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6310205459594727,
      "learning_rate": 1.9900819790289464e-05,
      "loss": 0.7254,
      "step": 1170
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6808377504348755,
      "learning_rate": 1.990053341165792e-05,
      "loss": 0.6614,
      "step": 1171
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6928189992904663,
      "learning_rate": 1.9900246622235657e-05,
      "loss": 0.6734,
      "step": 1172
    },
    {
      "epoch": 0.07,
      "grad_norm": 31.48361587524414,
      "learning_rate": 1.9899959422034576e-05,
      "loss": 0.72,
      "step": 1173
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1472177505493164,
      "learning_rate": 1.9899671811066596e-05,
      "loss": 0.7202,
      "step": 1174
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7789509296417236,
      "learning_rate": 1.989938378934365e-05,
      "loss": 0.7106,
      "step": 1175
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6612391471862793,
      "learning_rate": 1.9899095356877685e-05,
      "loss": 0.6922,
      "step": 1176
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0626232624053955,
      "learning_rate": 1.9898806513680667e-05,
      "loss": 0.6709,
      "step": 1177
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7897058725357056,
      "learning_rate": 1.989851725976459e-05,
      "loss": 0.7177,
      "step": 1178
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7483829259872437,
      "learning_rate": 1.989822759514145e-05,
      "loss": 0.6579,
      "step": 1179
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.824679970741272,
      "learning_rate": 1.9897937519823265e-05,
      "loss": 0.7246,
      "step": 1180
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7328813076019287,
      "learning_rate": 1.989764703382207e-05,
      "loss": 0.6774,
      "step": 1181
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.64930260181427,
      "learning_rate": 1.9897356137149918e-05,
      "loss": 0.6973,
      "step": 1182
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6350966691970825,
      "learning_rate": 1.9897064829818884e-05,
      "loss": 0.7251,
      "step": 1183
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.777754306793213,
      "learning_rate": 1.989677311184105e-05,
      "loss": 0.6787,
      "step": 1184
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6148046255111694,
      "learning_rate": 1.989648098322852e-05,
      "loss": 0.647,
      "step": 1185
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.800220012664795,
      "learning_rate": 1.989618844399342e-05,
      "loss": 0.7151,
      "step": 1186
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.633367896080017,
      "learning_rate": 1.9895895494147878e-05,
      "loss": 0.6883,
      "step": 1187
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6916463375091553,
      "learning_rate": 1.9895602133704058e-05,
      "loss": 0.6821,
      "step": 1188
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7242777347564697,
      "learning_rate": 1.9895308362674125e-05,
      "loss": 0.6903,
      "step": 1189
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9523122310638428,
      "learning_rate": 1.9895014181070277e-05,
      "loss": 0.6912,
      "step": 1190
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6696746349334717,
      "learning_rate": 1.989471958890471e-05,
      "loss": 0.6888,
      "step": 1191
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6752276420593262,
      "learning_rate": 1.989442458618966e-05,
      "loss": 0.6751,
      "step": 1192
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9331010580062866,
      "learning_rate": 1.9894129172937354e-05,
      "loss": 0.6874,
      "step": 1193
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6298333406448364,
      "learning_rate": 1.9893833349160055e-05,
      "loss": 0.6991,
      "step": 1194
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.676154375076294,
      "learning_rate": 1.9893537114870037e-05,
      "loss": 0.6766,
      "step": 1195
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7215545177459717,
      "learning_rate": 1.9893240470079593e-05,
      "loss": 0.6813,
      "step": 1196
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6445990800857544,
      "learning_rate": 1.9892943414801028e-05,
      "loss": 0.7148,
      "step": 1197
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8685849905014038,
      "learning_rate": 1.989264594904667e-05,
      "loss": 0.6882,
      "step": 1198
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8426058292388916,
      "learning_rate": 1.989234807282886e-05,
      "loss": 0.7058,
      "step": 1199
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6496106386184692,
      "learning_rate": 1.9892049786159955e-05,
      "loss": 0.7023,
      "step": 1200
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7984846830368042,
      "learning_rate": 1.9891751089052335e-05,
      "loss": 0.6654,
      "step": 1201
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.668593406677246,
      "learning_rate": 1.9891451981518395e-05,
      "loss": 0.6708,
      "step": 1202
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9407117366790771,
      "learning_rate": 1.989115246357054e-05,
      "loss": 0.6687,
      "step": 1203
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6798063516616821,
      "learning_rate": 1.9890852535221205e-05,
      "loss": 0.7112,
      "step": 1204
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7784608602523804,
      "learning_rate": 1.9890552196482825e-05,
      "loss": 0.6778,
      "step": 1205
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6652014255523682,
      "learning_rate": 1.989025144736787e-05,
      "loss": 0.7023,
      "step": 1206
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5983134508132935,
      "learning_rate": 1.9889950287888814e-05,
      "loss": 0.6631,
      "step": 1207
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7217609882354736,
      "learning_rate": 1.9889648718058156e-05,
      "loss": 0.713,
      "step": 1208
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6976872682571411,
      "learning_rate": 1.9889346737888404e-05,
      "loss": 0.6569,
      "step": 1209
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.531137228012085,
      "learning_rate": 1.9889044347392094e-05,
      "loss": 0.7033,
      "step": 1210
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.788779377937317,
      "learning_rate": 1.9888741546581766e-05,
      "loss": 0.6589,
      "step": 1211
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6894391775131226,
      "learning_rate": 1.988843833546999e-05,
      "loss": 0.7144,
      "step": 1212
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6136099100112915,
      "learning_rate": 1.988813471406934e-05,
      "loss": 0.6902,
      "step": 1213
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9311188459396362,
      "learning_rate": 1.9887830682392417e-05,
      "loss": 0.6707,
      "step": 1214
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.862844467163086,
      "learning_rate": 1.988752624045184e-05,
      "loss": 0.6759,
      "step": 1215
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.705223798751831,
      "learning_rate": 1.9887221388260234e-05,
      "loss": 0.6843,
      "step": 1216
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7269412279129028,
      "learning_rate": 1.9886916125830252e-05,
      "loss": 0.6879,
      "step": 1217
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0294559001922607,
      "learning_rate": 1.988661045317456e-05,
      "loss": 0.6856,
      "step": 1218
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7178713083267212,
      "learning_rate": 1.988630437030584e-05,
      "loss": 0.6773,
      "step": 1219
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8992741107940674,
      "learning_rate": 1.9885997877236788e-05,
      "loss": 0.6859,
      "step": 1220
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.8270227909088135,
      "learning_rate": 1.9885690973980127e-05,
      "loss": 1.072,
      "step": 1221
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0796172618865967,
      "learning_rate": 1.9885383660548586e-05,
      "loss": 0.6578,
      "step": 1222
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9107335805892944,
      "learning_rate": 1.9885075936954922e-05,
      "loss": 0.6918,
      "step": 1223
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.666792392730713,
      "learning_rate": 1.98847678032119e-05,
      "loss": 0.7063,
      "step": 1224
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9492583274841309,
      "learning_rate": 1.9884459259332303e-05,
      "loss": 0.6675,
      "step": 1225
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.872671127319336,
      "learning_rate": 1.9884150305328935e-05,
      "loss": 0.6601,
      "step": 1226
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.654727578163147,
      "learning_rate": 1.9883840941214612e-05,
      "loss": 0.6678,
      "step": 1227
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8166271448135376,
      "learning_rate": 1.9883531167002176e-05,
      "loss": 0.7734,
      "step": 1228
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.840002179145813,
      "learning_rate": 1.9883220982704475e-05,
      "loss": 0.6849,
      "step": 1229
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.710781455039978,
      "learning_rate": 1.988291038833438e-05,
      "loss": 0.6611,
      "step": 1230
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.854064702987671,
      "learning_rate": 1.988259938390478e-05,
      "loss": 0.6622,
      "step": 1231
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6251081228256226,
      "learning_rate": 1.988228796942858e-05,
      "loss": 0.6753,
      "step": 1232
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8816078901290894,
      "learning_rate": 1.98819761449187e-05,
      "loss": 0.6885,
      "step": 1233
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6808801889419556,
      "learning_rate": 1.9881663910388073e-05,
      "loss": 0.6757,
      "step": 1234
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7704333066940308,
      "learning_rate": 1.988135126584966e-05,
      "loss": 0.7646,
      "step": 1235
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0060746669769287,
      "learning_rate": 1.9881038211316433e-05,
      "loss": 0.7099,
      "step": 1236
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7479369640350342,
      "learning_rate": 1.988072474680138e-05,
      "loss": 0.6779,
      "step": 1237
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7784976959228516,
      "learning_rate": 1.9880410872317505e-05,
      "loss": 0.6618,
      "step": 1238
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0403666496276855,
      "learning_rate": 1.9880096587877835e-05,
      "loss": 0.68,
      "step": 1239
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8173388242721558,
      "learning_rate": 1.9879781893495405e-05,
      "loss": 0.7152,
      "step": 1240
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.007858991622925,
      "learning_rate": 1.987946678918328e-05,
      "loss": 0.737,
      "step": 1241
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.033848285675049,
      "learning_rate": 1.9879151274954532e-05,
      "loss": 0.6614,
      "step": 1242
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7132678031921387,
      "learning_rate": 1.9878835350822245e-05,
      "loss": 0.6654,
      "step": 1243
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.873489499092102,
      "learning_rate": 1.9878519016799535e-05,
      "loss": 0.6331,
      "step": 1244
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0437674522399902,
      "learning_rate": 1.9878202272899526e-05,
      "loss": 0.7245,
      "step": 1245
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6644331216812134,
      "learning_rate": 1.9877885119135354e-05,
      "loss": 1.057,
      "step": 1246
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1334333419799805,
      "learning_rate": 1.987756755552019e-05,
      "loss": 0.6833,
      "step": 1247
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2199980020523071,
      "learning_rate": 1.9877249582067198e-05,
      "loss": 0.9984,
      "step": 1248
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.839856505393982,
      "learning_rate": 1.987693119878958e-05,
      "loss": 0.6939,
      "step": 1249
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5147405862808228,
      "learning_rate": 1.9876612405700543e-05,
      "loss": 1.0141,
      "step": 1250
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8348065614700317,
      "learning_rate": 1.9876293202813314e-05,
      "loss": 0.7039,
      "step": 1251
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7440354824066162,
      "learning_rate": 1.9875973590141136e-05,
      "loss": 0.7021,
      "step": 1252
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6266220808029175,
      "learning_rate": 1.987565356769727e-05,
      "loss": 0.6782,
      "step": 1253
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6644283533096313,
      "learning_rate": 1.9875333135495e-05,
      "loss": 0.6822,
      "step": 1254
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7056148052215576,
      "learning_rate": 1.987501229354762e-05,
      "loss": 0.7354,
      "step": 1255
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7157524824142456,
      "learning_rate": 1.987469104186844e-05,
      "loss": 0.6383,
      "step": 1256
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5590277910232544,
      "learning_rate": 1.9874369380470787e-05,
      "loss": 0.667,
      "step": 1257
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.602418303489685,
      "learning_rate": 1.9874047309368008e-05,
      "loss": 0.6571,
      "step": 1258
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7716552019119263,
      "learning_rate": 1.9873724828573468e-05,
      "loss": 0.7131,
      "step": 1259
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6187098026275635,
      "learning_rate": 1.987340193810055e-05,
      "loss": 0.682,
      "step": 1260
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.743456482887268,
      "learning_rate": 1.9873078637962645e-05,
      "loss": 0.6778,
      "step": 1261
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6336501836776733,
      "learning_rate": 1.9872754928173172e-05,
      "loss": 0.7384,
      "step": 1262
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7874077558517456,
      "learning_rate": 1.9872430808745566e-05,
      "loss": 0.6802,
      "step": 1263
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0859220027923584,
      "learning_rate": 1.9872106279693263e-05,
      "loss": 1.0335,
      "step": 1264
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9231141805648804,
      "learning_rate": 1.987178134102974e-05,
      "loss": 0.6564,
      "step": 1265
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6995854377746582,
      "learning_rate": 1.9871455992768474e-05,
      "loss": 0.6673,
      "step": 1266
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6766226291656494,
      "learning_rate": 1.987113023492296e-05,
      "loss": 0.6812,
      "step": 1267
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.727195382118225,
      "learning_rate": 1.9870804067506727e-05,
      "loss": 0.651,
      "step": 1268
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7491729259490967,
      "learning_rate": 1.9870477490533296e-05,
      "loss": 0.6849,
      "step": 1269
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6870864629745483,
      "learning_rate": 1.9870150504016227e-05,
      "loss": 0.7005,
      "step": 1270
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6453654766082764,
      "learning_rate": 1.9869823107969078e-05,
      "loss": 0.6551,
      "step": 1271
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4753305912017822,
      "learning_rate": 1.9869495302405436e-05,
      "loss": 0.9978,
      "step": 1272
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7956271171569824,
      "learning_rate": 1.9869167087338908e-05,
      "loss": 0.6749,
      "step": 1273
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9415932893753052,
      "learning_rate": 1.9868838462783108e-05,
      "loss": 0.6539,
      "step": 1274
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6331417560577393,
      "learning_rate": 1.986850942875167e-05,
      "loss": 0.6826,
      "step": 1275
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8157217502593994,
      "learning_rate": 1.9868179985258245e-05,
      "loss": 0.6701,
      "step": 1276
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6707521677017212,
      "learning_rate": 1.9867850132316507e-05,
      "loss": 0.6774,
      "step": 1277
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6601240634918213,
      "learning_rate": 1.9867519869940138e-05,
      "loss": 0.6767,
      "step": 1278
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.717079758644104,
      "learning_rate": 1.986718919814284e-05,
      "loss": 0.6472,
      "step": 1279
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.690219521522522,
      "learning_rate": 1.9866858116938343e-05,
      "loss": 0.6631,
      "step": 1280
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.728824496269226,
      "learning_rate": 1.9866526626340374e-05,
      "loss": 0.6863,
      "step": 1281
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6970407962799072,
      "learning_rate": 1.9866194726362692e-05,
      "loss": 0.666,
      "step": 1282
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7883789539337158,
      "learning_rate": 1.9865862417019065e-05,
      "loss": 0.6379,
      "step": 1283
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5709294080734253,
      "learning_rate": 1.986552969832328e-05,
      "loss": 0.7014,
      "step": 1284
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8286926746368408,
      "learning_rate": 1.986519657028915e-05,
      "loss": 0.6726,
      "step": 1285
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5845330953598022,
      "learning_rate": 1.986486303293049e-05,
      "loss": 0.6847,
      "step": 1286
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7132502794265747,
      "learning_rate": 1.9864529086261136e-05,
      "loss": 0.7029,
      "step": 1287
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.813341498374939,
      "learning_rate": 1.9864194730294953e-05,
      "loss": 0.6813,
      "step": 1288
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5488287210464478,
      "learning_rate": 1.9863859965045812e-05,
      "loss": 0.6653,
      "step": 1289
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8220925331115723,
      "learning_rate": 1.9863524790527596e-05,
      "loss": 0.6533,
      "step": 1290
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.407740831375122,
      "learning_rate": 1.986318920675422e-05,
      "loss": 0.6656,
      "step": 1291
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5227657556533813,
      "learning_rate": 1.9862853213739603e-05,
      "loss": 0.6883,
      "step": 1292
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7690705060958862,
      "learning_rate": 1.986251681149769e-05,
      "loss": 0.6595,
      "step": 1293
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6066558361053467,
      "learning_rate": 1.9862180000042436e-05,
      "loss": 0.6244,
      "step": 1294
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5100969076156616,
      "learning_rate": 1.9861842779387814e-05,
      "loss": 0.691,
      "step": 1295
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7225457429885864,
      "learning_rate": 1.986150514954782e-05,
      "loss": 0.6871,
      "step": 1296
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6383625268936157,
      "learning_rate": 1.9861167110536463e-05,
      "loss": 0.6385,
      "step": 1297
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9078819751739502,
      "learning_rate": 1.9860828662367765e-05,
      "loss": 0.974,
      "step": 1298
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4957845211029053,
      "learning_rate": 1.986048980505577e-05,
      "loss": 0.6867,
      "step": 1299
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0772175788879395,
      "learning_rate": 1.9860150538614543e-05,
      "loss": 0.6996,
      "step": 1300
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6982231140136719,
      "learning_rate": 1.9859810863058153e-05,
      "loss": 0.682,
      "step": 1301
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8082648515701294,
      "learning_rate": 1.98594707784007e-05,
      "loss": 0.6689,
      "step": 1302
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7532119750976562,
      "learning_rate": 1.985913028465629e-05,
      "loss": 0.7075,
      "step": 1303
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6563501358032227,
      "learning_rate": 1.9858789381839055e-05,
      "loss": 0.6434,
      "step": 1304
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0031886100769043,
      "learning_rate": 1.9858448069963137e-05,
      "loss": 0.6388,
      "step": 1305
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.811314582824707,
      "learning_rate": 1.9858106349042696e-05,
      "loss": 0.7388,
      "step": 1306
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8476868867874146,
      "learning_rate": 1.9857764219091912e-05,
      "loss": 0.653,
      "step": 1307
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.600237250328064,
      "learning_rate": 1.9857421680124984e-05,
      "loss": 0.7025,
      "step": 1308
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7846038341522217,
      "learning_rate": 1.985707873215612e-05,
      "loss": 0.6889,
      "step": 1309
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7380037307739258,
      "learning_rate": 1.985673537519955e-05,
      "loss": 0.6794,
      "step": 1310
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6224476099014282,
      "learning_rate": 1.9856391609269522e-05,
      "loss": 0.6875,
      "step": 1311
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8437808752059937,
      "learning_rate": 1.9856047434380304e-05,
      "loss": 0.6643,
      "step": 1312
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6556832790374756,
      "learning_rate": 1.9855702850546165e-05,
      "loss": 0.6868,
      "step": 1313
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6481789350509644,
      "learning_rate": 1.9855357857781415e-05,
      "loss": 0.648,
      "step": 1314
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7046302556991577,
      "learning_rate": 1.985501245610036e-05,
      "loss": 0.6747,
      "step": 1315
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.815792202949524,
      "learning_rate": 1.985466664551733e-05,
      "loss": 0.6634,
      "step": 1316
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6627494096755981,
      "learning_rate": 1.985432042604668e-05,
      "loss": 0.7251,
      "step": 1317
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7401753664016724,
      "learning_rate": 1.9853973797702774e-05,
      "loss": 0.7071,
      "step": 1318
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7472789287567139,
      "learning_rate": 1.985362676049999e-05,
      "loss": 0.657,
      "step": 1319
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6508598327636719,
      "learning_rate": 1.9853279314452733e-05,
      "loss": 0.7012,
      "step": 1320
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.728501319885254,
      "learning_rate": 1.9852931459575414e-05,
      "loss": 0.6386,
      "step": 1321
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7703889608383179,
      "learning_rate": 1.9852583195882468e-05,
      "loss": 0.6764,
      "step": 1322
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6811625957489014,
      "learning_rate": 1.9852234523388342e-05,
      "loss": 0.651,
      "step": 1323
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7110469341278076,
      "learning_rate": 1.985188544210751e-05,
      "loss": 0.6669,
      "step": 1324
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6631176471710205,
      "learning_rate": 1.9851535952054452e-05,
      "loss": 0.6635,
      "step": 1325
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8085379600524902,
      "learning_rate": 1.9851186053243667e-05,
      "loss": 0.6409,
      "step": 1326
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6380712985992432,
      "learning_rate": 1.9850835745689676e-05,
      "loss": 0.6555,
      "step": 1327
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.749403476715088,
      "learning_rate": 1.985048502940701e-05,
      "loss": 0.6638,
      "step": 1328
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6188281774520874,
      "learning_rate": 1.9850133904410226e-05,
      "loss": 0.6704,
      "step": 1329
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6485813856124878,
      "learning_rate": 1.984978237071389e-05,
      "loss": 0.6466,
      "step": 1330
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5871226787567139,
      "learning_rate": 1.9849430428332593e-05,
      "loss": 0.676,
      "step": 1331
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6671695709228516,
      "learning_rate": 1.9849078077280927e-05,
      "loss": 0.7307,
      "step": 1332
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6420643329620361,
      "learning_rate": 1.984872531757352e-05,
      "loss": 0.6691,
      "step": 1333
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6601427793502808,
      "learning_rate": 1.9848372149225004e-05,
      "loss": 0.6416,
      "step": 1334
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7026411294937134,
      "learning_rate": 1.984801857225004e-05,
      "loss": 0.6715,
      "step": 1335
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7206051349639893,
      "learning_rate": 1.984766458666329e-05,
      "loss": 0.6522,
      "step": 1336
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8699655532836914,
      "learning_rate": 1.9847310192479445e-05,
      "loss": 0.7176,
      "step": 1337
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6067838668823242,
      "learning_rate": 1.984695538971321e-05,
      "loss": 0.7502,
      "step": 1338
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6482692956924438,
      "learning_rate": 1.9846600178379305e-05,
      "loss": 0.7156,
      "step": 1339
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6981178522109985,
      "learning_rate": 1.9846244558492468e-05,
      "loss": 0.6727,
      "step": 1340
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6191799640655518,
      "learning_rate": 1.9845888530067458e-05,
      "loss": 0.6596,
      "step": 1341
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6409897804260254,
      "learning_rate": 1.9845532093119044e-05,
      "loss": 0.6758,
      "step": 1342
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7351735830307007,
      "learning_rate": 1.9845175247662015e-05,
      "loss": 0.6686,
      "step": 1343
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6277070045471191,
      "learning_rate": 1.9844817993711176e-05,
      "loss": 0.6694,
      "step": 1344
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7011929750442505,
      "learning_rate": 1.9844460331281355e-05,
      "loss": 0.6713,
      "step": 1345
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.713097333908081,
      "learning_rate": 1.984410226038739e-05,
      "loss": 0.6333,
      "step": 1346
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6461799144744873,
      "learning_rate": 1.9843743781044133e-05,
      "loss": 0.6596,
      "step": 1347
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6632845401763916,
      "learning_rate": 1.9843384893266465e-05,
      "loss": 0.6646,
      "step": 1348
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7591474056243896,
      "learning_rate": 1.9843025597069274e-05,
      "loss": 0.6725,
      "step": 1349
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.038256883621216,
      "learning_rate": 1.9842665892467466e-05,
      "loss": 1.0274,
      "step": 1350
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0001511573791504,
      "learning_rate": 1.984230577947597e-05,
      "loss": 0.731,
      "step": 1351
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9722126722335815,
      "learning_rate": 1.9841945258109723e-05,
      "loss": 0.6803,
      "step": 1352
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7442388534545898,
      "learning_rate": 1.984158432838369e-05,
      "loss": 0.6984,
      "step": 1353
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.817391037940979,
      "learning_rate": 1.9841222990312835e-05,
      "loss": 0.6823,
      "step": 1354
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2978460788726807,
      "learning_rate": 1.9840861243912165e-05,
      "loss": 1.0701,
      "step": 1355
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.034287214279175,
      "learning_rate": 1.984049908919668e-05,
      "loss": 0.7197,
      "step": 1356
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6672170162200928,
      "learning_rate": 1.9840136526181412e-05,
      "loss": 0.6726,
      "step": 1357
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.650750756263733,
      "learning_rate": 1.98397735548814e-05,
      "loss": 0.6755,
      "step": 1358
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3848289251327515,
      "learning_rate": 1.9839410175311703e-05,
      "loss": 0.9861,
      "step": 1359
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0126655101776123,
      "learning_rate": 1.9839046387487402e-05,
      "loss": 0.7441,
      "step": 1360
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7545353174209595,
      "learning_rate": 1.9838682191423592e-05,
      "loss": 0.6621,
      "step": 1361
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7046868801116943,
      "learning_rate": 1.983831758713538e-05,
      "loss": 0.6624,
      "step": 1362
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7811291217803955,
      "learning_rate": 1.9837952574637896e-05,
      "loss": 0.73,
      "step": 1363
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8360605239868164,
      "learning_rate": 1.983758715394629e-05,
      "loss": 0.6895,
      "step": 1364
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.669838547706604,
      "learning_rate": 1.983722132507572e-05,
      "loss": 0.6298,
      "step": 1365
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8046419620513916,
      "learning_rate": 1.9836855088041357e-05,
      "loss": 0.6803,
      "step": 1366
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7477329969406128,
      "learning_rate": 1.983648844285841e-05,
      "loss": 0.7221,
      "step": 1367
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8537929058074951,
      "learning_rate": 1.9836121389542086e-05,
      "loss": 0.6535,
      "step": 1368
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6442363262176514,
      "learning_rate": 1.9835753928107613e-05,
      "loss": 0.7022,
      "step": 1369
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6209567785263062,
      "learning_rate": 1.983538605857024e-05,
      "loss": 0.6794,
      "step": 1370
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.68607759475708,
      "learning_rate": 1.9835017780945226e-05,
      "loss": 0.6692,
      "step": 1371
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.74774169921875,
      "learning_rate": 1.983464909524786e-05,
      "loss": 0.6772,
      "step": 1372
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7550628185272217,
      "learning_rate": 1.9834280001493434e-05,
      "loss": 0.6918,
      "step": 1373
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7444489002227783,
      "learning_rate": 1.9833910499697262e-05,
      "loss": 0.6452,
      "step": 1374
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7787927389144897,
      "learning_rate": 1.9833540589874678e-05,
      "loss": 0.6722,
      "step": 1375
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6237311363220215,
      "learning_rate": 1.9833170272041026e-05,
      "loss": 0.6526,
      "step": 1376
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8754725456237793,
      "learning_rate": 1.9832799546211678e-05,
      "loss": 0.6567,
      "step": 1377
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6300108432769775,
      "learning_rate": 1.9832428412402005e-05,
      "loss": 0.6931,
      "step": 1378
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5660367012023926,
      "learning_rate": 1.983205687062742e-05,
      "loss": 0.6881,
      "step": 1379
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6856118440628052,
      "learning_rate": 1.9831684920903326e-05,
      "loss": 0.6768,
      "step": 1380
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5064581632614136,
      "learning_rate": 1.9831312563245167e-05,
      "loss": 0.6204,
      "step": 1381
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.732558012008667,
      "learning_rate": 1.9830939797668382e-05,
      "loss": 0.6907,
      "step": 1382
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7319738864898682,
      "learning_rate": 1.9830566624188445e-05,
      "loss": 0.6616,
      "step": 1383
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6846400499343872,
      "learning_rate": 1.983019304282084e-05,
      "loss": 0.6543,
      "step": 1384
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7393333911895752,
      "learning_rate": 1.982981905358106e-05,
      "loss": 0.6603,
      "step": 1385
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6206746101379395,
      "learning_rate": 1.9829444656484633e-05,
      "loss": 0.671,
      "step": 1386
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5704858303070068,
      "learning_rate": 1.9829069851547085e-05,
      "loss": 0.7013,
      "step": 1387
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.645172357559204,
      "learning_rate": 1.9828694638783974e-05,
      "loss": 0.6803,
      "step": 1388
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7173068523406982,
      "learning_rate": 1.9828319018210863e-05,
      "loss": 0.6344,
      "step": 1389
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.628185510635376,
      "learning_rate": 1.9827942989843338e-05,
      "loss": 0.6571,
      "step": 1390
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.738417148590088,
      "learning_rate": 1.9827566553697e-05,
      "loss": 0.6724,
      "step": 1391
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6314656734466553,
      "learning_rate": 1.9827189709787476e-05,
      "loss": 0.6892,
      "step": 1392
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.762066125869751,
      "learning_rate": 1.982681245813039e-05,
      "loss": 0.7126,
      "step": 1393
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6333775520324707,
      "learning_rate": 1.9826434798741404e-05,
      "loss": 0.6294,
      "step": 1394
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.67708158493042,
      "learning_rate": 1.9826056731636183e-05,
      "loss": 0.6244,
      "step": 1395
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7091723680496216,
      "learning_rate": 1.9825678256830417e-05,
      "loss": 0.7239,
      "step": 1396
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5081104040145874,
      "learning_rate": 1.9825299374339803e-05,
      "loss": 0.7345,
      "step": 1397
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7426434755325317,
      "learning_rate": 1.982492008418007e-05,
      "loss": 0.6647,
      "step": 1398
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7115391492843628,
      "learning_rate": 1.9824540386366953e-05,
      "loss": 0.6707,
      "step": 1399
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7732995748519897,
      "learning_rate": 1.9824160280916203e-05,
      "loss": 0.6595,
      "step": 1400
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6102439165115356,
      "learning_rate": 1.9823779767843594e-05,
      "loss": 0.6629,
      "step": 1401
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6064151525497437,
      "learning_rate": 1.9823398847164913e-05,
      "loss": 0.5898,
      "step": 1402
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5325616598129272,
      "learning_rate": 1.9823017518895968e-05,
      "loss": 0.6541,
      "step": 1403
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.64641273021698,
      "learning_rate": 1.9822635783052572e-05,
      "loss": 0.6484,
      "step": 1404
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7091612815856934,
      "learning_rate": 1.9822253639650574e-05,
      "loss": 0.6764,
      "step": 1405
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5012768507003784,
      "learning_rate": 1.982187108870583e-05,
      "loss": 0.6903,
      "step": 1406
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.548126220703125,
      "learning_rate": 1.9821488130234203e-05,
      "loss": 0.6769,
      "step": 1407
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6412440538406372,
      "learning_rate": 1.9821104764251594e-05,
      "loss": 0.6342,
      "step": 1408
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7538541555404663,
      "learning_rate": 1.98207209907739e-05,
      "loss": 0.64,
      "step": 1409
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6651753187179565,
      "learning_rate": 1.982033680981705e-05,
      "loss": 0.6666,
      "step": 1410
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6909059286117554,
      "learning_rate": 1.9819952221396983e-05,
      "loss": 0.6782,
      "step": 1411
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6549454927444458,
      "learning_rate": 1.9819567225529653e-05,
      "loss": 0.6448,
      "step": 1412
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.533031702041626,
      "learning_rate": 1.9819181822231043e-05,
      "loss": 0.6893,
      "step": 1413
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8645522594451904,
      "learning_rate": 1.9818796011517134e-05,
      "loss": 1.0471,
      "step": 1414
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9028167724609375,
      "learning_rate": 1.9818409793403938e-05,
      "loss": 0.6797,
      "step": 1415
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8813507556915283,
      "learning_rate": 1.9818023167907485e-05,
      "loss": 0.6773,
      "step": 1416
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.60419499874115,
      "learning_rate": 1.9817636135043808e-05,
      "loss": 0.664,
      "step": 1417
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6892749071121216,
      "learning_rate": 1.9817248694828972e-05,
      "loss": 0.6873,
      "step": 1418
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7793371677398682,
      "learning_rate": 1.9816860847279048e-05,
      "loss": 0.6805,
      "step": 1419
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.619826316833496,
      "learning_rate": 1.9816472592410133e-05,
      "loss": 0.6693,
      "step": 1420
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.590438723564148,
      "learning_rate": 1.9816083930238334e-05,
      "loss": 0.6687,
      "step": 1421
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6209373474121094,
      "learning_rate": 1.9815694860779775e-05,
      "loss": 0.6458,
      "step": 1422
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5479549169540405,
      "learning_rate": 1.98153053840506e-05,
      "loss": 0.6782,
      "step": 1423
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6195387840270996,
      "learning_rate": 1.9814915500066975e-05,
      "loss": 0.6263,
      "step": 1424
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.521360158920288,
      "learning_rate": 1.9814525208845073e-05,
      "loss": 0.6587,
      "step": 1425
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5814142227172852,
      "learning_rate": 1.9814134510401085e-05,
      "loss": 0.6785,
      "step": 1426
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.633685827255249,
      "learning_rate": 1.9813743404751222e-05,
      "loss": 0.6392,
      "step": 1427
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6569552421569824,
      "learning_rate": 1.981335189191172e-05,
      "loss": 0.7053,
      "step": 1428
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5710116624832153,
      "learning_rate": 1.9812959971898813e-05,
      "loss": 0.6525,
      "step": 1429
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5576682090759277,
      "learning_rate": 1.9812567644728767e-05,
      "loss": 0.6599,
      "step": 1430
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6264026165008545,
      "learning_rate": 1.981217491041786e-05,
      "loss": 0.7167,
      "step": 1431
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6671613454818726,
      "learning_rate": 1.9811781768982392e-05,
      "loss": 0.6877,
      "step": 1432
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5300873517990112,
      "learning_rate": 1.981138822043867e-05,
      "loss": 0.6774,
      "step": 1433
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5439194440841675,
      "learning_rate": 1.981099426480302e-05,
      "loss": 0.6519,
      "step": 1434
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5583678483963013,
      "learning_rate": 1.9810599902091793e-05,
      "loss": 0.7017,
      "step": 1435
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5886632204055786,
      "learning_rate": 1.981020513232135e-05,
      "loss": 0.6742,
      "step": 1436
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.703321099281311,
      "learning_rate": 1.9809809955508075e-05,
      "loss": 0.6477,
      "step": 1437
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5666453838348389,
      "learning_rate": 1.9809414371668356e-05,
      "loss": 0.6724,
      "step": 1438
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6916533708572388,
      "learning_rate": 1.9809018380818612e-05,
      "loss": 0.6601,
      "step": 1439
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6417124271392822,
      "learning_rate": 1.9808621982975277e-05,
      "loss": 0.672,
      "step": 1440
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5757957696914673,
      "learning_rate": 1.980822517815479e-05,
      "loss": 0.6589,
      "step": 1441
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5994127988815308,
      "learning_rate": 1.980782796637362e-05,
      "loss": 0.6503,
      "step": 1442
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5372828245162964,
      "learning_rate": 1.9807430347648246e-05,
      "loss": 0.6452,
      "step": 1443
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5850380659103394,
      "learning_rate": 1.9807032321995166e-05,
      "loss": 0.6589,
      "step": 1444
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.65659761428833,
      "learning_rate": 1.9806633889430898e-05,
      "loss": 0.6167,
      "step": 1445
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8234935998916626,
      "learning_rate": 1.9806235049971972e-05,
      "loss": 0.6537,
      "step": 1446
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1936118602752686,
      "learning_rate": 1.980583580363493e-05,
      "loss": 1.0568,
      "step": 1447
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.053528070449829,
      "learning_rate": 1.9805436150436352e-05,
      "loss": 0.6573,
      "step": 1448
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8209097385406494,
      "learning_rate": 1.9805036090392808e-05,
      "loss": 0.6732,
      "step": 1449
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7125083208084106,
      "learning_rate": 1.9804635623520898e-05,
      "loss": 0.6373,
      "step": 1450
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1921886205673218,
      "learning_rate": 1.9804234749837247e-05,
      "loss": 0.9573,
      "step": 1451
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0633559226989746,
      "learning_rate": 1.9803833469358475e-05,
      "loss": 0.6666,
      "step": 1452
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6569914817810059,
      "learning_rate": 1.9803431782101244e-05,
      "loss": 0.6897,
      "step": 1453
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6644225120544434,
      "learning_rate": 1.9803029688082213e-05,
      "loss": 0.6618,
      "step": 1454
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8147666454315186,
      "learning_rate": 1.980262718731807e-05,
      "loss": 0.6687,
      "step": 1455
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7971131801605225,
      "learning_rate": 1.9802224279825514e-05,
      "loss": 0.6898,
      "step": 1456
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.670626163482666,
      "learning_rate": 1.980182096562126e-05,
      "loss": 0.6229,
      "step": 1457
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.846803069114685,
      "learning_rate": 1.980141724472204e-05,
      "loss": 0.6364,
      "step": 1458
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6899917125701904,
      "learning_rate": 1.9801013117144617e-05,
      "loss": 0.6758,
      "step": 1459
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6901779174804688,
      "learning_rate": 1.9800608582905747e-05,
      "loss": 0.6403,
      "step": 1460
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6796395778656006,
      "learning_rate": 1.980020364202222e-05,
      "loss": 0.6166,
      "step": 1461
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5835729837417603,
      "learning_rate": 1.9799798294510836e-05,
      "loss": 0.6699,
      "step": 1462
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7270216941833496,
      "learning_rate": 1.9799392540388416e-05,
      "loss": 0.6811,
      "step": 1463
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5993850231170654,
      "learning_rate": 1.979898637967179e-05,
      "loss": 0.672,
      "step": 1464
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6011788845062256,
      "learning_rate": 1.9798579812377817e-05,
      "loss": 0.6098,
      "step": 1465
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.569272994995117,
      "learning_rate": 1.9798172838523366e-05,
      "loss": 1.0089,
      "step": 1466
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.033214807510376,
      "learning_rate": 1.979776545812532e-05,
      "loss": 0.6832,
      "step": 1467
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.744287133216858,
      "learning_rate": 1.979735767120058e-05,
      "loss": 0.6723,
      "step": 1468
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6718004941940308,
      "learning_rate": 1.9796949477766065e-05,
      "loss": 0.7065,
      "step": 1469
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.780448079109192,
      "learning_rate": 1.9796540877838718e-05,
      "loss": 0.6957,
      "step": 1470
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.620569109916687,
      "learning_rate": 1.9796131871435492e-05,
      "loss": 0.6323,
      "step": 1471
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7284064292907715,
      "learning_rate": 1.9795722458573352e-05,
      "loss": 0.6418,
      "step": 1472
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7479524612426758,
      "learning_rate": 1.9795312639269287e-05,
      "loss": 0.6787,
      "step": 1473
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6849160194396973,
      "learning_rate": 1.9794902413540304e-05,
      "loss": 0.6393,
      "step": 1474
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4266606569290161,
      "learning_rate": 1.9794491781403425e-05,
      "loss": 1.0159,
      "step": 1475
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9860154390335083,
      "learning_rate": 1.979408074287568e-05,
      "loss": 0.6888,
      "step": 1476
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8498986959457397,
      "learning_rate": 1.979366929797413e-05,
      "loss": 0.7001,
      "step": 1477
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6818820238113403,
      "learning_rate": 1.9793257446715847e-05,
      "loss": 0.6538,
      "step": 1478
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7024776935577393,
      "learning_rate": 1.9792845189117915e-05,
      "loss": 0.6741,
      "step": 1479
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6243112087249756,
      "learning_rate": 1.9792432525197444e-05,
      "loss": 0.6231,
      "step": 1480
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.595219373703003,
      "learning_rate": 1.9792019454971555e-05,
      "loss": 0.6596,
      "step": 1481
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5053290128707886,
      "learning_rate": 1.9791605978457382e-05,
      "loss": 1.0014,
      "step": 1482
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9561986923217773,
      "learning_rate": 1.979119209567209e-05,
      "loss": 0.6461,
      "step": 1483
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6934150457382202,
      "learning_rate": 1.9790777806632848e-05,
      "loss": 0.665,
      "step": 1484
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5342904329299927,
      "learning_rate": 1.9790363111356838e-05,
      "loss": 0.6851,
      "step": 1485
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6573164463043213,
      "learning_rate": 1.9789948009861276e-05,
      "loss": 0.6539,
      "step": 1486
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6610969305038452,
      "learning_rate": 1.9789532502163383e-05,
      "loss": 0.6884,
      "step": 1487
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5179758071899414,
      "learning_rate": 1.9789116588280398e-05,
      "loss": 0.7027,
      "step": 1488
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4832727909088135,
      "learning_rate": 1.9788700268229577e-05,
      "loss": 0.6654,
      "step": 1489
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6127502918243408,
      "learning_rate": 1.9788283542028196e-05,
      "loss": 0.7222,
      "step": 1490
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.698611855506897,
      "learning_rate": 1.9787866409693544e-05,
      "loss": 0.6532,
      "step": 1491
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.641985535621643,
      "learning_rate": 1.9787448871242932e-05,
      "loss": 0.6618,
      "step": 1492
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6168243885040283,
      "learning_rate": 1.9787030926693683e-05,
      "loss": 0.6153,
      "step": 1493
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7281417846679688,
      "learning_rate": 1.978661257606313e-05,
      "loss": 0.6664,
      "step": 1494
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6787383556365967,
      "learning_rate": 1.9786193819368646e-05,
      "loss": 0.6585,
      "step": 1495
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7179930210113525,
      "learning_rate": 1.9785774656627593e-05,
      "loss": 0.665,
      "step": 1496
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8219584226608276,
      "learning_rate": 1.9785355087857373e-05,
      "loss": 0.6821,
      "step": 1497
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6316653490066528,
      "learning_rate": 1.9784935113075385e-05,
      "loss": 0.6831,
      "step": 1498
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.874664306640625,
      "learning_rate": 1.978451473229906e-05,
      "loss": 0.6144,
      "step": 1499
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8060673475265503,
      "learning_rate": 1.978409394554584e-05,
      "loss": 0.6357,
      "step": 1500
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.585785984992981,
      "learning_rate": 1.9783672752833185e-05,
      "loss": 0.6538,
      "step": 1501
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6434869766235352,
      "learning_rate": 1.978325115417857e-05,
      "loss": 0.6279,
      "step": 1502
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7838428020477295,
      "learning_rate": 1.9782829149599485e-05,
      "loss": 0.67,
      "step": 1503
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6268937587738037,
      "learning_rate": 1.9782406739113444e-05,
      "loss": 0.5986,
      "step": 1504
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6554566621780396,
      "learning_rate": 1.9781983922737972e-05,
      "loss": 0.6459,
      "step": 1505
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7380794286727905,
      "learning_rate": 1.978156070049061e-05,
      "loss": 0.645,
      "step": 1506
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8210142850875854,
      "learning_rate": 1.9781137072388924e-05,
      "loss": 0.9955,
      "step": 1507
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1479177474975586,
      "learning_rate": 1.978071303845049e-05,
      "loss": 0.6227,
      "step": 1508
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7862834930419922,
      "learning_rate": 1.978028859869289e-05,
      "loss": 0.6537,
      "step": 1509
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.543481707572937,
      "learning_rate": 1.9779863753133754e-05,
      "loss": 0.6809,
      "step": 1510
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8099427223205566,
      "learning_rate": 1.9779438501790693e-05,
      "loss": 0.6634,
      "step": 1511
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.725780725479126,
      "learning_rate": 1.977901284468136e-05,
      "loss": 0.6284,
      "step": 1512
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.414629340171814,
      "learning_rate": 1.9778586781823417e-05,
      "loss": 0.9275,
      "step": 1513
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.114278793334961,
      "learning_rate": 1.977816031323454e-05,
      "loss": 0.716,
      "step": 1514
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9020662307739258,
      "learning_rate": 1.977773343893242e-05,
      "loss": 0.6889,
      "step": 1515
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5840262174606323,
      "learning_rate": 1.9777306158934776e-05,
      "loss": 0.6557,
      "step": 1516
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.74437415599823,
      "learning_rate": 1.977687847325933e-05,
      "loss": 0.6698,
      "step": 1517
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.820623517036438,
      "learning_rate": 1.977645038192383e-05,
      "loss": 0.6502,
      "step": 1518
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.182354211807251,
      "learning_rate": 1.9776021884946043e-05,
      "loss": 0.9402,
      "step": 1519
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.652285575866699,
      "learning_rate": 1.9775592982343742e-05,
      "loss": 0.6333,
      "step": 1520
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8976705074310303,
      "learning_rate": 1.9775163674134724e-05,
      "loss": 0.6719,
      "step": 1521
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6418026685714722,
      "learning_rate": 1.97747339603368e-05,
      "loss": 0.6542,
      "step": 1522
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.721541166305542,
      "learning_rate": 1.9774303840967808e-05,
      "loss": 0.6662,
      "step": 1523
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.608540415763855,
      "learning_rate": 1.9773873316045584e-05,
      "loss": 0.6615,
      "step": 1524
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7159230709075928,
      "learning_rate": 1.9773442385587996e-05,
      "loss": 0.7071,
      "step": 1525
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7181227207183838,
      "learning_rate": 1.9773011049612924e-05,
      "loss": 0.6638,
      "step": 1526
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6926676034927368,
      "learning_rate": 1.9772579308138267e-05,
      "loss": 0.6535,
      "step": 1527
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6025508642196655,
      "learning_rate": 1.9772147161181935e-05,
      "loss": 0.6511,
      "step": 1528
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6070988178253174,
      "learning_rate": 1.977171460876186e-05,
      "loss": 0.6899,
      "step": 1529
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6766750812530518,
      "learning_rate": 1.977128165089599e-05,
      "loss": 0.6887,
      "step": 1530
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6955245733261108,
      "learning_rate": 1.9770848287602287e-05,
      "loss": 0.674,
      "step": 1531
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7100074291229248,
      "learning_rate": 1.9770414518898734e-05,
      "loss": 0.6471,
      "step": 1532
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7030901908874512,
      "learning_rate": 1.976998034480333e-05,
      "loss": 0.6489,
      "step": 1533
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5804429054260254,
      "learning_rate": 1.976954576533409e-05,
      "loss": 0.6677,
      "step": 1534
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6554450988769531,
      "learning_rate": 1.976911078050904e-05,
      "loss": 0.6667,
      "step": 1535
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.614225149154663,
      "learning_rate": 1.976867539034623e-05,
      "loss": 0.6122,
      "step": 1536
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.726146936416626,
      "learning_rate": 1.976823959486373e-05,
      "loss": 0.6023,
      "step": 1537
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7007582187652588,
      "learning_rate": 1.9767803394079618e-05,
      "loss": 0.6258,
      "step": 1538
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.669629454612732,
      "learning_rate": 1.9767366788011993e-05,
      "loss": 0.6698,
      "step": 1539
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.549206256866455,
      "learning_rate": 1.9766929776678975e-05,
      "loss": 0.6274,
      "step": 1540
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7309976816177368,
      "learning_rate": 1.976649236009869e-05,
      "loss": 0.6318,
      "step": 1541
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7907415628433228,
      "learning_rate": 1.9766054538289287e-05,
      "loss": 0.9893,
      "step": 1542
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.519210934638977,
      "learning_rate": 1.9765616311268938e-05,
      "loss": 0.9907,
      "step": 1543
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2560312747955322,
      "learning_rate": 1.976517767905582e-05,
      "loss": 0.6431,
      "step": 1544
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9621665477752686,
      "learning_rate": 1.9764738641668137e-05,
      "loss": 0.7042,
      "step": 1545
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8651726245880127,
      "learning_rate": 1.9764299199124104e-05,
      "loss": 0.6768,
      "step": 1546
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7561309337615967,
      "learning_rate": 1.9763859351441952e-05,
      "loss": 0.6739,
      "step": 1547
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8029088973999023,
      "learning_rate": 1.9763419098639934e-05,
      "loss": 0.693,
      "step": 1548
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9105240106582642,
      "learning_rate": 1.9762978440736318e-05,
      "loss": 0.6867,
      "step": 1549
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.835360050201416,
      "learning_rate": 1.9762537377749383e-05,
      "loss": 0.6605,
      "step": 1550
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9947419166564941,
      "learning_rate": 1.9762095909697432e-05,
      "loss": 0.6467,
      "step": 1551
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7650270462036133,
      "learning_rate": 1.9761654036598784e-05,
      "loss": 0.6359,
      "step": 1552
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6854472160339355,
      "learning_rate": 1.976121175847177e-05,
      "loss": 0.6796,
      "step": 1553
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.786903977394104,
      "learning_rate": 1.9760769075334742e-05,
      "loss": 0.6528,
      "step": 1554
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.761546015739441,
      "learning_rate": 1.9760325987206064e-05,
      "loss": 0.6592,
      "step": 1555
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.835752010345459,
      "learning_rate": 1.975988249410413e-05,
      "loss": 0.6579,
      "step": 1556
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7432100772857666,
      "learning_rate": 1.9759438596047337e-05,
      "loss": 0.6525,
      "step": 1557
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6313180923461914,
      "learning_rate": 1.97589942930541e-05,
      "loss": 0.6505,
      "step": 1558
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7356852293014526,
      "learning_rate": 1.9758549585142853e-05,
      "loss": 0.645,
      "step": 1559
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7601234912872314,
      "learning_rate": 1.9758104472332057e-05,
      "loss": 0.6394,
      "step": 1560
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6468322277069092,
      "learning_rate": 1.975765895464017e-05,
      "loss": 0.6702,
      "step": 1561
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7219297885894775,
      "learning_rate": 1.975721303208568e-05,
      "loss": 0.6936,
      "step": 1562
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6160279512405396,
      "learning_rate": 1.9756766704687093e-05,
      "loss": 0.6681,
      "step": 1563
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.667873740196228,
      "learning_rate": 1.9756319972462922e-05,
      "loss": 0.6319,
      "step": 1564
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.599122166633606,
      "learning_rate": 1.975587283543171e-05,
      "loss": 0.6379,
      "step": 1565
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6422098875045776,
      "learning_rate": 1.9755425293612007e-05,
      "loss": 0.6494,
      "step": 1566
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.672215223312378,
      "learning_rate": 1.975497734702238e-05,
      "loss": 0.6651,
      "step": 1567
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.541667103767395,
      "learning_rate": 1.9754528995681414e-05,
      "loss": 0.6778,
      "step": 1568
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6662954092025757,
      "learning_rate": 1.975408023960771e-05,
      "loss": 0.6462,
      "step": 1569
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5911654233932495,
      "learning_rate": 1.97536310788199e-05,
      "loss": 0.6554,
      "step": 1570
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1780052185058594,
      "learning_rate": 1.9753181513336607e-05,
      "loss": 0.6477,
      "step": 1571
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7928295135498047,
      "learning_rate": 1.9752731543176492e-05,
      "loss": 0.6436,
      "step": 1572
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6071467399597168,
      "learning_rate": 1.9752281168358222e-05,
      "loss": 0.648,
      "step": 1573
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.739298939704895,
      "learning_rate": 1.975183038890048e-05,
      "loss": 0.6718,
      "step": 1574
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6801389455795288,
      "learning_rate": 1.975137920482198e-05,
      "loss": 0.6521,
      "step": 1575
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6923803091049194,
      "learning_rate": 1.975092761614143e-05,
      "loss": 0.6651,
      "step": 1576
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.887859582901001,
      "learning_rate": 1.975047562287758e-05,
      "loss": 0.6262,
      "step": 1577
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7345370054244995,
      "learning_rate": 1.9750023225049175e-05,
      "loss": 0.6762,
      "step": 1578
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.694947361946106,
      "learning_rate": 1.974957042267499e-05,
      "loss": 0.6884,
      "step": 1579
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7780413627624512,
      "learning_rate": 1.9749117215773807e-05,
      "loss": 0.6824,
      "step": 1580
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.664587378501892,
      "learning_rate": 1.9748663604364437e-05,
      "loss": 0.6599,
      "step": 1581
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7408926486968994,
      "learning_rate": 1.9748209588465698e-05,
      "loss": 0.6346,
      "step": 1582
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8129066228866577,
      "learning_rate": 1.974775516809643e-05,
      "loss": 0.6376,
      "step": 1583
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6305886507034302,
      "learning_rate": 1.9747300343275484e-05,
      "loss": 0.6392,
      "step": 1584
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8905110359191895,
      "learning_rate": 1.9746845114021736e-05,
      "loss": 0.66,
      "step": 1585
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.648985743522644,
      "learning_rate": 1.974638948035407e-05,
      "loss": 0.6631,
      "step": 1586
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.592076301574707,
      "learning_rate": 1.9745933442291393e-05,
      "loss": 0.6687,
      "step": 1587
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7427781820297241,
      "learning_rate": 1.9745476999852626e-05,
      "loss": 0.6932,
      "step": 1588
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5623663663864136,
      "learning_rate": 1.974502015305671e-05,
      "loss": 0.6462,
      "step": 1589
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5598841905593872,
      "learning_rate": 1.9744562901922597e-05,
      "loss": 0.626,
      "step": 1590
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5936390161514282,
      "learning_rate": 1.9744105246469264e-05,
      "loss": 0.6655,
      "step": 1591
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5728251934051514,
      "learning_rate": 1.9743647186715695e-05,
      "loss": 0.7004,
      "step": 1592
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.582640290260315,
      "learning_rate": 1.9743188722680896e-05,
      "loss": 0.6378,
      "step": 1593
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6411596536636353,
      "learning_rate": 1.974272985438389e-05,
      "loss": 0.7394,
      "step": 1594
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4921879768371582,
      "learning_rate": 1.974227058184372e-05,
      "loss": 0.6382,
      "step": 1595
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.564147710800171,
      "learning_rate": 1.9741810905079443e-05,
      "loss": 0.6609,
      "step": 1596
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.668597936630249,
      "learning_rate": 1.9741350824110122e-05,
      "loss": 0.6358,
      "step": 1597
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.260915756225586,
      "learning_rate": 1.9740890338954856e-05,
      "loss": 1.1078,
      "step": 1598
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8936446905136108,
      "learning_rate": 1.9740429449632747e-05,
      "loss": 0.6552,
      "step": 1599
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7482792139053345,
      "learning_rate": 1.9739968156162917e-05,
      "loss": 0.6741,
      "step": 1600
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6678540706634521,
      "learning_rate": 1.973950645856451e-05,
      "loss": 0.6482,
      "step": 1601
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6439703702926636,
      "learning_rate": 1.9739044356856684e-05,
      "loss": 0.6446,
      "step": 1602
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7077596187591553,
      "learning_rate": 1.9738581851058606e-05,
      "loss": 0.7042,
      "step": 1603
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7106540203094482,
      "learning_rate": 1.9738118941189468e-05,
      "loss": 0.6595,
      "step": 1604
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.692631721496582,
      "learning_rate": 1.973765562726848e-05,
      "loss": 0.63,
      "step": 1605
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6871174573898315,
      "learning_rate": 1.9737191909314863e-05,
      "loss": 0.6199,
      "step": 1606
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5670678615570068,
      "learning_rate": 1.973672778734786e-05,
      "loss": 0.6469,
      "step": 1607
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5690906047821045,
      "learning_rate": 1.9736263261386725e-05,
      "loss": 0.643,
      "step": 1608
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6141633987426758,
      "learning_rate": 1.9735798331450734e-05,
      "loss": 0.6332,
      "step": 1609
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5695313215255737,
      "learning_rate": 1.9735332997559177e-05,
      "loss": 0.6487,
      "step": 1610
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5985397100448608,
      "learning_rate": 1.9734867259731363e-05,
      "loss": 0.6268,
      "step": 1611
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5526015758514404,
      "learning_rate": 1.9734401117986616e-05,
      "loss": 0.6748,
      "step": 1612
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6305898427963257,
      "learning_rate": 1.9733934572344276e-05,
      "loss": 0.6378,
      "step": 1613
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.578449010848999,
      "learning_rate": 1.9733467622823702e-05,
      "loss": 0.6147,
      "step": 1614
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5521190166473389,
      "learning_rate": 1.973300026944427e-05,
      "loss": 0.6521,
      "step": 1615
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.720716953277588,
      "learning_rate": 1.9732532512225366e-05,
      "loss": 0.602,
      "step": 1616
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6097791194915771,
      "learning_rate": 1.97320643511864e-05,
      "loss": 0.6805,
      "step": 1617
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6763701438903809,
      "learning_rate": 1.9731595786346796e-05,
      "loss": 0.6271,
      "step": 1618
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8110159635543823,
      "learning_rate": 1.9731126817726003e-05,
      "loss": 0.72,
      "step": 1619
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.617815613746643,
      "learning_rate": 1.9730657445343472e-05,
      "loss": 0.9686,
      "step": 1620
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.297443151473999,
      "learning_rate": 1.973018766921868e-05,
      "loss": 0.6475,
      "step": 1621
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.585216760635376,
      "learning_rate": 1.972971748937112e-05,
      "loss": 0.6897,
      "step": 1622
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.733683466911316,
      "learning_rate": 1.9729246905820293e-05,
      "loss": 0.6293,
      "step": 1623
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7001099586486816,
      "learning_rate": 1.972877591858574e-05,
      "loss": 0.7262,
      "step": 1624
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6537922620773315,
      "learning_rate": 1.9728304527686987e-05,
      "loss": 0.5992,
      "step": 1625
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6452440023422241,
      "learning_rate": 1.9727832733143603e-05,
      "loss": 0.6194,
      "step": 1626
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7490686178207397,
      "learning_rate": 1.9727360534975158e-05,
      "loss": 0.65,
      "step": 1627
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6597626209259033,
      "learning_rate": 1.972688793320125e-05,
      "loss": 0.6916,
      "step": 1628
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8600612878799438,
      "learning_rate": 1.9726414927841483e-05,
      "loss": 0.6657,
      "step": 1629
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8314331769943237,
      "learning_rate": 1.9725941518915483e-05,
      "loss": 0.6191,
      "step": 1630
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6574243307113647,
      "learning_rate": 1.97254677064429e-05,
      "loss": 0.6516,
      "step": 1631
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7753320932388306,
      "learning_rate": 1.9724993490443382e-05,
      "loss": 0.6599,
      "step": 1632
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6391408443450928,
      "learning_rate": 1.972451887093661e-05,
      "loss": 0.6751,
      "step": 1633
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6915323734283447,
      "learning_rate": 1.9724043847942278e-05,
      "loss": 0.6855,
      "step": 1634
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6595652103424072,
      "learning_rate": 1.9723568421480096e-05,
      "loss": 0.6349,
      "step": 1635
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.883194923400879,
      "learning_rate": 1.972309259156979e-05,
      "loss": 0.6202,
      "step": 1636
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6914689540863037,
      "learning_rate": 1.97226163582311e-05,
      "loss": 0.6192,
      "step": 1637
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.642015814781189,
      "learning_rate": 1.972213972148379e-05,
      "loss": 0.6674,
      "step": 1638
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5521681308746338,
      "learning_rate": 1.9721662681347635e-05,
      "loss": 0.6375,
      "step": 1639
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6664503812789917,
      "learning_rate": 1.9721185237842426e-05,
      "loss": 0.6116,
      "step": 1640
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4929553270339966,
      "learning_rate": 1.9720707390987974e-05,
      "loss": 0.6732,
      "step": 1641
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5855880975723267,
      "learning_rate": 1.972022914080411e-05,
      "loss": 0.6493,
      "step": 1642
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6868764162063599,
      "learning_rate": 1.9719750487310672e-05,
      "loss": 1.0193,
      "step": 1643
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.825431227684021,
      "learning_rate": 1.971927143052752e-05,
      "loss": 0.6744,
      "step": 1644
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6578447818756104,
      "learning_rate": 1.9718791970474535e-05,
      "loss": 0.6475,
      "step": 1645
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5310996770858765,
      "learning_rate": 1.971831210717161e-05,
      "loss": 0.682,
      "step": 1646
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.723103642463684,
      "learning_rate": 1.971783184063865e-05,
      "loss": 0.6472,
      "step": 1647
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6293331384658813,
      "learning_rate": 1.971735117089559e-05,
      "loss": 0.6396,
      "step": 1648
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8720080852508545,
      "learning_rate": 1.9716870097962368e-05,
      "loss": 0.6873,
      "step": 1649
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.802715539932251,
      "learning_rate": 1.971638862185895e-05,
      "loss": 0.6538,
      "step": 1650
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6592984199523926,
      "learning_rate": 1.9715906742605303e-05,
      "loss": 0.6414,
      "step": 1651
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.704066276550293,
      "learning_rate": 1.9715424460221433e-05,
      "loss": 0.6163,
      "step": 1652
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7567613124847412,
      "learning_rate": 1.9714941774727344e-05,
      "loss": 0.6446,
      "step": 1653
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6658477783203125,
      "learning_rate": 1.9714458686143065e-05,
      "loss": 0.6876,
      "step": 1654
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7998842000961304,
      "learning_rate": 1.971397519448864e-05,
      "loss": 0.6393,
      "step": 1655
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7463414669036865,
      "learning_rate": 1.9713491299784132e-05,
      "loss": 0.591,
      "step": 1656
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.728734016418457,
      "learning_rate": 1.9713007002049615e-05,
      "loss": 0.6755,
      "step": 1657
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7016966342926025,
      "learning_rate": 1.9712522301305186e-05,
      "loss": 0.5976,
      "step": 1658
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6629395484924316,
      "learning_rate": 1.9712037197570955e-05,
      "loss": 0.6807,
      "step": 1659
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6564538478851318,
      "learning_rate": 1.9711551690867054e-05,
      "loss": 0.6954,
      "step": 1660
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6952989101409912,
      "learning_rate": 1.9711065781213618e-05,
      "loss": 0.6816,
      "step": 1661
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6479401588439941,
      "learning_rate": 1.9710579468630814e-05,
      "loss": 0.6954,
      "step": 1662
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6613272428512573,
      "learning_rate": 1.9710092753138824e-05,
      "loss": 0.6637,
      "step": 1663
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7879266738891602,
      "learning_rate": 1.9709605634757836e-05,
      "loss": 0.6215,
      "step": 1664
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.637647032737732,
      "learning_rate": 1.9709118113508063e-05,
      "loss": 0.6717,
      "step": 1665
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0160958766937256,
      "learning_rate": 1.970863018940974e-05,
      "loss": 0.666,
      "step": 1666
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9312900304794312,
      "learning_rate": 1.97081418624831e-05,
      "loss": 0.6246,
      "step": 1667
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7194887399673462,
      "learning_rate": 1.9707653132748412e-05,
      "loss": 0.687,
      "step": 1668
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.675284743309021,
      "learning_rate": 1.9707164000225953e-05,
      "loss": 0.6204,
      "step": 1669
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6711689233779907,
      "learning_rate": 1.9706674464936017e-05,
      "loss": 1.0417,
      "step": 1670
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.052928924560547,
      "learning_rate": 1.970618452689892e-05,
      "loss": 0.641,
      "step": 1671
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9928721189498901,
      "learning_rate": 1.970569418613498e-05,
      "loss": 0.6585,
      "step": 1672
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4802119731903076,
      "learning_rate": 1.9705203442664555e-05,
      "loss": 0.623,
      "step": 1673
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9752463102340698,
      "learning_rate": 1.9704712296507997e-05,
      "loss": 0.6627,
      "step": 1674
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9150620698928833,
      "learning_rate": 1.970422074768569e-05,
      "loss": 0.6246,
      "step": 1675
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5911688804626465,
      "learning_rate": 1.970372879621803e-05,
      "loss": 0.6498,
      "step": 1676
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5608569383621216,
      "learning_rate": 1.9703236442125425e-05,
      "loss": 0.6501,
      "step": 1677
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6265324354171753,
      "learning_rate": 1.9702743685428305e-05,
      "loss": 0.6297,
      "step": 1678
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6301324367523193,
      "learning_rate": 1.970225052614711e-05,
      "loss": 0.6396,
      "step": 1679
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.555293083190918,
      "learning_rate": 1.9701756964302316e-05,
      "loss": 0.6807,
      "step": 1680
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6986455917358398,
      "learning_rate": 1.970126299991439e-05,
      "loss": 0.6404,
      "step": 1681
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6642838716506958,
      "learning_rate": 1.970076863300383e-05,
      "loss": 0.6389,
      "step": 1682
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4321805238723755,
      "learning_rate": 1.9700273863591152e-05,
      "loss": 0.6296,
      "step": 1683
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5799496173858643,
      "learning_rate": 1.9699778691696877e-05,
      "loss": 0.6859,
      "step": 1684
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.793521523475647,
      "learning_rate": 1.9699283117341558e-05,
      "loss": 0.6687,
      "step": 1685
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.590891718864441,
      "learning_rate": 1.969878714054575e-05,
      "loss": 0.6753,
      "step": 1686
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.504174828529358,
      "learning_rate": 1.9698290761330042e-05,
      "loss": 0.6786,
      "step": 1687
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.754580020904541,
      "learning_rate": 1.9697793979715025e-05,
      "loss": 0.6197,
      "step": 1688
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6168477535247803,
      "learning_rate": 1.969729679572131e-05,
      "loss": 0.6231,
      "step": 1689
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7471781969070435,
      "learning_rate": 1.9696799209369524e-05,
      "loss": 0.9798,
      "step": 1690
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0651707649230957,
      "learning_rate": 1.9696301220680315e-05,
      "loss": 0.6329,
      "step": 1691
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8067413568496704,
      "learning_rate": 1.9695802829674346e-05,
      "loss": 0.6564,
      "step": 1692
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6010533571243286,
      "learning_rate": 1.9695304036372298e-05,
      "loss": 0.6518,
      "step": 1693
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.771457314491272,
      "learning_rate": 1.9694804840794862e-05,
      "loss": 0.7342,
      "step": 1694
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8962322473526,
      "learning_rate": 1.9694305242962757e-05,
      "loss": 0.6642,
      "step": 1695
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6708745956420898,
      "learning_rate": 1.9693805242896708e-05,
      "loss": 0.667,
      "step": 1696
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6634050607681274,
      "learning_rate": 1.9693304840617456e-05,
      "loss": 0.6712,
      "step": 1697
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6325279474258423,
      "learning_rate": 1.9692804036145776e-05,
      "loss": 0.6834,
      "step": 1698
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6230462789535522,
      "learning_rate": 1.9692302829502435e-05,
      "loss": 0.6474,
      "step": 1699
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6422661542892456,
      "learning_rate": 1.9691801220708234e-05,
      "loss": 1.0155,
      "step": 1700
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0153439044952393,
      "learning_rate": 1.9691299209783992e-05,
      "loss": 0.6565,
      "step": 1701
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9209907054901123,
      "learning_rate": 1.9690796796750527e-05,
      "loss": 0.6215,
      "step": 1702
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6476970911026,
      "learning_rate": 1.969029398162869e-05,
      "loss": 0.5918,
      "step": 1703
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6047745943069458,
      "learning_rate": 1.9689790764439346e-05,
      "loss": 0.6445,
      "step": 1704
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6348490715026855,
      "learning_rate": 1.968928714520337e-05,
      "loss": 0.6522,
      "step": 1705
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.900153398513794,
      "learning_rate": 1.9688783123941663e-05,
      "loss": 0.6274,
      "step": 1706
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7029013633728027,
      "learning_rate": 1.9688278700675133e-05,
      "loss": 0.6133,
      "step": 1707
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.702127456665039,
      "learning_rate": 1.968777387542471e-05,
      "loss": 0.661,
      "step": 1708
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7150899171829224,
      "learning_rate": 1.9687268648211342e-05,
      "loss": 0.6467,
      "step": 1709
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6185747385025024,
      "learning_rate": 1.9686763019055993e-05,
      "loss": 0.6628,
      "step": 1710
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.72216796875,
      "learning_rate": 1.968625698797964e-05,
      "loss": 0.6297,
      "step": 1711
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3391928672790527,
      "learning_rate": 1.968575055500328e-05,
      "loss": 0.953,
      "step": 1712
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9440511465072632,
      "learning_rate": 1.9685243720147927e-05,
      "loss": 0.6319,
      "step": 1713
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6379855871200562,
      "learning_rate": 1.968473648343461e-05,
      "loss": 0.6482,
      "step": 1714
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5140790939331055,
      "learning_rate": 1.9684228844884367e-05,
      "loss": 0.6453,
      "step": 1715
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2250005006790161,
      "learning_rate": 1.9683720804518277e-05,
      "loss": 0.9854,
      "step": 1716
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.904559850692749,
      "learning_rate": 1.9683212362357406e-05,
      "loss": 0.683,
      "step": 1717
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7203562259674072,
      "learning_rate": 1.968270351842285e-05,
      "loss": 0.6432,
      "step": 1718
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5460872650146484,
      "learning_rate": 1.9682194272735735e-05,
      "loss": 0.6317,
      "step": 1719
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6739368438720703,
      "learning_rate": 1.968168462531718e-05,
      "loss": 0.6309,
      "step": 1720
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6952288150787354,
      "learning_rate": 1.968117457618833e-05,
      "loss": 0.6108,
      "step": 1721
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.660170316696167,
      "learning_rate": 1.9680664125370357e-05,
      "loss": 0.6369,
      "step": 1722
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5519218444824219,
      "learning_rate": 1.968015327288443e-05,
      "loss": 0.6778,
      "step": 1723
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.69758141040802,
      "learning_rate": 1.9679642018751753e-05,
      "loss": 0.6324,
      "step": 1724
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.607015609741211,
      "learning_rate": 1.967913036299353e-05,
      "loss": 0.6507,
      "step": 1725
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6277527809143066,
      "learning_rate": 1.9678618305631003e-05,
      "loss": 0.663,
      "step": 1726
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.648137092590332,
      "learning_rate": 1.9678105846685407e-05,
      "loss": 0.655,
      "step": 1727
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6273852586746216,
      "learning_rate": 1.9677592986178008e-05,
      "loss": 0.6618,
      "step": 1728
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.175276041030884,
      "learning_rate": 1.9677079724130094e-05,
      "loss": 0.6373,
      "step": 1729
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5227817296981812,
      "learning_rate": 1.9676566060562945e-05,
      "loss": 1.0293,
      "step": 1730
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8226498365402222,
      "learning_rate": 1.9676051995497886e-05,
      "loss": 0.668,
      "step": 1731
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.661276936531067,
      "learning_rate": 1.9675537528956242e-05,
      "loss": 0.6363,
      "step": 1732
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5406074523925781,
      "learning_rate": 1.9675022660959366e-05,
      "loss": 0.6553,
      "step": 1733
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5423011779785156,
      "learning_rate": 1.967450739152861e-05,
      "loss": 0.6133,
      "step": 1734
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4963219165802002,
      "learning_rate": 1.967399172068536e-05,
      "loss": 0.6319,
      "step": 1735
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2590924501419067,
      "learning_rate": 1.9673475648451008e-05,
      "loss": 0.9637,
      "step": 1736
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0335071086883545,
      "learning_rate": 1.9672959174846974e-05,
      "loss": 0.7186,
      "step": 1737
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.940742015838623,
      "learning_rate": 1.9672442299894678e-05,
      "loss": 0.6549,
      "step": 1738
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6176341772079468,
      "learning_rate": 1.9671925023615572e-05,
      "loss": 0.6257,
      "step": 1739
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6359614133834839,
      "learning_rate": 1.9671407346031118e-05,
      "loss": 0.653,
      "step": 1740
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6932148933410645,
      "learning_rate": 1.9670889267162796e-05,
      "loss": 0.6579,
      "step": 1741
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6944035291671753,
      "learning_rate": 1.96703707870321e-05,
      "loss": 0.6637,
      "step": 1742
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6330058574676514,
      "learning_rate": 1.9669851905660545e-05,
      "loss": 0.6423,
      "step": 1743
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6732476949691772,
      "learning_rate": 1.9669332623069656e-05,
      "loss": 0.644,
      "step": 1744
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6160789728164673,
      "learning_rate": 1.9668812939280985e-05,
      "loss": 0.6127,
      "step": 1745
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6117100715637207,
      "learning_rate": 1.9668292854316086e-05,
      "loss": 0.6879,
      "step": 1746
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6819188594818115,
      "learning_rate": 1.9667772368196547e-05,
      "loss": 0.9798,
      "step": 1747
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8549996614456177,
      "learning_rate": 1.9667251480943962e-05,
      "loss": 0.6316,
      "step": 1748
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6812409162521362,
      "learning_rate": 1.9666730192579942e-05,
      "loss": 0.6228,
      "step": 1749
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6554440259933472,
      "learning_rate": 1.9666208503126115e-05,
      "loss": 0.6706,
      "step": 1750
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.604457974433899,
      "learning_rate": 1.9665686412604127e-05,
      "loss": 0.6224,
      "step": 1751
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6696721315383911,
      "learning_rate": 1.9665163921035642e-05,
      "loss": 0.6556,
      "step": 1752
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.592177152633667,
      "learning_rate": 1.966464102844234e-05,
      "loss": 0.6138,
      "step": 1753
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9946403503417969,
      "learning_rate": 1.9664117734845917e-05,
      "loss": 0.6582,
      "step": 1754
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5058695077896118,
      "learning_rate": 1.9663594040268078e-05,
      "loss": 0.6589,
      "step": 1755
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.646147608757019,
      "learning_rate": 1.9663069944730564e-05,
      "loss": 0.6536,
      "step": 1756
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5897963047027588,
      "learning_rate": 1.966254544825511e-05,
      "loss": 0.6184,
      "step": 1757
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5361393690109253,
      "learning_rate": 1.9662020550863487e-05,
      "loss": 0.644,
      "step": 1758
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9069424867630005,
      "learning_rate": 1.9661495252577464e-05,
      "loss": 0.9827,
      "step": 1759
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.192535400390625,
      "learning_rate": 1.9660969553418848e-05,
      "loss": 0.5994,
      "step": 1760
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9708360433578491,
      "learning_rate": 1.9660443453409442e-05,
      "loss": 0.6156,
      "step": 1761
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.466417670249939,
      "learning_rate": 1.965991695257108e-05,
      "loss": 0.6267,
      "step": 1762
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8612103462219238,
      "learning_rate": 1.9659390050925606e-05,
      "loss": 0.6577,
      "step": 1763
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9019954204559326,
      "learning_rate": 1.965886274849488e-05,
      "loss": 0.6477,
      "step": 1764
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.644053339958191,
      "learning_rate": 1.965833504530078e-05,
      "loss": 0.6577,
      "step": 1765
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8114558458328247,
      "learning_rate": 1.9657806941365207e-05,
      "loss": 0.6495,
      "step": 1766
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5788687467575073,
      "learning_rate": 1.965727843671007e-05,
      "loss": 0.6367,
      "step": 1767
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.548911452293396,
      "learning_rate": 1.96567495313573e-05,
      "loss": 0.6297,
      "step": 1768
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7241593599319458,
      "learning_rate": 1.9656220225328834e-05,
      "loss": 0.6313,
      "step": 1769
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6516237258911133,
      "learning_rate": 1.965569051864664e-05,
      "loss": 0.6471,
      "step": 1770
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6286687850952148,
      "learning_rate": 1.96551604113327e-05,
      "loss": 0.6713,
      "step": 1771
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.756709098815918,
      "learning_rate": 1.9654629903409003e-05,
      "loss": 0.6249,
      "step": 1772
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5594900846481323,
      "learning_rate": 1.9654098994897564e-05,
      "loss": 0.6278,
      "step": 1773
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5309473276138306,
      "learning_rate": 1.9653567685820406e-05,
      "loss": 0.6284,
      "step": 1774
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6644238233566284,
      "learning_rate": 1.9653035976199585e-05,
      "loss": 0.642,
      "step": 1775
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5656847953796387,
      "learning_rate": 1.9652503866057148e-05,
      "loss": 0.6285,
      "step": 1776
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6060971021652222,
      "learning_rate": 1.9651971355415185e-05,
      "loss": 0.6312,
      "step": 1777
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6986069679260254,
      "learning_rate": 1.9651438444295788e-05,
      "loss": 0.6258,
      "step": 1778
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4703946113586426,
      "learning_rate": 1.9650905132721066e-05,
      "loss": 1.0554,
      "step": 1779
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8841336965560913,
      "learning_rate": 1.9650371420713145e-05,
      "loss": 0.656,
      "step": 1780
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5883135795593262,
      "learning_rate": 1.9649837308294176e-05,
      "loss": 0.6407,
      "step": 1781
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4851089715957642,
      "learning_rate": 1.9649302795486317e-05,
      "loss": 0.6376,
      "step": 1782
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.694785475730896,
      "learning_rate": 1.9648767882311745e-05,
      "loss": 0.657,
      "step": 1783
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5362141132354736,
      "learning_rate": 1.9648232568792657e-05,
      "loss": 0.6242,
      "step": 1784
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6404978036880493,
      "learning_rate": 1.964769685495126e-05,
      "loss": 0.645,
      "step": 1785
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6598621606826782,
      "learning_rate": 1.9647160740809786e-05,
      "loss": 0.6736,
      "step": 1786
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5532398223876953,
      "learning_rate": 1.9646624226390475e-05,
      "loss": 0.6592,
      "step": 1787
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6088274717330933,
      "learning_rate": 1.964608731171559e-05,
      "loss": 0.6037,
      "step": 1788
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7452713251113892,
      "learning_rate": 1.9645549996807415e-05,
      "loss": 0.6404,
      "step": 1789
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7307766675949097,
      "learning_rate": 1.9645012281688235e-05,
      "loss": 0.6396,
      "step": 1790
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5744413137435913,
      "learning_rate": 1.964447416638036e-05,
      "loss": 0.6181,
      "step": 1791
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7290503978729248,
      "learning_rate": 1.9643935650906127e-05,
      "loss": 0.6404,
      "step": 1792
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6087182760238647,
      "learning_rate": 1.9643396735287873e-05,
      "loss": 0.62,
      "step": 1793
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.588570475578308,
      "learning_rate": 1.964285741954796e-05,
      "loss": 0.6112,
      "step": 1794
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5601677894592285,
      "learning_rate": 1.9642317703708762e-05,
      "loss": 0.6257,
      "step": 1795
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6692818403244019,
      "learning_rate": 1.964177758779268e-05,
      "loss": 0.6372,
      "step": 1796
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.616292119026184,
      "learning_rate": 1.9641237071822115e-05,
      "loss": 0.6532,
      "step": 1797
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5141061544418335,
      "learning_rate": 1.9640696155819502e-05,
      "loss": 0.6296,
      "step": 1798
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.61655855178833,
      "learning_rate": 1.9640154839807285e-05,
      "loss": 0.6235,
      "step": 1799
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.559775710105896,
      "learning_rate": 1.9639613123807916e-05,
      "loss": 0.6391,
      "step": 1800
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4858742952346802,
      "learning_rate": 1.9639071007843878e-05,
      "loss": 0.6589,
      "step": 1801
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6440678834915161,
      "learning_rate": 1.9638528491937662e-05,
      "loss": 0.6193,
      "step": 1802
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5642192363739014,
      "learning_rate": 1.963798557611178e-05,
      "loss": 0.631,
      "step": 1803
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5793962478637695,
      "learning_rate": 1.963744226038876e-05,
      "loss": 0.6466,
      "step": 1804
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7545466423034668,
      "learning_rate": 1.9636898544791137e-05,
      "loss": 0.6844,
      "step": 1805
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5164222717285156,
      "learning_rate": 1.9636354429341483e-05,
      "loss": 0.6549,
      "step": 1806
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5362522602081299,
      "learning_rate": 1.963580991406236e-05,
      "loss": 0.6308,
      "step": 1807
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.484470009803772,
      "learning_rate": 1.9635264998976377e-05,
      "loss": 0.6511,
      "step": 1808
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4670648574829102,
      "learning_rate": 1.9634719684106128e-05,
      "loss": 0.6005,
      "step": 1809
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5545364618301392,
      "learning_rate": 1.9634173969474252e-05,
      "loss": 0.6191,
      "step": 1810
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.556053638458252,
      "learning_rate": 1.963362785510338e-05,
      "loss": 0.6637,
      "step": 1811
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6909821033477783,
      "learning_rate": 1.963308134101618e-05,
      "loss": 0.9734,
      "step": 1812
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8084651231765747,
      "learning_rate": 1.9632534427235326e-05,
      "loss": 0.6322,
      "step": 1813
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5841174125671387,
      "learning_rate": 1.9631987113783507e-05,
      "loss": 0.6396,
      "step": 1814
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5963627099990845,
      "learning_rate": 1.9631439400683433e-05,
      "loss": 0.6788,
      "step": 1815
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6051135063171387,
      "learning_rate": 1.9630891287957837e-05,
      "loss": 0.6677,
      "step": 1816
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5600180625915527,
      "learning_rate": 1.963034277562945e-05,
      "loss": 0.6679,
      "step": 1817
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6022696495056152,
      "learning_rate": 1.9629793863721035e-05,
      "loss": 0.6582,
      "step": 1818
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6949058771133423,
      "learning_rate": 1.962924455225537e-05,
      "loss": 0.6748,
      "step": 1819
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6855392456054688,
      "learning_rate": 1.962869484125525e-05,
      "loss": 0.6419,
      "step": 1820
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5071145296096802,
      "learning_rate": 1.9628144730743472e-05,
      "loss": 0.6734,
      "step": 1821
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6063677072525024,
      "learning_rate": 1.962759422074287e-05,
      "loss": 0.6546,
      "step": 1822
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6032187938690186,
      "learning_rate": 1.962704331127628e-05,
      "loss": 0.6543,
      "step": 1823
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5566415786743164,
      "learning_rate": 1.9626492002366567e-05,
      "loss": 0.6229,
      "step": 1824
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5659444332122803,
      "learning_rate": 1.9625940294036602e-05,
      "loss": 0.6498,
      "step": 1825
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.206806182861328,
      "learning_rate": 1.9625388186309275e-05,
      "loss": 1.0169,
      "step": 1826
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.915038824081421,
      "learning_rate": 1.9624835679207495e-05,
      "loss": 0.6207,
      "step": 1827
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8198981285095215,
      "learning_rate": 1.962428277275419e-05,
      "loss": 0.6011,
      "step": 1828
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.474717378616333,
      "learning_rate": 1.9623729466972295e-05,
      "loss": 0.6294,
      "step": 1829
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7050652503967285,
      "learning_rate": 1.9623175761884774e-05,
      "loss": 0.6684,
      "step": 1830
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6713130474090576,
      "learning_rate": 1.96226216575146e-05,
      "loss": 0.6582,
      "step": 1831
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.501413106918335,
      "learning_rate": 1.9622067153884754e-05,
      "loss": 0.6098,
      "step": 1832
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6282951831817627,
      "learning_rate": 1.962151225101826e-05,
      "loss": 0.6572,
      "step": 1833
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5814961194992065,
      "learning_rate": 1.9620956948938126e-05,
      "loss": 0.6277,
      "step": 1834
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5846585035324097,
      "learning_rate": 1.96204012476674e-05,
      "loss": 0.6509,
      "step": 1835
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6567327976226807,
      "learning_rate": 1.961984514722914e-05,
      "loss": 0.6468,
      "step": 1836
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7218972444534302,
      "learning_rate": 1.9619288647646417e-05,
      "loss": 0.6182,
      "step": 1837
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.473063349723816,
      "learning_rate": 1.9618731748942324e-05,
      "loss": 0.6193,
      "step": 1838
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.514858365058899,
      "learning_rate": 1.9618174451139963e-05,
      "loss": 0.6296,
      "step": 1839
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6165987253189087,
      "learning_rate": 1.9617616754262464e-05,
      "loss": 0.6455,
      "step": 1840
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.615574836730957,
      "learning_rate": 1.961705865833296e-05,
      "loss": 0.6143,
      "step": 1841
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5541223287582397,
      "learning_rate": 1.9616500163374615e-05,
      "loss": 0.6354,
      "step": 1842
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.670170545578003,
      "learning_rate": 1.9615941269410594e-05,
      "loss": 0.6283,
      "step": 1843
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6299015283584595,
      "learning_rate": 1.961538197646409e-05,
      "loss": 0.6216,
      "step": 1844
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6043411493301392,
      "learning_rate": 1.961482228455831e-05,
      "loss": 0.6469,
      "step": 1845
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5554760694503784,
      "learning_rate": 1.9614262193716475e-05,
      "loss": 1.0181,
      "step": 1846
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7528374195098877,
      "learning_rate": 1.9613701703961826e-05,
      "loss": 0.7208,
      "step": 1847
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.758342981338501,
      "learning_rate": 1.9613140815317614e-05,
      "loss": 0.6218,
      "step": 1848
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4911788702011108,
      "learning_rate": 1.9612579527807116e-05,
      "loss": 0.6671,
      "step": 1849
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5705171823501587,
      "learning_rate": 1.961201784145362e-05,
      "loss": 0.6189,
      "step": 1850
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.586118221282959,
      "learning_rate": 1.961145575628043e-05,
      "loss": 0.6355,
      "step": 1851
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4930561780929565,
      "learning_rate": 1.961089327231087e-05,
      "loss": 0.602,
      "step": 1852
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6895866394042969,
      "learning_rate": 1.961033038956828e-05,
      "loss": 0.626,
      "step": 1853
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.590221643447876,
      "learning_rate": 1.960976710807601e-05,
      "loss": 0.634,
      "step": 1854
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6278198957443237,
      "learning_rate": 1.9609203427857433e-05,
      "loss": 0.6541,
      "step": 1855
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5800740718841553,
      "learning_rate": 1.9608639348935938e-05,
      "loss": 0.6437,
      "step": 1856
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7173928022384644,
      "learning_rate": 1.960807487133493e-05,
      "loss": 0.6547,
      "step": 1857
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6017751693725586,
      "learning_rate": 1.960750999507783e-05,
      "loss": 0.6475,
      "step": 1858
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.764112114906311,
      "learning_rate": 1.9606944720188073e-05,
      "loss": 1.0225,
      "step": 1859
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9528067111968994,
      "learning_rate": 1.960637904668912e-05,
      "loss": 0.6659,
      "step": 1860
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.652669906616211,
      "learning_rate": 1.9605812974604434e-05,
      "loss": 0.6597,
      "step": 1861
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.266555666923523,
      "learning_rate": 1.9605246503957508e-05,
      "loss": 1.0126,
      "step": 1862
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8664138317108154,
      "learning_rate": 1.9604679634771842e-05,
      "loss": 0.6089,
      "step": 1863
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7629282474517822,
      "learning_rate": 1.960411236707096e-05,
      "loss": 0.6516,
      "step": 1864
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4764189720153809,
      "learning_rate": 1.9603544700878395e-05,
      "loss": 0.5973,
      "step": 1865
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.732073426246643,
      "learning_rate": 1.96029766362177e-05,
      "loss": 0.6373,
      "step": 1866
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6391854286193848,
      "learning_rate": 1.9602408173112453e-05,
      "loss": 0.6304,
      "step": 1867
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.621830701828003,
      "learning_rate": 1.9601839311586236e-05,
      "loss": 0.6496,
      "step": 1868
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6034437417984009,
      "learning_rate": 1.9601270051662646e-05,
      "loss": 0.6099,
      "step": 1869
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6384997367858887,
      "learning_rate": 1.960070039336531e-05,
      "loss": 0.6332,
      "step": 1870
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3295719623565674,
      "learning_rate": 1.9600130336717864e-05,
      "loss": 1.0562,
      "step": 1871
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9372717142105103,
      "learning_rate": 1.9599559881743957e-05,
      "loss": 0.5968,
      "step": 1872
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7629262208938599,
      "learning_rate": 1.959898902846726e-05,
      "loss": 0.6518,
      "step": 1873
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6259537935256958,
      "learning_rate": 1.9598417776911455e-05,
      "loss": 0.6145,
      "step": 1874
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6194084882736206,
      "learning_rate": 1.959784612710025e-05,
      "loss": 0.6576,
      "step": 1875
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6512900590896606,
      "learning_rate": 1.9597274079057365e-05,
      "loss": 0.609,
      "step": 1876
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6131761074066162,
      "learning_rate": 1.959670163280653e-05,
      "loss": 0.6491,
      "step": 1877
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7188040018081665,
      "learning_rate": 1.9596128788371494e-05,
      "loss": 0.6274,
      "step": 1878
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.501775860786438,
      "learning_rate": 1.9595555545776036e-05,
      "loss": 0.6183,
      "step": 1879
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6853128671646118,
      "learning_rate": 1.959498190504393e-05,
      "loss": 0.6713,
      "step": 1880
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.587325930595398,
      "learning_rate": 1.9594407866198984e-05,
      "loss": 0.6453,
      "step": 1881
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5629048347473145,
      "learning_rate": 1.9593833429265014e-05,
      "loss": 0.6949,
      "step": 1882
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6835944652557373,
      "learning_rate": 1.9593258594265854e-05,
      "loss": 0.6146,
      "step": 1883
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5746147632598877,
      "learning_rate": 1.959268336122536e-05,
      "loss": 0.6389,
      "step": 1884
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5325813293457031,
      "learning_rate": 1.959210773016739e-05,
      "loss": 0.5906,
      "step": 1885
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7668867111206055,
      "learning_rate": 1.9591531701115834e-05,
      "loss": 0.6134,
      "step": 1886
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6509819030761719,
      "learning_rate": 1.959095527409459e-05,
      "loss": 0.6133,
      "step": 1887
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.596922516822815,
      "learning_rate": 1.9590378449127576e-05,
      "loss": 0.6265,
      "step": 1888
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7999322414398193,
      "learning_rate": 1.958980122623873e-05,
      "loss": 0.6056,
      "step": 1889
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7176101207733154,
      "learning_rate": 1.9589223605451993e-05,
      "loss": 0.6026,
      "step": 1890
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5205647945404053,
      "learning_rate": 1.9588645586791338e-05,
      "loss": 0.6152,
      "step": 1891
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8721591234207153,
      "learning_rate": 1.9588067170280744e-05,
      "loss": 1.0285,
      "step": 1892
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.349778175354004,
      "learning_rate": 1.9587488355944216e-05,
      "loss": 0.6447,
      "step": 1893
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8500139713287354,
      "learning_rate": 1.9586909143805766e-05,
      "loss": 0.6135,
      "step": 1894
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.994543433189392,
      "learning_rate": 1.958632953388943e-05,
      "loss": 0.6684,
      "step": 1895
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0513572692871094,
      "learning_rate": 1.9585749526219248e-05,
      "loss": 0.649,
      "step": 1896
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.690402865409851,
      "learning_rate": 1.9585169120819298e-05,
      "loss": 0.6504,
      "step": 1897
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.433510661125183,
      "learning_rate": 1.9584588317713652e-05,
      "loss": 1.0114,
      "step": 1898
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5326340198516846,
      "learning_rate": 1.9584007116926413e-05,
      "loss": 0.6595,
      "step": 1899
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3439433574676514,
      "learning_rate": 1.95834255184817e-05,
      "loss": 0.6199,
      "step": 1900
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.651286244392395,
      "learning_rate": 1.9582843522403632e-05,
      "loss": 0.6056,
      "step": 1901
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.861422061920166,
      "learning_rate": 1.9582261128716374e-05,
      "loss": 0.6166,
      "step": 1902
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9665899276733398,
      "learning_rate": 1.9581678337444075e-05,
      "loss": 0.6084,
      "step": 1903
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5779290199279785,
      "learning_rate": 1.9581095148610925e-05,
      "loss": 0.6337,
      "step": 1904
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7935161590576172,
      "learning_rate": 1.958051156224112e-05,
      "loss": 0.6412,
      "step": 1905
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7666065692901611,
      "learning_rate": 1.9579927578358876e-05,
      "loss": 0.6678,
      "step": 1906
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7972745895385742,
      "learning_rate": 1.9579343196988417e-05,
      "loss": 0.6277,
      "step": 1907
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6601494550704956,
      "learning_rate": 1.9578758418153994e-05,
      "loss": 0.6679,
      "step": 1908
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.695910930633545,
      "learning_rate": 1.957817324187987e-05,
      "loss": 0.6317,
      "step": 1909
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8784078359603882,
      "learning_rate": 1.957758766819033e-05,
      "loss": 0.6483,
      "step": 1910
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6866759061813354,
      "learning_rate": 1.9577001697109662e-05,
      "loss": 0.6593,
      "step": 1911
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6632134914398193,
      "learning_rate": 1.9576415328662186e-05,
      "loss": 0.6658,
      "step": 1912
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8437700271606445,
      "learning_rate": 1.9575828562872224e-05,
      "loss": 0.6626,
      "step": 1913
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.132338762283325,
      "learning_rate": 1.957524139976413e-05,
      "loss": 0.6456,
      "step": 1914
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6644741296768188,
      "learning_rate": 1.957465383936226e-05,
      "loss": 0.6488,
      "step": 1915
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6177040338516235,
      "learning_rate": 1.9574065881690998e-05,
      "loss": 0.6258,
      "step": 1916
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.584883689880371,
      "learning_rate": 1.9573477526774735e-05,
      "loss": 0.6332,
      "step": 1917
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6740140914916992,
      "learning_rate": 1.9572888774637886e-05,
      "loss": 0.6495,
      "step": 1918
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6064567565917969,
      "learning_rate": 1.957229962530488e-05,
      "loss": 0.6422,
      "step": 1919
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7750422954559326,
      "learning_rate": 1.9571710078800158e-05,
      "loss": 0.9775,
      "step": 1920
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.842453956604004,
      "learning_rate": 1.9571120135148184e-05,
      "loss": 0.6289,
      "step": 1921
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7397902011871338,
      "learning_rate": 1.9570529794373435e-05,
      "loss": 0.6489,
      "step": 1922
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.585623860359192,
      "learning_rate": 1.9569939056500408e-05,
      "loss": 0.6184,
      "step": 1923
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.597822904586792,
      "learning_rate": 1.956934792155361e-05,
      "loss": 0.643,
      "step": 1924
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5695255994796753,
      "learning_rate": 1.9568756389557572e-05,
      "loss": 0.6331,
      "step": 1925
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7787524461746216,
      "learning_rate": 1.956816446053683e-05,
      "loss": 0.6046,
      "step": 1926
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6429986953735352,
      "learning_rate": 1.9567572134515953e-05,
      "loss": 0.6599,
      "step": 1927
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5467984676361084,
      "learning_rate": 1.9566979411519512e-05,
      "loss": 0.6314,
      "step": 1928
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5258978605270386,
      "learning_rate": 1.9566386291572106e-05,
      "loss": 0.6595,
      "step": 1929
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6358094215393066,
      "learning_rate": 1.956579277469834e-05,
      "loss": 0.6421,
      "step": 1930
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5614904165267944,
      "learning_rate": 1.956519886092284e-05,
      "loss": 0.5953,
      "step": 1931
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5178961753845215,
      "learning_rate": 1.9564604550270248e-05,
      "loss": 0.6753,
      "step": 1932
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.571166753768921,
      "learning_rate": 1.9564009842765225e-05,
      "loss": 0.6135,
      "step": 1933
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5090402364730835,
      "learning_rate": 1.956341473843245e-05,
      "loss": 0.6291,
      "step": 1934
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4808114767074585,
      "learning_rate": 1.9562819237296606e-05,
      "loss": 0.6125,
      "step": 1935
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5914050340652466,
      "learning_rate": 1.9562223339382405e-05,
      "loss": 0.6602,
      "step": 1936
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5272834300994873,
      "learning_rate": 1.956162704471458e-05,
      "loss": 0.631,
      "step": 1937
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.556242823600769,
      "learning_rate": 1.9561030353317863e-05,
      "loss": 0.6341,
      "step": 1938
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5670592784881592,
      "learning_rate": 1.956043326521701e-05,
      "loss": 0.6318,
      "step": 1939
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4983116388320923,
      "learning_rate": 1.9559835780436806e-05,
      "loss": 0.6345,
      "step": 1940
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5088508129119873,
      "learning_rate": 1.9559237899002032e-05,
      "loss": 0.6267,
      "step": 1941
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5172817707061768,
      "learning_rate": 1.95586396209375e-05,
      "loss": 0.6077,
      "step": 1942
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4853143692016602,
      "learning_rate": 1.955804094626803e-05,
      "loss": 0.6548,
      "step": 1943
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.611865758895874,
      "learning_rate": 1.9557441875018464e-05,
      "loss": 0.6426,
      "step": 1944
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6276917457580566,
      "learning_rate": 1.955684240721366e-05,
      "loss": 0.6461,
      "step": 1945
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5248284339904785,
      "learning_rate": 1.955624254287849e-05,
      "loss": 0.6521,
      "step": 1946
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4849127531051636,
      "learning_rate": 1.9555642282037842e-05,
      "loss": 0.6006,
      "step": 1947
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5630793571472168,
      "learning_rate": 1.9555041624716625e-05,
      "loss": 0.6244,
      "step": 1948
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.553552508354187,
      "learning_rate": 1.955444057093976e-05,
      "loss": 0.6335,
      "step": 1949
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5073318481445312,
      "learning_rate": 1.9553839120732182e-05,
      "loss": 0.6238,
      "step": 1950
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5160834789276123,
      "learning_rate": 1.955323727411885e-05,
      "loss": 0.642,
      "step": 1951
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.558510661125183,
      "learning_rate": 1.9552635031124736e-05,
      "loss": 0.629,
      "step": 1952
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8121320009231567,
      "learning_rate": 1.9552032391774825e-05,
      "loss": 1.0083,
      "step": 1953
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.186514139175415,
      "learning_rate": 1.9551429356094125e-05,
      "loss": 0.6075,
      "step": 1954
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9256466627120972,
      "learning_rate": 1.9550825924107654e-05,
      "loss": 0.626,
      "step": 1955
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6473146677017212,
      "learning_rate": 1.9550222095840455e-05,
      "loss": 0.636,
      "step": 1956
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9306665658950806,
      "learning_rate": 1.9549617871317578e-05,
      "loss": 0.6333,
      "step": 1957
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8775427341461182,
      "learning_rate": 1.954901325056409e-05,
      "loss": 0.6446,
      "step": 1958
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.64423406124115,
      "learning_rate": 1.9548408233605083e-05,
      "loss": 0.6287,
      "step": 1959
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6634118556976318,
      "learning_rate": 1.954780282046566e-05,
      "loss": 0.6429,
      "step": 1960
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7322773933410645,
      "learning_rate": 1.9547197011170938e-05,
      "loss": 0.641,
      "step": 1961
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8744863271713257,
      "learning_rate": 1.9546590805746054e-05,
      "loss": 0.6609,
      "step": 1962
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8398678302764893,
      "learning_rate": 1.9545984204216162e-05,
      "loss": 0.6251,
      "step": 1963
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.685462474822998,
      "learning_rate": 1.9545377206606425e-05,
      "loss": 0.6325,
      "step": 1964
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6006460189819336,
      "learning_rate": 1.9544769812942038e-05,
      "loss": 0.6432,
      "step": 1965
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7592488527297974,
      "learning_rate": 1.95441620232482e-05,
      "loss": 0.6241,
      "step": 1966
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.648193359375,
      "learning_rate": 1.9543553837550124e-05,
      "loss": 0.6158,
      "step": 1967
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5135140419006348,
      "learning_rate": 1.9542945255873046e-05,
      "loss": 0.629,
      "step": 1968
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7147685289382935,
      "learning_rate": 1.9542336278242225e-05,
      "loss": 0.6416,
      "step": 1969
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5346014499664307,
      "learning_rate": 1.954172690468292e-05,
      "loss": 0.6479,
      "step": 1970
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5609712600708008,
      "learning_rate": 1.9541117135220412e-05,
      "loss": 0.6194,
      "step": 1971
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5905380249023438,
      "learning_rate": 1.9540506969880013e-05,
      "loss": 0.633,
      "step": 1972
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6129168272018433,
      "learning_rate": 1.9539896408687032e-05,
      "loss": 0.6144,
      "step": 1973
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.507163643836975,
      "learning_rate": 1.9539285451666807e-05,
      "loss": 0.65,
      "step": 1974
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5820428133010864,
      "learning_rate": 1.9538674098844685e-05,
      "loss": 0.6226,
      "step": 1975
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6012574434280396,
      "learning_rate": 1.9538062350246028e-05,
      "loss": 0.634,
      "step": 1976
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6581209897994995,
      "learning_rate": 1.9537450205896227e-05,
      "loss": 0.6521,
      "step": 1977
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5401358604431152,
      "learning_rate": 1.953683766582067e-05,
      "loss": 0.6025,
      "step": 1978
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.741025447845459,
      "learning_rate": 1.9536224730044785e-05,
      "loss": 0.6311,
      "step": 1979
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.491789698600769,
      "learning_rate": 1.9535611398594e-05,
      "loss": 0.6276,
      "step": 1980
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.589314579963684,
      "learning_rate": 1.9534997671493755e-05,
      "loss": 0.6139,
      "step": 1981
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5815136432647705,
      "learning_rate": 1.9534383548769523e-05,
      "loss": 0.6198,
      "step": 1982
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6330947875976562,
      "learning_rate": 1.953376903044678e-05,
      "loss": 0.6339,
      "step": 1983
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5516939163208008,
      "learning_rate": 1.953315411655103e-05,
      "loss": 0.6354,
      "step": 1984
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5854264497756958,
      "learning_rate": 1.953253880710778e-05,
      "loss": 0.6274,
      "step": 1985
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5277897119522095,
      "learning_rate": 1.9531923102142563e-05,
      "loss": 0.5894,
      "step": 1986
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5024960041046143,
      "learning_rate": 1.953130700168093e-05,
      "loss": 0.6567,
      "step": 1987
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5665045976638794,
      "learning_rate": 1.9530690505748436e-05,
      "loss": 0.616,
      "step": 1988
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5654850006103516,
      "learning_rate": 1.9530073614370666e-05,
      "loss": 0.6136,
      "step": 1989
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5363049507141113,
      "learning_rate": 1.9529456327573214e-05,
      "loss": 0.5813,
      "step": 1990
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5615665912628174,
      "learning_rate": 1.9528838645381693e-05,
      "loss": 0.6424,
      "step": 1991
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5880707502365112,
      "learning_rate": 1.9528220567821728e-05,
      "loss": 0.6011,
      "step": 1992
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.592686414718628,
      "learning_rate": 1.9527602094918974e-05,
      "loss": 0.638,
      "step": 1993
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.62847101688385,
      "learning_rate": 1.952698322669908e-05,
      "loss": 0.6371,
      "step": 1994
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5807461738586426,
      "learning_rate": 1.9526363963187736e-05,
      "loss": 0.6137,
      "step": 1995
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4654966592788696,
      "learning_rate": 1.9525744304410626e-05,
      "loss": 0.6211,
      "step": 1996
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7051575183868408,
      "learning_rate": 1.952512425039347e-05,
      "loss": 0.5953,
      "step": 1997
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.703739047050476,
      "learning_rate": 1.9524503801161984e-05,
      "loss": 0.6055,
      "step": 1998
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5044026374816895,
      "learning_rate": 1.952388295674192e-05,
      "loss": 0.6133,
      "step": 1999
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3999308347702026,
      "learning_rate": 1.952326171715904e-05,
      "loss": 0.6173,
      "step": 2000
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1164071559906006,
      "learning_rate": 1.952264008243911e-05,
      "loss": 0.9917,
      "step": 2001
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8772205114364624,
      "learning_rate": 1.9522018052607935e-05,
      "loss": 0.6455,
      "step": 2002
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2607421875,
      "learning_rate": 1.952139562769131e-05,
      "loss": 1.0091,
      "step": 2003
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8890597820281982,
      "learning_rate": 1.9520772807715075e-05,
      "loss": 0.6532,
      "step": 2004
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5741878747940063,
      "learning_rate": 1.9520149592705064e-05,
      "loss": 0.6322,
      "step": 2005
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5779742002487183,
      "learning_rate": 1.9519525982687137e-05,
      "loss": 0.6336,
      "step": 2006
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6615947484970093,
      "learning_rate": 1.951890197768717e-05,
      "loss": 0.6392,
      "step": 2007
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6825981140136719,
      "learning_rate": 1.951827757773105e-05,
      "loss": 0.6145,
      "step": 2008
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6543662548065186,
      "learning_rate": 1.9517652782844685e-05,
      "loss": 0.6371,
      "step": 2009
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6794084310531616,
      "learning_rate": 1.9517027593054004e-05,
      "loss": 0.5988,
      "step": 2010
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6033406257629395,
      "learning_rate": 1.9516402008384942e-05,
      "loss": 0.6256,
      "step": 2011
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.783955693244934,
      "learning_rate": 1.951577602886346e-05,
      "loss": 0.6457,
      "step": 2012
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6825066804885864,
      "learning_rate": 1.9515149654515528e-05,
      "loss": 0.6005,
      "step": 2013
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.365427017211914,
      "learning_rate": 1.9514522885367135e-05,
      "loss": 1.0484,
      "step": 2014
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0611231327056885,
      "learning_rate": 1.9513895721444286e-05,
      "loss": 0.6263,
      "step": 2015
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9567289352416992,
      "learning_rate": 1.9513268162773008e-05,
      "loss": 0.6358,
      "step": 2016
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5318909883499146,
      "learning_rate": 1.9512640209379334e-05,
      "loss": 0.6242,
      "step": 2017
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7264320850372314,
      "learning_rate": 1.9512011861289324e-05,
      "loss": 0.6641,
      "step": 2018
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6272245645523071,
      "learning_rate": 1.9511383118529045e-05,
      "loss": 0.6771,
      "step": 2019
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6002869606018066,
      "learning_rate": 1.951075398112459e-05,
      "loss": 0.629,
      "step": 2020
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.631064772605896,
      "learning_rate": 1.9510124449102053e-05,
      "loss": 0.6924,
      "step": 2021
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5585236549377441,
      "learning_rate": 1.9509494522487564e-05,
      "loss": 0.6334,
      "step": 2022
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6957229375839233,
      "learning_rate": 1.9508864201307257e-05,
      "loss": 0.6286,
      "step": 2023
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.787550449371338,
      "learning_rate": 1.9508233485587286e-05,
      "loss": 0.6192,
      "step": 2024
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.549975872039795,
      "learning_rate": 1.9507602375353816e-05,
      "loss": 0.616,
      "step": 2025
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5789157152175903,
      "learning_rate": 1.9506970870633036e-05,
      "loss": 0.5952,
      "step": 2026
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5774344205856323,
      "learning_rate": 1.950633897145115e-05,
      "loss": 0.6133,
      "step": 2027
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5235916376113892,
      "learning_rate": 1.9505706677834374e-05,
      "loss": 0.6312,
      "step": 2028
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.635716438293457,
      "learning_rate": 1.9505073989808946e-05,
      "loss": 0.6233,
      "step": 2029
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6143194437026978,
      "learning_rate": 1.9504440907401113e-05,
      "loss": 0.609,
      "step": 2030
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5870429277420044,
      "learning_rate": 1.9503807430637145e-05,
      "loss": 0.5969,
      "step": 2031
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.572247862815857,
      "learning_rate": 1.9503173559543327e-05,
      "loss": 0.6501,
      "step": 2032
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5434691905975342,
      "learning_rate": 1.9502539294145955e-05,
      "loss": 0.6397,
      "step": 2033
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6650903224945068,
      "learning_rate": 1.9501904634471355e-05,
      "loss": 0.6599,
      "step": 2034
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6539006233215332,
      "learning_rate": 1.9501269580545853e-05,
      "loss": 0.6372,
      "step": 2035
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.555932879447937,
      "learning_rate": 1.9500634132395797e-05,
      "loss": 0.6259,
      "step": 2036
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3049196004867554,
      "learning_rate": 1.949999829004756e-05,
      "loss": 0.9907,
      "step": 2037
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8829948902130127,
      "learning_rate": 1.9499362053527516e-05,
      "loss": 0.6533,
      "step": 2038
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6787971258163452,
      "learning_rate": 1.9498725422862068e-05,
      "loss": 0.6236,
      "step": 2039
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5640015602111816,
      "learning_rate": 1.9498088398077634e-05,
      "loss": 0.6184,
      "step": 2040
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.618179440498352,
      "learning_rate": 1.949745097920064e-05,
      "loss": 0.5964,
      "step": 2041
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8276844024658203,
      "learning_rate": 1.9496813166257533e-05,
      "loss": 0.5992,
      "step": 2042
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6502633094787598,
      "learning_rate": 1.9496174959274784e-05,
      "loss": 0.6321,
      "step": 2043
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5390795469284058,
      "learning_rate": 1.9495536358278865e-05,
      "loss": 0.6365,
      "step": 2044
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6547483205795288,
      "learning_rate": 1.949489736329628e-05,
      "loss": 0.6462,
      "step": 2045
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.697027325630188,
      "learning_rate": 1.9494257974353536e-05,
      "loss": 0.6378,
      "step": 2046
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5540897846221924,
      "learning_rate": 1.9493618191477163e-05,
      "loss": 0.6499,
      "step": 2047
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6216444969177246,
      "learning_rate": 1.9492978014693712e-05,
      "loss": 0.6416,
      "step": 2048
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6015392541885376,
      "learning_rate": 1.9492337444029742e-05,
      "loss": 0.6321,
      "step": 2049
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6295539140701294,
      "learning_rate": 1.949169647951183e-05,
      "loss": 0.6073,
      "step": 2050
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5550427436828613,
      "learning_rate": 1.949105512116657e-05,
      "loss": 0.6074,
      "step": 2051
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.575610637664795,
      "learning_rate": 1.9490413369020577e-05,
      "loss": 0.6629,
      "step": 2052
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.621018648147583,
      "learning_rate": 1.9489771223100477e-05,
      "loss": 0.6771,
      "step": 2053
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5555435419082642,
      "learning_rate": 1.948912868343291e-05,
      "loss": 0.5632,
      "step": 2054
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5193355083465576,
      "learning_rate": 1.9488485750044545e-05,
      "loss": 0.6237,
      "step": 2055
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5394132137298584,
      "learning_rate": 1.9487842422962048e-05,
      "loss": 0.6298,
      "step": 2056
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5366710424423218,
      "learning_rate": 1.948719870221212e-05,
      "loss": 0.6112,
      "step": 2057
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4655029773712158,
      "learning_rate": 1.9486554587821467e-05,
      "loss": 0.6472,
      "step": 2058
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4816045761108398,
      "learning_rate": 1.9485910079816812e-05,
      "loss": 0.6437,
      "step": 2059
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5147433280944824,
      "learning_rate": 1.9485265178224896e-05,
      "loss": 0.5963,
      "step": 2060
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.489362120628357,
      "learning_rate": 1.9484619883072487e-05,
      "loss": 0.6253,
      "step": 2061
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5276708602905273,
      "learning_rate": 1.9483974194386347e-05,
      "loss": 0.6497,
      "step": 2062
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.579416036605835,
      "learning_rate": 1.9483328112193275e-05,
      "loss": 0.6282,
      "step": 2063
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5614020824432373,
      "learning_rate": 1.9482681636520073e-05,
      "loss": 0.6158,
      "step": 2064
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4950448274612427,
      "learning_rate": 1.948203476739357e-05,
      "loss": 0.5984,
      "step": 2065
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5622445344924927,
      "learning_rate": 1.94813875048406e-05,
      "loss": 0.6284,
      "step": 2066
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6128485202789307,
      "learning_rate": 1.9480739848888022e-05,
      "loss": 0.6253,
      "step": 2067
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.596803069114685,
      "learning_rate": 1.9480091799562706e-05,
      "loss": 0.6598,
      "step": 2068
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4532365798950195,
      "learning_rate": 1.9479443356891545e-05,
      "loss": 0.6084,
      "step": 2069
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4890092611312866,
      "learning_rate": 1.947879452090144e-05,
      "loss": 0.6357,
      "step": 2070
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5227746963500977,
      "learning_rate": 1.9478145291619317e-05,
      "loss": 0.6496,
      "step": 2071
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6288138628005981,
      "learning_rate": 1.9477495669072112e-05,
      "loss": 0.646,
      "step": 2072
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4327794313430786,
      "learning_rate": 1.947684565328677e-05,
      "loss": 0.6303,
      "step": 2073
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.564518690109253,
      "learning_rate": 1.9476195244290277e-05,
      "loss": 0.6214,
      "step": 2074
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5217660665512085,
      "learning_rate": 1.947554444210961e-05,
      "loss": 0.6521,
      "step": 2075
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5279710292816162,
      "learning_rate": 1.947489324677177e-05,
      "loss": 0.648,
      "step": 2076
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5969204902648926,
      "learning_rate": 1.9474241658303784e-05,
      "loss": 0.6451,
      "step": 2077
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5327110290527344,
      "learning_rate": 1.9473589676732683e-05,
      "loss": 0.6318,
      "step": 2078
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7412254810333252,
      "learning_rate": 1.9472937302085518e-05,
      "loss": 0.6217,
      "step": 2079
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.501205325126648,
      "learning_rate": 1.947228453438936e-05,
      "loss": 0.6343,
      "step": 2080
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4588029384613037,
      "learning_rate": 1.9471631373671288e-05,
      "loss": 0.6525,
      "step": 2081
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5416847467422485,
      "learning_rate": 1.947097781995841e-05,
      "loss": 0.641,
      "step": 2082
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6450968980789185,
      "learning_rate": 1.9470323873277838e-05,
      "loss": 0.598,
      "step": 2083
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4462167024612427,
      "learning_rate": 1.946966953365671e-05,
      "loss": 0.6291,
      "step": 2084
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5298186540603638,
      "learning_rate": 1.9469014801122174e-05,
      "loss": 0.6039,
      "step": 2085
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5451021194458008,
      "learning_rate": 1.946835967570139e-05,
      "loss": 0.6161,
      "step": 2086
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4718440771102905,
      "learning_rate": 1.946770415742155e-05,
      "loss": 0.6209,
      "step": 2087
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2153127193450928,
      "learning_rate": 1.9467048246309846e-05,
      "loss": 0.6424,
      "step": 2088
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5848004817962646,
      "learning_rate": 1.9466391942393495e-05,
      "loss": 0.6301,
      "step": 2089
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.538470983505249,
      "learning_rate": 1.9465735245699726e-05,
      "loss": 0.6526,
      "step": 2090
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4983986616134644,
      "learning_rate": 1.946507815625579e-05,
      "loss": 0.622,
      "step": 2091
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7089701890945435,
      "learning_rate": 1.946442067408895e-05,
      "loss": 0.6295,
      "step": 2092
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7355395555496216,
      "learning_rate": 1.9463762799226487e-05,
      "loss": 0.6367,
      "step": 2093
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5703219175338745,
      "learning_rate": 1.9463104531695694e-05,
      "loss": 0.6224,
      "step": 2094
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5910365581512451,
      "learning_rate": 1.9462445871523887e-05,
      "loss": 0.6401,
      "step": 2095
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.50626540184021,
      "learning_rate": 1.9461786818738393e-05,
      "loss": 0.6078,
      "step": 2096
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4691215753555298,
      "learning_rate": 1.9461127373366556e-05,
      "loss": 0.6106,
      "step": 2097
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.687273621559143,
      "learning_rate": 1.9460467535435744e-05,
      "loss": 0.609,
      "step": 2098
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5621130466461182,
      "learning_rate": 1.9459807304973327e-05,
      "loss": 0.6263,
      "step": 2099
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.697347640991211,
      "learning_rate": 1.94591466820067e-05,
      "loss": 0.6342,
      "step": 2100
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5835057497024536,
      "learning_rate": 1.945848566656328e-05,
      "loss": 0.6138,
      "step": 2101
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4626880884170532,
      "learning_rate": 1.9457824258670487e-05,
      "loss": 0.6084,
      "step": 2102
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4451435804367065,
      "learning_rate": 1.945716245835577e-05,
      "loss": 1.0016,
      "step": 2103
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8573567867279053,
      "learning_rate": 1.945650026564658e-05,
      "loss": 0.6108,
      "step": 2104
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5926761627197266,
      "learning_rate": 1.94558376805704e-05,
      "loss": 0.6049,
      "step": 2105
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5854655504226685,
      "learning_rate": 1.945517470315472e-05,
      "loss": 0.6315,
      "step": 2106
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8556722402572632,
      "learning_rate": 1.9454511333427047e-05,
      "loss": 0.618,
      "step": 2107
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7009766101837158,
      "learning_rate": 1.9453847571414903e-05,
      "loss": 0.638,
      "step": 2108
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6693135499954224,
      "learning_rate": 1.9453183417145833e-05,
      "loss": 0.6117,
      "step": 2109
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8093798160552979,
      "learning_rate": 1.9452518870647395e-05,
      "loss": 0.6544,
      "step": 2110
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.626098871231079,
      "learning_rate": 1.9451853931947157e-05,
      "loss": 0.6456,
      "step": 2111
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8292959928512573,
      "learning_rate": 1.945118860107271e-05,
      "loss": 0.5968,
      "step": 2112
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8346123695373535,
      "learning_rate": 1.945052287805166e-05,
      "loss": 0.6272,
      "step": 2113
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.568688154220581,
      "learning_rate": 1.9449856762911633e-05,
      "loss": 0.6244,
      "step": 2114
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6316767930984497,
      "learning_rate": 1.9449190255680262e-05,
      "loss": 0.6015,
      "step": 2115
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7269213199615479,
      "learning_rate": 1.9448523356385204e-05,
      "loss": 0.6134,
      "step": 2116
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5239670276641846,
      "learning_rate": 1.9447856065054133e-05,
      "loss": 0.6448,
      "step": 2117
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6384798288345337,
      "learning_rate": 1.9447188381714726e-05,
      "loss": 0.6282,
      "step": 2118
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6632312536239624,
      "learning_rate": 1.94465203063947e-05,
      "loss": 0.6139,
      "step": 2119
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5005584955215454,
      "learning_rate": 1.944585183912176e-05,
      "loss": 0.6192,
      "step": 2120
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5032638311386108,
      "learning_rate": 1.9445182979923657e-05,
      "loss": 0.6071,
      "step": 2121
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5562142133712769,
      "learning_rate": 1.944451372882813e-05,
      "loss": 0.5925,
      "step": 2122
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6150034666061401,
      "learning_rate": 1.9443844085862956e-05,
      "loss": 0.6263,
      "step": 2123
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6703675985336304,
      "learning_rate": 1.9443174051055913e-05,
      "loss": 0.6318,
      "step": 2124
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5116405487060547,
      "learning_rate": 1.944250362443481e-05,
      "loss": 0.5996,
      "step": 2125
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6488840579986572,
      "learning_rate": 1.9441832806027457e-05,
      "loss": 0.6359,
      "step": 2126
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5780251026153564,
      "learning_rate": 1.944116159586169e-05,
      "loss": 0.6218,
      "step": 2127
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5428133010864258,
      "learning_rate": 1.9440489993965365e-05,
      "loss": 0.5904,
      "step": 2128
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5759080648422241,
      "learning_rate": 1.9439818000366334e-05,
      "loss": 0.6521,
      "step": 2129
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5187612771987915,
      "learning_rate": 1.943914561509249e-05,
      "loss": 0.5891,
      "step": 2130
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.51810884475708,
      "learning_rate": 1.9438472838171733e-05,
      "loss": 0.6178,
      "step": 2131
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5143911838531494,
      "learning_rate": 1.943779966963197e-05,
      "loss": 0.606,
      "step": 2132
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5986415147781372,
      "learning_rate": 1.943712610950113e-05,
      "loss": 0.6209,
      "step": 2133
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5741605758666992,
      "learning_rate": 1.9436452157807173e-05,
      "loss": 0.611,
      "step": 2134
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5349584817886353,
      "learning_rate": 1.943577781457805e-05,
      "loss": 0.6619,
      "step": 2135
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5265934467315674,
      "learning_rate": 1.943510307984175e-05,
      "loss": 0.6403,
      "step": 2136
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5860872268676758,
      "learning_rate": 1.9434427953626262e-05,
      "loss": 0.6452,
      "step": 2137
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6774765253067017,
      "learning_rate": 1.94337524359596e-05,
      "loss": 0.6202,
      "step": 2138
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5022788047790527,
      "learning_rate": 1.9433076526869797e-05,
      "loss": 0.5781,
      "step": 2139
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6363409757614136,
      "learning_rate": 1.943240022638489e-05,
      "loss": 0.5832,
      "step": 2140
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6179769039154053,
      "learning_rate": 1.9431723534532944e-05,
      "loss": 0.5942,
      "step": 2141
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6077916622161865,
      "learning_rate": 1.9431046451342038e-05,
      "loss": 0.6242,
      "step": 2142
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6564122438430786,
      "learning_rate": 1.9430368976840264e-05,
      "loss": 0.6808,
      "step": 2143
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5644254684448242,
      "learning_rate": 1.942969111105573e-05,
      "loss": 0.6162,
      "step": 2144
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4485716819763184,
      "learning_rate": 1.9429012854016564e-05,
      "loss": 0.5936,
      "step": 2145
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5532244443893433,
      "learning_rate": 1.9428334205750904e-05,
      "loss": 0.58,
      "step": 2146
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7307673692703247,
      "learning_rate": 1.942765516628691e-05,
      "loss": 0.6203,
      "step": 2147
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5923959016799927,
      "learning_rate": 1.942697573565276e-05,
      "loss": 0.6136,
      "step": 2148
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5915660858154297,
      "learning_rate": 1.9426295913876644e-05,
      "loss": 0.6329,
      "step": 2149
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6290392875671387,
      "learning_rate": 1.9425615700986764e-05,
      "loss": 0.6042,
      "step": 2150
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6955839395523071,
      "learning_rate": 1.942493509701135e-05,
      "loss": 0.6713,
      "step": 2151
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6215991973876953,
      "learning_rate": 1.9424254101978637e-05,
      "loss": 0.6067,
      "step": 2152
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.545032262802124,
      "learning_rate": 1.9423572715916884e-05,
      "loss": 0.5981,
      "step": 2153
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6136852502822876,
      "learning_rate": 1.9422890938854357e-05,
      "loss": 0.6081,
      "step": 2154
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6567344665527344,
      "learning_rate": 1.9422208770819352e-05,
      "loss": 0.6682,
      "step": 2155
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.561257004737854,
      "learning_rate": 1.9421526211840163e-05,
      "loss": 0.624,
      "step": 2156
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3651517629623413,
      "learning_rate": 1.9420843261945125e-05,
      "loss": 1.0079,
      "step": 2157
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8105283975601196,
      "learning_rate": 1.942015992116256e-05,
      "loss": 0.6292,
      "step": 2158
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.754955530166626,
      "learning_rate": 1.941947618952083e-05,
      "loss": 0.6161,
      "step": 2159
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5734602212905884,
      "learning_rate": 1.9418792067048305e-05,
      "loss": 0.6297,
      "step": 2160
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.848710298538208,
      "learning_rate": 1.9418107553773365e-05,
      "loss": 0.6192,
      "step": 2161
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6480590105056763,
      "learning_rate": 1.9417422649724417e-05,
      "loss": 0.6044,
      "step": 2162
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6029777526855469,
      "learning_rate": 1.9416737354929872e-05,
      "loss": 0.5994,
      "step": 2163
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5868021249771118,
      "learning_rate": 1.941605166941817e-05,
      "loss": 0.6421,
      "step": 2164
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6081749200820923,
      "learning_rate": 1.9415365593217757e-05,
      "loss": 0.6088,
      "step": 2165
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.616589069366455,
      "learning_rate": 1.9414679126357107e-05,
      "loss": 0.6183,
      "step": 2166
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.652693271636963,
      "learning_rate": 1.9413992268864692e-05,
      "loss": 0.6222,
      "step": 2167
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6783654689788818,
      "learning_rate": 1.941330502076902e-05,
      "loss": 0.6199,
      "step": 2168
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.646398901939392,
      "learning_rate": 1.9412617382098603e-05,
      "loss": 0.6097,
      "step": 2169
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6701369285583496,
      "learning_rate": 1.9411929352881968e-05,
      "loss": 0.5849,
      "step": 2170
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5128997564315796,
      "learning_rate": 1.9411240933147674e-05,
      "loss": 0.6312,
      "step": 2171
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6064194440841675,
      "learning_rate": 1.941055212292427e-05,
      "loss": 0.6521,
      "step": 2172
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5506190061569214,
      "learning_rate": 1.9409862922240347e-05,
      "loss": 0.6439,
      "step": 2173
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5735567808151245,
      "learning_rate": 1.94091733311245e-05,
      "loss": 0.6395,
      "step": 2174
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5291481018066406,
      "learning_rate": 1.9408483349605337e-05,
      "loss": 0.6457,
      "step": 2175
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6313422918319702,
      "learning_rate": 1.9407792977711485e-05,
      "loss": 0.6394,
      "step": 2176
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5589385032653809,
      "learning_rate": 1.9407102215471597e-05,
      "loss": 0.6101,
      "step": 2177
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6827667951583862,
      "learning_rate": 1.9406411062914325e-05,
      "loss": 0.9789,
      "step": 2178
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6928924322128296,
      "learning_rate": 1.9405719520068355e-05,
      "loss": 0.5835,
      "step": 2179
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.551108479499817,
      "learning_rate": 1.9405027586962372e-05,
      "loss": 0.608,
      "step": 2180
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4540997743606567,
      "learning_rate": 1.940433526362509e-05,
      "loss": 0.6252,
      "step": 2181
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6520020961761475,
      "learning_rate": 1.9403642550085236e-05,
      "loss": 0.6154,
      "step": 2182
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4508482217788696,
      "learning_rate": 1.9402949446371548e-05,
      "loss": 0.6214,
      "step": 2183
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3815840482711792,
      "learning_rate": 1.9402255952512784e-05,
      "loss": 0.5958,
      "step": 2184
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5863791704177856,
      "learning_rate": 1.9401562068537722e-05,
      "loss": 0.6439,
      "step": 2185
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5038628578186035,
      "learning_rate": 1.940086779447515e-05,
      "loss": 0.6615,
      "step": 2186
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5322744846343994,
      "learning_rate": 1.940017313035388e-05,
      "loss": 0.6142,
      "step": 2187
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5540446043014526,
      "learning_rate": 1.9399478076202726e-05,
      "loss": 0.6029,
      "step": 2188
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5842989683151245,
      "learning_rate": 1.9398782632050532e-05,
      "loss": 0.5903,
      "step": 2189
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.440250277519226,
      "learning_rate": 1.939808679792615e-05,
      "loss": 0.6277,
      "step": 2190
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5276347398757935,
      "learning_rate": 1.939739057385846e-05,
      "loss": 0.633,
      "step": 2191
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.543866515159607,
      "learning_rate": 1.9396693959876338e-05,
      "loss": 0.6225,
      "step": 2192
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5061378479003906,
      "learning_rate": 1.9395996956008694e-05,
      "loss": 0.615,
      "step": 2193
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6614958047866821,
      "learning_rate": 1.939529956228445e-05,
      "loss": 0.6416,
      "step": 2194
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4952212572097778,
      "learning_rate": 1.9394601778732535e-05,
      "loss": 0.6278,
      "step": 2195
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6383262872695923,
      "learning_rate": 1.939390360538191e-05,
      "loss": 1.0123,
      "step": 2196
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6494656801223755,
      "learning_rate": 1.9393205042261536e-05,
      "loss": 0.6184,
      "step": 2197
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6688262224197388,
      "learning_rate": 1.93925060894004e-05,
      "loss": 0.5865,
      "step": 2198
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5248286724090576,
      "learning_rate": 1.9391806746827506e-05,
      "loss": 0.6356,
      "step": 2199
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6618144512176514,
      "learning_rate": 1.9391107014571867e-05,
      "loss": 0.6214,
      "step": 2200
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4580692052841187,
      "learning_rate": 1.9390406892662517e-05,
      "loss": 0.6203,
      "step": 2201
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6306661367416382,
      "learning_rate": 1.9389706381128507e-05,
      "loss": 0.6458,
      "step": 2202
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6408339738845825,
      "learning_rate": 1.9389005479998898e-05,
      "loss": 0.5773,
      "step": 2203
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4225049018859863,
      "learning_rate": 1.938830418930278e-05,
      "loss": 0.5928,
      "step": 2204
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7228615283966064,
      "learning_rate": 1.9387602509069243e-05,
      "loss": 0.6218,
      "step": 2205
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4734214544296265,
      "learning_rate": 1.9386900439327406e-05,
      "loss": 0.5928,
      "step": 2206
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5698354244232178,
      "learning_rate": 1.9386197980106393e-05,
      "loss": 0.6575,
      "step": 2207
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6423178911209106,
      "learning_rate": 1.9385495131435355e-05,
      "loss": 0.6692,
      "step": 2208
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5160804986953735,
      "learning_rate": 1.9384791893343454e-05,
      "loss": 0.609,
      "step": 2209
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5519816875457764,
      "learning_rate": 1.9384088265859868e-05,
      "loss": 0.6224,
      "step": 2210
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5687322616577148,
      "learning_rate": 1.9383384249013793e-05,
      "loss": 0.619,
      "step": 2211
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5204588174819946,
      "learning_rate": 1.9382679842834438e-05,
      "loss": 0.5997,
      "step": 2212
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6315104961395264,
      "learning_rate": 1.938197504735103e-05,
      "loss": 0.6243,
      "step": 2213
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.634140133857727,
      "learning_rate": 1.9381269862592816e-05,
      "loss": 0.6203,
      "step": 2214
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4194525480270386,
      "learning_rate": 1.938056428858905e-05,
      "loss": 0.5992,
      "step": 2215
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.542722225189209,
      "learning_rate": 1.9379858325369007e-05,
      "loss": 0.6056,
      "step": 2216
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5177345275878906,
      "learning_rate": 1.9379151972961984e-05,
      "loss": 0.6188,
      "step": 2217
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.470818281173706,
      "learning_rate": 1.9378445231397286e-05,
      "loss": 0.6158,
      "step": 2218
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6271642446517944,
      "learning_rate": 1.9377738100704238e-05,
      "loss": 0.6434,
      "step": 2219
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5824729204177856,
      "learning_rate": 1.9377030580912183e-05,
      "loss": 0.6137,
      "step": 2220
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5100091695785522,
      "learning_rate": 1.9376322672050468e-05,
      "loss": 0.5993,
      "step": 2221
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5890158414840698,
      "learning_rate": 1.9375614374148476e-05,
      "loss": 0.5954,
      "step": 2222
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7972471714019775,
      "learning_rate": 1.9374905687235584e-05,
      "loss": 0.6521,
      "step": 2223
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5683163404464722,
      "learning_rate": 1.9374196611341212e-05,
      "loss": 0.6231,
      "step": 2224
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5518593788146973,
      "learning_rate": 1.9373487146494766e-05,
      "loss": 0.6173,
      "step": 2225
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3829302787780762,
      "learning_rate": 1.9372777292725692e-05,
      "loss": 0.5871,
      "step": 2226
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5352575778961182,
      "learning_rate": 1.937206705006344e-05,
      "loss": 0.6431,
      "step": 2227
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4213719367980957,
      "learning_rate": 1.9371356418537477e-05,
      "loss": 0.6196,
      "step": 2228
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4865413904190063,
      "learning_rate": 1.9370645398177294e-05,
      "loss": 0.6355,
      "step": 2229
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5139191150665283,
      "learning_rate": 1.9369933989012386e-05,
      "loss": 0.6178,
      "step": 2230
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8763431310653687,
      "learning_rate": 1.9369222191072274e-05,
      "loss": 1.0134,
      "step": 2231
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6679569482803345,
      "learning_rate": 1.9368510004386492e-05,
      "loss": 0.6032,
      "step": 2232
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5963534116744995,
      "learning_rate": 1.9367797428984593e-05,
      "loss": 0.6227,
      "step": 2233
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5579580068588257,
      "learning_rate": 1.9367084464896136e-05,
      "loss": 0.6064,
      "step": 2234
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4785785675048828,
      "learning_rate": 1.936637111215071e-05,
      "loss": 0.6281,
      "step": 2235
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6118611097335815,
      "learning_rate": 1.9365657370777908e-05,
      "loss": 0.6063,
      "step": 2236
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.645931601524353,
      "learning_rate": 1.9364943240807346e-05,
      "loss": 0.6285,
      "step": 2237
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4865622520446777,
      "learning_rate": 1.9364228722268653e-05,
      "loss": 0.5719,
      "step": 2238
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5945409536361694,
      "learning_rate": 1.9363513815191482e-05,
      "loss": 0.6113,
      "step": 2239
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5513724088668823,
      "learning_rate": 1.936279851960549e-05,
      "loss": 0.6007,
      "step": 2240
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5164145231246948,
      "learning_rate": 1.9362082835540353e-05,
      "loss": 0.6066,
      "step": 2241
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7621350288391113,
      "learning_rate": 1.9361366763025776e-05,
      "loss": 0.6073,
      "step": 2242
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5641072988510132,
      "learning_rate": 1.936065030209146e-05,
      "loss": 0.6318,
      "step": 2243
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6076730489730835,
      "learning_rate": 1.9359933452767138e-05,
      "loss": 0.6409,
      "step": 2244
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.655665636062622,
      "learning_rate": 1.9359216215082556e-05,
      "loss": 0.6685,
      "step": 2245
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6272333860397339,
      "learning_rate": 1.9358498589067466e-05,
      "loss": 0.585,
      "step": 2246
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5595942735671997,
      "learning_rate": 1.9357780574751646e-05,
      "loss": 0.6051,
      "step": 2247
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4784764051437378,
      "learning_rate": 1.935706217216489e-05,
      "loss": 0.6099,
      "step": 2248
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4675875902175903,
      "learning_rate": 1.9356343381337e-05,
      "loss": 0.6086,
      "step": 2249
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5277502536773682,
      "learning_rate": 1.935562420229781e-05,
      "loss": 0.6042,
      "step": 2250
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5350748300552368,
      "learning_rate": 1.9354904635077153e-05,
      "loss": 0.6782,
      "step": 2251
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5736196041107178,
      "learning_rate": 1.9354184679704887e-05,
      "loss": 0.6456,
      "step": 2252
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4396493434906006,
      "learning_rate": 1.9353464336210885e-05,
      "loss": 0.5841,
      "step": 2253
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.612680435180664,
      "learning_rate": 1.935274360462503e-05,
      "loss": 0.633,
      "step": 2254
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5033316612243652,
      "learning_rate": 1.9352022484977232e-05,
      "loss": 0.6852,
      "step": 2255
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.508664608001709,
      "learning_rate": 1.9351300977297415e-05,
      "loss": 0.5626,
      "step": 2256
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5332050323486328,
      "learning_rate": 1.9350579081615504e-05,
      "loss": 0.5771,
      "step": 2257
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4660749435424805,
      "learning_rate": 1.9349856797961458e-05,
      "loss": 0.589,
      "step": 2258
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5424751043319702,
      "learning_rate": 1.934913412636525e-05,
      "loss": 0.6588,
      "step": 2259
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5768108367919922,
      "learning_rate": 1.9348411066856862e-05,
      "loss": 0.656,
      "step": 2260
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7739640474319458,
      "learning_rate": 1.934768761946629e-05,
      "loss": 0.6178,
      "step": 2261
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.554707646369934,
      "learning_rate": 1.934696378422356e-05,
      "loss": 0.6259,
      "step": 2262
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5784962177276611,
      "learning_rate": 1.93462395611587e-05,
      "loss": 0.6444,
      "step": 2263
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.521371603012085,
      "learning_rate": 1.9345514950301755e-05,
      "loss": 0.6054,
      "step": 2264
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5022733211517334,
      "learning_rate": 1.93447899516828e-05,
      "loss": 0.5772,
      "step": 2265
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.586758017539978,
      "learning_rate": 1.9344064565331908e-05,
      "loss": 0.6527,
      "step": 2266
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5266096591949463,
      "learning_rate": 1.934333879127918e-05,
      "loss": 0.6404,
      "step": 2267
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6026909351348877,
      "learning_rate": 1.9342612629554734e-05,
      "loss": 0.6028,
      "step": 2268
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.577659010887146,
      "learning_rate": 1.9341886080188694e-05,
      "loss": 0.622,
      "step": 2269
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5254322290420532,
      "learning_rate": 1.934115914321121e-05,
      "loss": 0.5866,
      "step": 2270
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5260272026062012,
      "learning_rate": 1.9340431818652438e-05,
      "loss": 0.5684,
      "step": 2271
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5084657669067383,
      "learning_rate": 1.9339704106542557e-05,
      "loss": 0.6362,
      "step": 2272
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5420022010803223,
      "learning_rate": 1.933897600691177e-05,
      "loss": 0.6074,
      "step": 2273
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6056326627731323,
      "learning_rate": 1.9338247519790275e-05,
      "loss": 0.667,
      "step": 2274
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.505638837814331,
      "learning_rate": 1.9337518645208308e-05,
      "loss": 0.626,
      "step": 2275
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6409956216812134,
      "learning_rate": 1.93367893831961e-05,
      "loss": 0.6116,
      "step": 2276
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5418176651000977,
      "learning_rate": 1.9336059733783925e-05,
      "loss": 0.6043,
      "step": 2277
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7287936210632324,
      "learning_rate": 1.9335329697002044e-05,
      "loss": 1.0202,
      "step": 2278
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.878553032875061,
      "learning_rate": 1.9334599272880754e-05,
      "loss": 0.6237,
      "step": 2279
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6721268892288208,
      "learning_rate": 1.933386846145036e-05,
      "loss": 0.6358,
      "step": 2280
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5115876197814941,
      "learning_rate": 1.9333137262741185e-05,
      "loss": 0.6464,
      "step": 2281
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.222598671913147,
      "learning_rate": 1.9332405676783565e-05,
      "loss": 0.975,
      "step": 2282
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.5545737743377686,
      "learning_rate": 1.933167370360786e-05,
      "loss": 0.6114,
      "step": 2283
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0481088161468506,
      "learning_rate": 1.933094134324444e-05,
      "loss": 0.6342,
      "step": 2284
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5914674997329712,
      "learning_rate": 1.9330208595723685e-05,
      "loss": 0.6168,
      "step": 2285
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7007449865341187,
      "learning_rate": 1.9329475461076005e-05,
      "loss": 0.599,
      "step": 2286
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9570701122283936,
      "learning_rate": 1.932874193933182e-05,
      "loss": 0.6143,
      "step": 2287
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.715856909751892,
      "learning_rate": 1.932800803052156e-05,
      "loss": 0.6268,
      "step": 2288
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6527570486068726,
      "learning_rate": 1.9327273734675677e-05,
      "loss": 0.6119,
      "step": 2289
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6993968486785889,
      "learning_rate": 1.932653905182464e-05,
      "loss": 0.6111,
      "step": 2290
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.626943826675415,
      "learning_rate": 1.9325803981998935e-05,
      "loss": 0.6159,
      "step": 2291
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5153734683990479,
      "learning_rate": 1.9325068525229055e-05,
      "loss": 0.6025,
      "step": 2292
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5357203483581543,
      "learning_rate": 1.932433268154552e-05,
      "loss": 0.6122,
      "step": 2293
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5445382595062256,
      "learning_rate": 1.932359645097886e-05,
      "loss": 0.6032,
      "step": 2294
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5544275045394897,
      "learning_rate": 1.9322859833559623e-05,
      "loss": 0.6256,
      "step": 2295
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4468469619750977,
      "learning_rate": 1.932212282931837e-05,
      "loss": 0.6067,
      "step": 2296
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4992117881774902,
      "learning_rate": 1.9321385438285686e-05,
      "loss": 0.6106,
      "step": 2297
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5735325813293457,
      "learning_rate": 1.9320647660492163e-05,
      "loss": 0.5852,
      "step": 2298
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5654598474502563,
      "learning_rate": 1.931990949596841e-05,
      "loss": 0.5887,
      "step": 2299
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.511053442955017,
      "learning_rate": 1.931917094474506e-05,
      "loss": 0.6254,
      "step": 2300
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5191912651062012,
      "learning_rate": 1.9318432006852756e-05,
      "loss": 0.6269,
      "step": 2301
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6623413562774658,
      "learning_rate": 1.9317692682322157e-05,
      "loss": 0.6418,
      "step": 2302
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4839829206466675,
      "learning_rate": 1.9316952971183935e-05,
      "loss": 0.6535,
      "step": 2303
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.62028968334198,
      "learning_rate": 1.931621287346879e-05,
      "loss": 0.6199,
      "step": 2304
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6613247394561768,
      "learning_rate": 1.9315472389207423e-05,
      "loss": 0.649,
      "step": 2305
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5483592748641968,
      "learning_rate": 1.9314731518430562e-05,
      "loss": 0.5899,
      "step": 2306
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.644867181777954,
      "learning_rate": 1.9313990261168946e-05,
      "loss": 0.5902,
      "step": 2307
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6282953023910522,
      "learning_rate": 1.9313248617453325e-05,
      "loss": 0.6096,
      "step": 2308
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4217380285263062,
      "learning_rate": 1.931250658731448e-05,
      "loss": 0.5883,
      "step": 2309
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.548506259918213,
      "learning_rate": 1.93117641707832e-05,
      "loss": 0.6072,
      "step": 2310
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5058172941207886,
      "learning_rate": 1.9311021367890278e-05,
      "loss": 0.6185,
      "step": 2311
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4909741878509521,
      "learning_rate": 1.9310278178666544e-05,
      "loss": 0.5941,
      "step": 2312
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.446386456489563,
      "learning_rate": 1.9309534603142836e-05,
      "loss": 0.5966,
      "step": 2313
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4006600379943848,
      "learning_rate": 1.9308790641349996e-05,
      "loss": 0.5952,
      "step": 2314
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.512519359588623,
      "learning_rate": 1.93080462933189e-05,
      "loss": 0.5928,
      "step": 2315
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4361430406570435,
      "learning_rate": 1.9307301559080428e-05,
      "loss": 0.6291,
      "step": 2316
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.50221586227417,
      "learning_rate": 1.9306556438665484e-05,
      "loss": 0.6341,
      "step": 2317
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4866808652877808,
      "learning_rate": 1.9305810932104984e-05,
      "loss": 0.5826,
      "step": 2318
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4415451288223267,
      "learning_rate": 1.930506503942986e-05,
      "loss": 0.6258,
      "step": 2319
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5443899631500244,
      "learning_rate": 1.930431876067106e-05,
      "loss": 0.6752,
      "step": 2320
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5233243703842163,
      "learning_rate": 1.9303572095859545e-05,
      "loss": 0.5995,
      "step": 2321
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4650784730911255,
      "learning_rate": 1.9302825045026302e-05,
      "loss": 0.6193,
      "step": 2322
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6113539934158325,
      "learning_rate": 1.9302077608202324e-05,
      "loss": 0.6154,
      "step": 2323
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6272902488708496,
      "learning_rate": 1.9301329785418622e-05,
      "loss": 0.6728,
      "step": 2324
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5646287202835083,
      "learning_rate": 1.930058157670623e-05,
      "loss": 0.6278,
      "step": 2325
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5378273725509644,
      "learning_rate": 1.9299832982096185e-05,
      "loss": 0.6046,
      "step": 2326
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5639182329177856,
      "learning_rate": 1.929908400161955e-05,
      "loss": 0.6095,
      "step": 2327
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.137174367904663,
      "learning_rate": 1.9298334635307407e-05,
      "loss": 0.9962,
      "step": 2328
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6466858386993408,
      "learning_rate": 1.929758488319084e-05,
      "loss": 0.6165,
      "step": 2329
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6564464569091797,
      "learning_rate": 1.9296834745300964e-05,
      "loss": 0.6095,
      "step": 2330
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5005992650985718,
      "learning_rate": 1.92960842216689e-05,
      "loss": 0.6314,
      "step": 2331
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5830085277557373,
      "learning_rate": 1.9295333312325796e-05,
      "loss": 0.6395,
      "step": 2332
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5341683626174927,
      "learning_rate": 1.9294582017302797e-05,
      "loss": 0.6487,
      "step": 2333
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5539835691452026,
      "learning_rate": 1.929383033663108e-05,
      "loss": 0.6265,
      "step": 2334
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5698232650756836,
      "learning_rate": 1.9293078270341837e-05,
      "loss": 0.5955,
      "step": 2335
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6210358142852783,
      "learning_rate": 1.929232581846627e-05,
      "loss": 0.6371,
      "step": 2336
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.435483455657959,
      "learning_rate": 1.92915729810356e-05,
      "loss": 0.6236,
      "step": 2337
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5136202573776245,
      "learning_rate": 1.929081975808106e-05,
      "loss": 0.583,
      "step": 2338
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6157639026641846,
      "learning_rate": 1.929006614963391e-05,
      "loss": 0.594,
      "step": 2339
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4422053098678589,
      "learning_rate": 1.9289312155725413e-05,
      "loss": 0.5988,
      "step": 2340
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.53079354763031,
      "learning_rate": 1.9288557776386855e-05,
      "loss": 0.6314,
      "step": 2341
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5235731601715088,
      "learning_rate": 1.928780301164954e-05,
      "loss": 0.6294,
      "step": 2342
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6126002073287964,
      "learning_rate": 1.9287047861544775e-05,
      "loss": 0.6435,
      "step": 2343
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5235556364059448,
      "learning_rate": 1.92862923261039e-05,
      "loss": 0.6299,
      "step": 2344
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4561413526535034,
      "learning_rate": 1.928553640535826e-05,
      "loss": 0.6064,
      "step": 2345
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.54944908618927,
      "learning_rate": 1.928478009933923e-05,
      "loss": 0.6141,
      "step": 2346
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5053770542144775,
      "learning_rate": 1.9284023408078173e-05,
      "loss": 0.6374,
      "step": 2347
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.636190414428711,
      "learning_rate": 1.92832663316065e-05,
      "loss": 0.6028,
      "step": 2348
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4996225833892822,
      "learning_rate": 1.928250886995562e-05,
      "loss": 0.609,
      "step": 2349
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.514270544052124,
      "learning_rate": 1.928175102315695e-05,
      "loss": 0.6449,
      "step": 2350
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.531002163887024,
      "learning_rate": 1.9280992791241952e-05,
      "loss": 0.6214,
      "step": 2351
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3543428182601929,
      "learning_rate": 1.9280234174242075e-05,
      "loss": 0.614,
      "step": 2352
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.382015347480774,
      "learning_rate": 1.92794751721888e-05,
      "loss": 0.6171,
      "step": 2353
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4326292276382446,
      "learning_rate": 1.927871578511362e-05,
      "loss": 0.5761,
      "step": 2354
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4400824308395386,
      "learning_rate": 1.9277956013048035e-05,
      "loss": 0.5686,
      "step": 2355
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3997737169265747,
      "learning_rate": 1.927719585602358e-05,
      "loss": 0.6165,
      "step": 2356
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4867438077926636,
      "learning_rate": 1.927643531407179e-05,
      "loss": 0.617,
      "step": 2357
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5183124542236328,
      "learning_rate": 1.9275674387224223e-05,
      "loss": 0.6244,
      "step": 2358
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4387589693069458,
      "learning_rate": 1.9274913075512448e-05,
      "loss": 0.6071,
      "step": 2359
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5167685747146606,
      "learning_rate": 1.9274151378968055e-05,
      "loss": 0.5811,
      "step": 2360
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5264452695846558,
      "learning_rate": 1.9273389297622652e-05,
      "loss": 0.5824,
      "step": 2361
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4390652179718018,
      "learning_rate": 1.9272626831507853e-05,
      "loss": 0.6218,
      "step": 2362
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5996726751327515,
      "learning_rate": 1.9271863980655298e-05,
      "loss": 0.577,
      "step": 2363
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5152233839035034,
      "learning_rate": 1.9271100745096634e-05,
      "loss": 0.635,
      "step": 2364
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4847359657287598,
      "learning_rate": 1.9270337124863538e-05,
      "loss": 0.6285,
      "step": 2365
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5146299600601196,
      "learning_rate": 1.9269573119987684e-05,
      "loss": 0.611,
      "step": 2366
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4672861099243164,
      "learning_rate": 1.9268808730500783e-05,
      "loss": 0.5981,
      "step": 2367
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5290496349334717,
      "learning_rate": 1.926804395643454e-05,
      "loss": 0.6179,
      "step": 2368
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4367902278900146,
      "learning_rate": 1.9267278797820693e-05,
      "loss": 0.5396,
      "step": 2369
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4417619705200195,
      "learning_rate": 1.9266513254690988e-05,
      "loss": 0.602,
      "step": 2370
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5394443273544312,
      "learning_rate": 1.9265747327077188e-05,
      "loss": 0.6172,
      "step": 2371
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5977145433425903,
      "learning_rate": 1.926498101501107e-05,
      "loss": 0.6364,
      "step": 2372
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6448732614517212,
      "learning_rate": 1.9264214318524435e-05,
      "loss": 0.6005,
      "step": 2373
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4894055128097534,
      "learning_rate": 1.9263447237649098e-05,
      "loss": 0.6193,
      "step": 2374
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5381349325180054,
      "learning_rate": 1.926267977241688e-05,
      "loss": 0.5883,
      "step": 2375
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4437185525894165,
      "learning_rate": 1.926191192285962e-05,
      "loss": 0.664,
      "step": 2376
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5775606632232666,
      "learning_rate": 1.9261143689009188e-05,
      "loss": 0.6112,
      "step": 2377
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.668076515197754,
      "learning_rate": 1.9260375070897455e-05,
      "loss": 0.6288,
      "step": 2378
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4243144989013672,
      "learning_rate": 1.925960606855631e-05,
      "loss": 0.6514,
      "step": 2379
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4015483856201172,
      "learning_rate": 1.9258836682017662e-05,
      "loss": 0.6039,
      "step": 2380
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.50315260887146,
      "learning_rate": 1.925806691131344e-05,
      "loss": 0.615,
      "step": 2381
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4616831541061401,
      "learning_rate": 1.9257296756475573e-05,
      "loss": 0.5953,
      "step": 2382
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4014395475387573,
      "learning_rate": 1.925652621753602e-05,
      "loss": 0.6051,
      "step": 2383
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6092764139175415,
      "learning_rate": 1.9255755294526756e-05,
      "loss": 0.6017,
      "step": 2384
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.549288272857666,
      "learning_rate": 1.925498398747976e-05,
      "loss": 0.6065,
      "step": 2385
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4869357347488403,
      "learning_rate": 1.9254212296427043e-05,
      "loss": 0.6058,
      "step": 2386
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.377463459968567,
      "learning_rate": 1.925344022140062e-05,
      "loss": 0.5832,
      "step": 2387
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.658864140510559,
      "learning_rate": 1.9252667762432526e-05,
      "loss": 0.6025,
      "step": 2388
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5924116373062134,
      "learning_rate": 1.9251894919554815e-05,
      "loss": 0.6309,
      "step": 2389
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.387746810913086,
      "learning_rate": 1.9251121692799547e-05,
      "loss": 0.6148,
      "step": 2390
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4913251399993896,
      "learning_rate": 1.9250348082198807e-05,
      "loss": 0.5996,
      "step": 2391
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5461797714233398,
      "learning_rate": 1.9249574087784697e-05,
      "loss": 0.6344,
      "step": 2392
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4351003170013428,
      "learning_rate": 1.9248799709589327e-05,
      "loss": 0.579,
      "step": 2393
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4702694416046143,
      "learning_rate": 1.9248024947644834e-05,
      "loss": 0.596,
      "step": 2394
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4517909288406372,
      "learning_rate": 1.9247249801983353e-05,
      "loss": 0.6293,
      "step": 2395
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4971274137496948,
      "learning_rate": 1.9246474272637058e-05,
      "loss": 0.5945,
      "step": 2396
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5007768869400024,
      "learning_rate": 1.924569835963812e-05,
      "loss": 0.5926,
      "step": 2397
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4077399969100952,
      "learning_rate": 1.9244922063018732e-05,
      "loss": 0.6352,
      "step": 2398
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.44891357421875,
      "learning_rate": 1.9244145382811113e-05,
      "loss": 0.6106,
      "step": 2399
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.560275912284851,
      "learning_rate": 1.9243368319047477e-05,
      "loss": 0.6115,
      "step": 2400
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5291625261306763,
      "learning_rate": 1.9242590871760076e-05,
      "loss": 0.5806,
      "step": 2401
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5004905462265015,
      "learning_rate": 1.924181304098116e-05,
      "loss": 0.6055,
      "step": 2402
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5817822217941284,
      "learning_rate": 1.9241034826743005e-05,
      "loss": 0.6002,
      "step": 2403
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.433213710784912,
      "learning_rate": 1.92402562290779e-05,
      "loss": 0.6285,
      "step": 2404
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5149391889572144,
      "learning_rate": 1.9239477248018157e-05,
      "loss": 0.6207,
      "step": 2405
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4456713199615479,
      "learning_rate": 1.9238697883596088e-05,
      "loss": 0.5972,
      "step": 2406
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4363977909088135,
      "learning_rate": 1.9237918135844037e-05,
      "loss": 0.6253,
      "step": 2407
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4811303615570068,
      "learning_rate": 1.9237138004794352e-05,
      "loss": 0.5899,
      "step": 2408
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5537971258163452,
      "learning_rate": 1.9236357490479402e-05,
      "loss": 0.5993,
      "step": 2409
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5398772954940796,
      "learning_rate": 1.9235576592931578e-05,
      "loss": 0.6053,
      "step": 2410
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4864085912704468,
      "learning_rate": 1.9234795312183273e-05,
      "loss": 0.5928,
      "step": 2411
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5874605178833008,
      "learning_rate": 1.923401364826691e-05,
      "loss": 0.626,
      "step": 2412
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4508129358291626,
      "learning_rate": 1.923323160121492e-05,
      "loss": 0.6295,
      "step": 2413
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7906601428985596,
      "learning_rate": 1.9232449171059747e-05,
      "loss": 0.6238,
      "step": 2414
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4685333967208862,
      "learning_rate": 1.9231666357833864e-05,
      "loss": 0.6184,
      "step": 2415
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3887059688568115,
      "learning_rate": 1.9230883161569745e-05,
      "loss": 0.5674,
      "step": 2416
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4386651515960693,
      "learning_rate": 1.923009958229989e-05,
      "loss": 0.593,
      "step": 2417
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4502928256988525,
      "learning_rate": 1.9229315620056805e-05,
      "loss": 0.6253,
      "step": 2418
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5585049390792847,
      "learning_rate": 1.9228531274873022e-05,
      "loss": 0.6104,
      "step": 2419
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4635206460952759,
      "learning_rate": 1.9227746546781087e-05,
      "loss": 0.5895,
      "step": 2420
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6700050830841064,
      "learning_rate": 1.9226961435813556e-05,
      "loss": 0.576,
      "step": 2421
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4668300151824951,
      "learning_rate": 1.9226175942003004e-05,
      "loss": 0.5909,
      "step": 2422
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.455237627029419,
      "learning_rate": 1.9225390065382026e-05,
      "loss": 0.5835,
      "step": 2423
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4211233854293823,
      "learning_rate": 1.922460380598323e-05,
      "loss": 0.6172,
      "step": 2424
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.505486011505127,
      "learning_rate": 1.9223817163839235e-05,
      "loss": 0.6115,
      "step": 2425
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4460523128509521,
      "learning_rate": 1.9223030138982685e-05,
      "loss": 0.6208,
      "step": 2426
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4291839599609375,
      "learning_rate": 1.922224273144623e-05,
      "loss": 0.6252,
      "step": 2427
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6062971353530884,
      "learning_rate": 1.9221454941262547e-05,
      "loss": 0.625,
      "step": 2428
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.477141261100769,
      "learning_rate": 1.9220666768464317e-05,
      "loss": 0.5984,
      "step": 2429
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4585373401641846,
      "learning_rate": 1.9219878213084245e-05,
      "loss": 0.6208,
      "step": 2430
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5565303564071655,
      "learning_rate": 1.9219089275155048e-05,
      "loss": 0.6216,
      "step": 2431
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5469727516174316,
      "learning_rate": 1.9218299954709465e-05,
      "loss": 0.6513,
      "step": 2432
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4789350032806396,
      "learning_rate": 1.9217510251780243e-05,
      "loss": 0.6419,
      "step": 2433
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6316391229629517,
      "learning_rate": 1.9216720166400148e-05,
      "loss": 0.6006,
      "step": 2434
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4642629623413086,
      "learning_rate": 1.9215929698601963e-05,
      "loss": 0.6256,
      "step": 2435
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5017132759094238,
      "learning_rate": 1.9215138848418485e-05,
      "loss": 0.5787,
      "step": 2436
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6344603300094604,
      "learning_rate": 1.9214347615882527e-05,
      "loss": 0.5982,
      "step": 2437
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5339826345443726,
      "learning_rate": 1.921355600102692e-05,
      "loss": 0.6095,
      "step": 2438
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4364025592803955,
      "learning_rate": 1.921276400388451e-05,
      "loss": 0.5829,
      "step": 2439
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4845138788223267,
      "learning_rate": 1.9211971624488158e-05,
      "loss": 0.594,
      "step": 2440
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5460282564163208,
      "learning_rate": 1.9211178862870742e-05,
      "loss": 0.5938,
      "step": 2441
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4408010244369507,
      "learning_rate": 1.9210385719065155e-05,
      "loss": 0.631,
      "step": 2442
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4818164110183716,
      "learning_rate": 1.9209592193104303e-05,
      "loss": 0.6318,
      "step": 2443
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.493893027305603,
      "learning_rate": 1.9208798285021114e-05,
      "loss": 0.6542,
      "step": 2444
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.536350131034851,
      "learning_rate": 1.9208003994848522e-05,
      "loss": 0.6187,
      "step": 2445
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.511641263961792,
      "learning_rate": 1.9207209322619494e-05,
      "loss": 0.5977,
      "step": 2446
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5253201723098755,
      "learning_rate": 1.9206414268367e-05,
      "loss": 0.6311,
      "step": 2447
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6009182929992676,
      "learning_rate": 1.9205618832124023e-05,
      "loss": 0.6313,
      "step": 2448
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9894636869430542,
      "learning_rate": 1.9204823013923567e-05,
      "loss": 0.6101,
      "step": 2449
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6600593328475952,
      "learning_rate": 1.9204026813798656e-05,
      "loss": 0.5936,
      "step": 2450
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.437001347541809,
      "learning_rate": 1.9203230231782328e-05,
      "loss": 0.6055,
      "step": 2451
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7028619050979614,
      "learning_rate": 1.9202433267907625e-05,
      "loss": 0.5943,
      "step": 2452
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6604769229888916,
      "learning_rate": 1.9201635922207624e-05,
      "loss": 0.6461,
      "step": 2453
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3871712684631348,
      "learning_rate": 1.9200838194715407e-05,
      "loss": 0.5741,
      "step": 2454
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5842937231063843,
      "learning_rate": 1.9200040085464064e-05,
      "loss": 0.5939,
      "step": 2455
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5590649843215942,
      "learning_rate": 1.9199241594486723e-05,
      "loss": 0.5875,
      "step": 2456
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4358222484588623,
      "learning_rate": 1.9198442721816504e-05,
      "loss": 0.6753,
      "step": 2457
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6076921224594116,
      "learning_rate": 1.919764346748656e-05,
      "loss": 0.605,
      "step": 2458
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4587904214859009,
      "learning_rate": 1.9196843831530052e-05,
      "loss": 0.587,
      "step": 2459
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.546059012413025,
      "learning_rate": 1.9196043813980158e-05,
      "loss": 0.6314,
      "step": 2460
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5131186246871948,
      "learning_rate": 1.9195243414870073e-05,
      "loss": 0.6518,
      "step": 2461
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5238817930221558,
      "learning_rate": 1.9194442634233e-05,
      "loss": 0.6152,
      "step": 2462
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4357571601867676,
      "learning_rate": 1.919364147210218e-05,
      "loss": 0.6055,
      "step": 2463
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5753512382507324,
      "learning_rate": 1.919283992851084e-05,
      "loss": 0.5964,
      "step": 2464
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1799981594085693,
      "learning_rate": 1.9192038003492245e-05,
      "loss": 0.5961,
      "step": 2465
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5774986743927002,
      "learning_rate": 1.9191235697079668e-05,
      "loss": 0.6096,
      "step": 2466
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4523048400878906,
      "learning_rate": 1.9190433009306394e-05,
      "loss": 0.5889,
      "step": 2467
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5807461738586426,
      "learning_rate": 1.918962994020573e-05,
      "loss": 0.5793,
      "step": 2468
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5685153007507324,
      "learning_rate": 1.9188826489810997e-05,
      "loss": 0.5926,
      "step": 2469
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.524795651435852,
      "learning_rate": 1.918802265815553e-05,
      "loss": 0.6109,
      "step": 2470
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5360502004623413,
      "learning_rate": 1.918721844527269e-05,
      "loss": 0.6315,
      "step": 2471
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5961655378341675,
      "learning_rate": 1.918641385119583e-05,
      "loss": 0.6275,
      "step": 2472
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.553107500076294,
      "learning_rate": 1.918560887595835e-05,
      "loss": 0.6256,
      "step": 2473
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4553871154785156,
      "learning_rate": 1.9184803519593637e-05,
      "loss": 0.6037,
      "step": 2474
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5381039381027222,
      "learning_rate": 1.9183997782135116e-05,
      "loss": 0.5875,
      "step": 2475
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4344239234924316,
      "learning_rate": 1.918319166361621e-05,
      "loss": 0.5629,
      "step": 2476
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4830608367919922,
      "learning_rate": 1.9182385164070374e-05,
      "loss": 0.6047,
      "step": 2477
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4425030946731567,
      "learning_rate": 1.9181578283531068e-05,
      "loss": 0.5939,
      "step": 2478
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.444130539894104,
      "learning_rate": 1.918077102203177e-05,
      "loss": 0.6192,
      "step": 2479
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4634510278701782,
      "learning_rate": 1.9179963379605975e-05,
      "loss": 0.5876,
      "step": 2480
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6589927673339844,
      "learning_rate": 1.9179155356287192e-05,
      "loss": 0.5696,
      "step": 2481
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4597342014312744,
      "learning_rate": 1.917834695210895e-05,
      "loss": 0.5894,
      "step": 2482
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4754056930541992,
      "learning_rate": 1.9177538167104792e-05,
      "loss": 0.6616,
      "step": 2483
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.480018138885498,
      "learning_rate": 1.9176729001308274e-05,
      "loss": 0.5992,
      "step": 2484
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3797298669815063,
      "learning_rate": 1.917591945475297e-05,
      "loss": 0.9706,
      "step": 2485
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5655217170715332,
      "learning_rate": 1.917510952747247e-05,
      "loss": 0.6021,
      "step": 2486
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.574790358543396,
      "learning_rate": 1.917429921950038e-05,
      "loss": 0.5611,
      "step": 2487
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5534396171569824,
      "learning_rate": 1.9173488530870316e-05,
      "loss": 0.5721,
      "step": 2488
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5364750623703003,
      "learning_rate": 1.917267746161592e-05,
      "loss": 0.6545,
      "step": 2489
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4561930894851685,
      "learning_rate": 1.9171866011770846e-05,
      "loss": 0.5664,
      "step": 2490
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4378368854522705,
      "learning_rate": 1.917105418136876e-05,
      "loss": 0.5845,
      "step": 2491
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.431589126586914,
      "learning_rate": 1.9170241970443344e-05,
      "loss": 0.6075,
      "step": 2492
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5323646068572998,
      "learning_rate": 1.91694293790283e-05,
      "loss": 0.657,
      "step": 2493
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5532639026641846,
      "learning_rate": 1.916861640715735e-05,
      "loss": 0.5952,
      "step": 2494
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.42643404006958,
      "learning_rate": 1.9167803054864212e-05,
      "loss": 0.6207,
      "step": 2495
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4506981372833252,
      "learning_rate": 1.9166989322182647e-05,
      "loss": 0.5691,
      "step": 2496
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.438138723373413,
      "learning_rate": 1.9166175209146412e-05,
      "loss": 0.6197,
      "step": 2497
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.537619948387146,
      "learning_rate": 1.9165360715789287e-05,
      "loss": 0.6421,
      "step": 2498
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5895620584487915,
      "learning_rate": 1.916454584214506e-05,
      "loss": 0.5579,
      "step": 2499
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.527675747871399,
      "learning_rate": 1.9163730588247556e-05,
      "loss": 0.592,
      "step": 2500
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4131884574890137,
      "learning_rate": 1.9162914954130588e-05,
      "loss": 0.5997,
      "step": 2501
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5239412784576416,
      "learning_rate": 1.9162098939828007e-05,
      "loss": 0.5977,
      "step": 2502
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.583330750465393,
      "learning_rate": 1.9161282545373666e-05,
      "loss": 0.6285,
      "step": 2503
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.53241765499115,
      "learning_rate": 1.916046577080144e-05,
      "loss": 1.0536,
      "step": 2504
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5948034524917603,
      "learning_rate": 1.915964861614522e-05,
      "loss": 0.5966,
      "step": 2505
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4881783723831177,
      "learning_rate": 1.9158831081438906e-05,
      "loss": 0.6302,
      "step": 2506
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.37205970287323,
      "learning_rate": 1.9158013166716424e-05,
      "loss": 0.5703,
      "step": 2507
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4493861198425293,
      "learning_rate": 1.915719487201171e-05,
      "loss": 0.5815,
      "step": 2508
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5756971836090088,
      "learning_rate": 1.9156376197358713e-05,
      "loss": 0.5752,
      "step": 2509
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5089221000671387,
      "learning_rate": 1.9155557142791407e-05,
      "loss": 0.6089,
      "step": 2510
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.510221242904663,
      "learning_rate": 1.9154737708343773e-05,
      "loss": 0.6355,
      "step": 2511
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4490450620651245,
      "learning_rate": 1.915391789404981e-05,
      "loss": 0.6005,
      "step": 2512
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5375386476516724,
      "learning_rate": 1.9153097699943534e-05,
      "loss": 0.5867,
      "step": 2513
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5604270696640015,
      "learning_rate": 1.9152277126058977e-05,
      "loss": 0.5961,
      "step": 2514
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5631698369979858,
      "learning_rate": 1.9151456172430186e-05,
      "loss": 0.6135,
      "step": 2515
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5395547151565552,
      "learning_rate": 1.9150634839091223e-05,
      "loss": 0.5635,
      "step": 2516
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4494221210479736,
      "learning_rate": 1.9149813126076168e-05,
      "loss": 0.5794,
      "step": 2517
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4101461172103882,
      "learning_rate": 1.9148991033419114e-05,
      "loss": 0.5828,
      "step": 2518
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4944210052490234,
      "learning_rate": 1.914816856115417e-05,
      "loss": 0.5962,
      "step": 2519
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5521448850631714,
      "learning_rate": 1.9147345709315468e-05,
      "loss": 0.6156,
      "step": 2520
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5558743476867676,
      "learning_rate": 1.914652247793714e-05,
      "loss": 0.6244,
      "step": 2521
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.566007137298584,
      "learning_rate": 1.9145698867053353e-05,
      "loss": 0.622,
      "step": 2522
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.423469066619873,
      "learning_rate": 1.9144874876698273e-05,
      "loss": 0.5932,
      "step": 2523
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4811160564422607,
      "learning_rate": 1.914405050690609e-05,
      "loss": 0.637,
      "step": 2524
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4733526706695557,
      "learning_rate": 1.914322575771101e-05,
      "loss": 0.5865,
      "step": 2525
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.510998010635376,
      "learning_rate": 1.9142400629147255e-05,
      "loss": 0.6156,
      "step": 2526
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5121806859970093,
      "learning_rate": 1.914157512124906e-05,
      "loss": 0.5755,
      "step": 2527
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4851166009902954,
      "learning_rate": 1.9140749234050672e-05,
      "loss": 0.6346,
      "step": 2528
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6343460083007812,
      "learning_rate": 1.9139922967586366e-05,
      "loss": 0.6233,
      "step": 2529
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6133333444595337,
      "learning_rate": 1.913909632189042e-05,
      "loss": 0.5748,
      "step": 2530
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5134391784667969,
      "learning_rate": 1.9138269296997135e-05,
      "loss": 0.6069,
      "step": 2531
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5956342220306396,
      "learning_rate": 1.9137441892940828e-05,
      "loss": 0.5762,
      "step": 2532
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4996156692504883,
      "learning_rate": 1.9136614109755823e-05,
      "loss": 0.6229,
      "step": 2533
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5074937343597412,
      "learning_rate": 1.9135785947476472e-05,
      "loss": 0.5899,
      "step": 2534
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5347157716751099,
      "learning_rate": 1.9134957406137134e-05,
      "loss": 0.6058,
      "step": 2535
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5146158933639526,
      "learning_rate": 1.913412848577219e-05,
      "loss": 0.5889,
      "step": 2536
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5605477094650269,
      "learning_rate": 1.913329918641603e-05,
      "loss": 0.6091,
      "step": 2537
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4637619256973267,
      "learning_rate": 1.9132469508103063e-05,
      "loss": 0.6429,
      "step": 2538
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5278902053833008,
      "learning_rate": 1.9131639450867715e-05,
      "loss": 0.5848,
      "step": 2539
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5053949356079102,
      "learning_rate": 1.9130809014744426e-05,
      "loss": 0.5631,
      "step": 2540
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4476031064987183,
      "learning_rate": 1.9129978199767657e-05,
      "loss": 0.5988,
      "step": 2541
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.477929711341858,
      "learning_rate": 1.9129147005971872e-05,
      "loss": 0.5751,
      "step": 2542
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5360432863235474,
      "learning_rate": 1.9128315433391565e-05,
      "loss": 0.5756,
      "step": 2543
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5448800325393677,
      "learning_rate": 1.9127483482061232e-05,
      "loss": 0.6058,
      "step": 2544
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4689604043960571,
      "learning_rate": 1.9126651152015404e-05,
      "loss": 0.6025,
      "step": 2545
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4851065874099731,
      "learning_rate": 1.9125818443288606e-05,
      "loss": 0.5589,
      "step": 2546
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4393044710159302,
      "learning_rate": 1.912498535591539e-05,
      "loss": 0.5981,
      "step": 2547
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6533845663070679,
      "learning_rate": 1.9124151889930326e-05,
      "loss": 0.5788,
      "step": 2548
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5081642866134644,
      "learning_rate": 1.9123318045367992e-05,
      "loss": 0.5998,
      "step": 2549
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5036571025848389,
      "learning_rate": 1.912248382226299e-05,
      "loss": 0.5823,
      "step": 2550
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4362430572509766,
      "learning_rate": 1.912164922064993e-05,
      "loss": 0.5663,
      "step": 2551
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5692672729492188,
      "learning_rate": 1.9120814240563435e-05,
      "loss": 0.6047,
      "step": 2552
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5211879014968872,
      "learning_rate": 1.9119978882038162e-05,
      "loss": 0.6042,
      "step": 2553
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5193638801574707,
      "learning_rate": 1.911914314510877e-05,
      "loss": 0.594,
      "step": 2554
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5669879913330078,
      "learning_rate": 1.9118307029809927e-05,
      "loss": 0.6349,
      "step": 2555
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4624587297439575,
      "learning_rate": 1.9117470536176334e-05,
      "loss": 0.6189,
      "step": 2556
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.468210220336914,
      "learning_rate": 1.911663366424269e-05,
      "loss": 0.6019,
      "step": 2557
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6235703229904175,
      "learning_rate": 1.911579641404372e-05,
      "loss": 0.5779,
      "step": 2558
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5042035579681396,
      "learning_rate": 1.911495878561417e-05,
      "loss": 0.6003,
      "step": 2559
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.432774305343628,
      "learning_rate": 1.911412077898879e-05,
      "loss": 0.573,
      "step": 2560
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4817785024642944,
      "learning_rate": 1.9113282394202347e-05,
      "loss": 0.6178,
      "step": 2561
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4506696462631226,
      "learning_rate": 1.9112443631289635e-05,
      "loss": 0.5803,
      "step": 2562
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5603543519973755,
      "learning_rate": 1.9111604490285448e-05,
      "loss": 0.5973,
      "step": 2563
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.503652572631836,
      "learning_rate": 1.9110764971224608e-05,
      "loss": 0.5949,
      "step": 2564
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4724234342575073,
      "learning_rate": 1.9109925074141946e-05,
      "loss": 0.6047,
      "step": 2565
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.570448637008667,
      "learning_rate": 1.910908479907231e-05,
      "loss": 0.6076,
      "step": 2566
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3829809427261353,
      "learning_rate": 1.910824414605057e-05,
      "loss": 0.556,
      "step": 2567
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6575740575790405,
      "learning_rate": 1.91074031151116e-05,
      "loss": 0.6033,
      "step": 2568
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.612359642982483,
      "learning_rate": 1.9106561706290296e-05,
      "loss": 0.6504,
      "step": 2569
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4066351652145386,
      "learning_rate": 1.9105719919621575e-05,
      "loss": 0.5763,
      "step": 2570
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5149750709533691,
      "learning_rate": 1.910487775514036e-05,
      "loss": 0.6112,
      "step": 2571
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5125123262405396,
      "learning_rate": 1.9104035212881597e-05,
      "loss": 0.6337,
      "step": 2572
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3735578060150146,
      "learning_rate": 1.9103192292880237e-05,
      "loss": 0.605,
      "step": 2573
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9796966314315796,
      "learning_rate": 1.9102348995171264e-05,
      "loss": 0.6327,
      "step": 2574
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.456304907798767,
      "learning_rate": 1.9101505319789663e-05,
      "loss": 1.0103,
      "step": 2575
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5952479839324951,
      "learning_rate": 1.9100661266770436e-05,
      "loss": 0.588,
      "step": 2576
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4987218379974365,
      "learning_rate": 1.9099816836148615e-05,
      "loss": 0.6203,
      "step": 2577
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5354808568954468,
      "learning_rate": 1.9098972027959227e-05,
      "loss": 0.618,
      "step": 2578
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4985610246658325,
      "learning_rate": 1.909812684223733e-05,
      "loss": 0.6024,
      "step": 2579
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5233196020126343,
      "learning_rate": 1.9097281279017988e-05,
      "loss": 0.5709,
      "step": 2580
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6239227056503296,
      "learning_rate": 1.9096435338336287e-05,
      "loss": 0.608,
      "step": 2581
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.526979684829712,
      "learning_rate": 1.9095589020227327e-05,
      "loss": 0.5857,
      "step": 2582
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5856205224990845,
      "learning_rate": 1.9094742324726225e-05,
      "loss": 0.5993,
      "step": 2583
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5977853536605835,
      "learning_rate": 1.909389525186811e-05,
      "loss": 0.6306,
      "step": 2584
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4624654054641724,
      "learning_rate": 1.9093047801688126e-05,
      "loss": 0.6427,
      "step": 2585
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.544296145439148,
      "learning_rate": 1.909219997422144e-05,
      "loss": 0.6131,
      "step": 2586
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6175764799118042,
      "learning_rate": 1.9091351769503223e-05,
      "loss": 0.629,
      "step": 2587
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4781023263931274,
      "learning_rate": 1.9090503187568678e-05,
      "loss": 0.5925,
      "step": 2588
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.403241515159607,
      "learning_rate": 1.9089654228453004e-05,
      "loss": 0.6238,
      "step": 2589
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6708924770355225,
      "learning_rate": 1.9088804892191435e-05,
      "loss": 1.0316,
      "step": 2590
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7007505893707275,
      "learning_rate": 1.9087955178819207e-05,
      "loss": 0.6187,
      "step": 2591
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5113048553466797,
      "learning_rate": 1.9087105088371577e-05,
      "loss": 0.6169,
      "step": 2592
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.535568118095398,
      "learning_rate": 1.908625462088382e-05,
      "loss": 0.5878,
      "step": 2593
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5159912109375,
      "learning_rate": 1.9085403776391213e-05,
      "loss": 0.5882,
      "step": 2594
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5884124040603638,
      "learning_rate": 1.9084552554929067e-05,
      "loss": 0.598,
      "step": 2595
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.377307653427124,
      "learning_rate": 1.90837009565327e-05,
      "loss": 0.5739,
      "step": 2596
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4777543544769287,
      "learning_rate": 1.9082848981237448e-05,
      "loss": 0.5952,
      "step": 2597
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5113310813903809,
      "learning_rate": 1.9081996629078655e-05,
      "loss": 0.5579,
      "step": 2598
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4310795068740845,
      "learning_rate": 1.9081143900091694e-05,
      "loss": 0.6208,
      "step": 2599
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.471413016319275,
      "learning_rate": 1.9080290794311943e-05,
      "loss": 0.5993,
      "step": 2600
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4870362281799316,
      "learning_rate": 1.9079437311774796e-05,
      "loss": 0.595,
      "step": 2601
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4310493469238281,
      "learning_rate": 1.907858345251567e-05,
      "loss": 0.5857,
      "step": 2602
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.447859287261963,
      "learning_rate": 1.907772921656999e-05,
      "loss": 0.6137,
      "step": 2603
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5454469919204712,
      "learning_rate": 1.90768746039732e-05,
      "loss": 0.5863,
      "step": 2604
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4963834285736084,
      "learning_rate": 1.9076019614760762e-05,
      "loss": 0.6104,
      "step": 2605
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4348490238189697,
      "learning_rate": 1.907516424896815e-05,
      "loss": 0.6309,
      "step": 2606
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6280394792556763,
      "learning_rate": 1.9074308506630853e-05,
      "loss": 0.5969,
      "step": 2607
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.474452018737793,
      "learning_rate": 1.907345238778438e-05,
      "loss": 0.62,
      "step": 2608
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.489879846572876,
      "learning_rate": 1.9072595892464248e-05,
      "loss": 0.5982,
      "step": 2609
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3741612434387207,
      "learning_rate": 1.9071739020706e-05,
      "loss": 0.5697,
      "step": 2610
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4662644863128662,
      "learning_rate": 1.9070881772545184e-05,
      "loss": 0.5903,
      "step": 2611
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5434162616729736,
      "learning_rate": 1.9070024148017375e-05,
      "loss": 0.6127,
      "step": 2612
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4756450653076172,
      "learning_rate": 1.906916614715815e-05,
      "loss": 0.6592,
      "step": 2613
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5157158374786377,
      "learning_rate": 1.9068307770003115e-05,
      "loss": 0.6339,
      "step": 2614
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4709371328353882,
      "learning_rate": 1.9067449016587884e-05,
      "loss": 0.5622,
      "step": 2615
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4929395914077759,
      "learning_rate": 1.906658988694809e-05,
      "loss": 0.6219,
      "step": 2616
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5726667642593384,
      "learning_rate": 1.9065730381119373e-05,
      "loss": 0.6242,
      "step": 2617
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4636157751083374,
      "learning_rate": 1.9064870499137402e-05,
      "loss": 0.6306,
      "step": 2618
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4098635911941528,
      "learning_rate": 1.9064010241037855e-05,
      "loss": 0.5719,
      "step": 2619
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4798038005828857,
      "learning_rate": 1.906314960685642e-05,
      "loss": 0.5588,
      "step": 2620
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.443837285041809,
      "learning_rate": 1.906228859662881e-05,
      "loss": 0.6048,
      "step": 2621
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5077731609344482,
      "learning_rate": 1.9061427210390754e-05,
      "loss": 0.6333,
      "step": 2622
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4945563077926636,
      "learning_rate": 1.9060565448177983e-05,
      "loss": 0.5817,
      "step": 2623
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3909411430358887,
      "learning_rate": 1.905970331002626e-05,
      "loss": 0.6145,
      "step": 2624
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5771422386169434,
      "learning_rate": 1.9058840795971356e-05,
      "loss": 0.6003,
      "step": 2625
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6698769330978394,
      "learning_rate": 1.9057977906049056e-05,
      "loss": 0.9945,
      "step": 2626
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4926669597625732,
      "learning_rate": 1.9057114640295165e-05,
      "loss": 0.6405,
      "step": 2627
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4785916805267334,
      "learning_rate": 1.90562509987455e-05,
      "loss": 0.6206,
      "step": 2628
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.51108717918396,
      "learning_rate": 1.9055386981435892e-05,
      "loss": 0.5772,
      "step": 2629
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3751181364059448,
      "learning_rate": 1.9054522588402196e-05,
      "loss": 0.5754,
      "step": 2630
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5293779373168945,
      "learning_rate": 1.905365781968028e-05,
      "loss": 0.5942,
      "step": 2631
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5487843751907349,
      "learning_rate": 1.9052792675306016e-05,
      "loss": 0.5811,
      "step": 2632
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5736932754516602,
      "learning_rate": 1.9051927155315302e-05,
      "loss": 0.5761,
      "step": 2633
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5901035070419312,
      "learning_rate": 1.9051061259744054e-05,
      "loss": 0.633,
      "step": 2634
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5028983354568481,
      "learning_rate": 1.90501949886282e-05,
      "loss": 0.6049,
      "step": 2635
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5973985195159912,
      "learning_rate": 1.9049328342003678e-05,
      "loss": 0.5774,
      "step": 2636
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5354748964309692,
      "learning_rate": 1.9048461319906454e-05,
      "loss": 0.5863,
      "step": 2637
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4636213779449463,
      "learning_rate": 1.9047593922372494e-05,
      "loss": 0.5816,
      "step": 2638
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.560141921043396,
      "learning_rate": 1.904672614943779e-05,
      "loss": 0.552,
      "step": 2639
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5788511037826538,
      "learning_rate": 1.9045858001138353e-05,
      "loss": 0.6099,
      "step": 2640
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4759544134140015,
      "learning_rate": 1.9044989477510197e-05,
      "loss": 0.5667,
      "step": 2641
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5445307493209839,
      "learning_rate": 1.9044120578589366e-05,
      "loss": 0.6214,
      "step": 2642
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3756812810897827,
      "learning_rate": 1.904325130441191e-05,
      "loss": 0.5834,
      "step": 2643
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5165467262268066,
      "learning_rate": 1.904238165501389e-05,
      "loss": 0.6323,
      "step": 2644
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4216899871826172,
      "learning_rate": 1.9041511630431394e-05,
      "loss": 0.5931,
      "step": 2645
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.360261082649231,
      "learning_rate": 1.904064123070052e-05,
      "loss": 0.5954,
      "step": 2646
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4825878143310547,
      "learning_rate": 1.9039770455857386e-05,
      "loss": 0.5971,
      "step": 2647
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.581060767173767,
      "learning_rate": 1.9038899305938118e-05,
      "loss": 0.6133,
      "step": 2648
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.474710464477539,
      "learning_rate": 1.9038027780978865e-05,
      "loss": 0.5873,
      "step": 2649
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5316871404647827,
      "learning_rate": 1.9037155881015785e-05,
      "loss": 0.5709,
      "step": 2650
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.651142954826355,
      "learning_rate": 1.9036283606085057e-05,
      "loss": 0.5962,
      "step": 2651
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.562027931213379,
      "learning_rate": 1.9035410956222868e-05,
      "loss": 0.587,
      "step": 2652
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4507049322128296,
      "learning_rate": 1.9034537931465432e-05,
      "loss": 0.6201,
      "step": 2653
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6060783863067627,
      "learning_rate": 1.903366453184897e-05,
      "loss": 0.6281,
      "step": 2654
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3997527360916138,
      "learning_rate": 1.9032790757409724e-05,
      "loss": 0.5871,
      "step": 2655
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.377685546875,
      "learning_rate": 1.903191660818394e-05,
      "loss": 0.6155,
      "step": 2656
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5195633172988892,
      "learning_rate": 1.9031042084207895e-05,
      "loss": 0.574,
      "step": 2657
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4222480058670044,
      "learning_rate": 1.9030167185517877e-05,
      "loss": 0.576,
      "step": 2658
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5865216255187988,
      "learning_rate": 1.9029291912150177e-05,
      "loss": 0.6114,
      "step": 2659
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3836464881896973,
      "learning_rate": 1.9028416264141122e-05,
      "loss": 0.593,
      "step": 2660
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4588080644607544,
      "learning_rate": 1.9027540241527037e-05,
      "loss": 0.5956,
      "step": 2661
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4706465005874634,
      "learning_rate": 1.9026663844344276e-05,
      "loss": 0.6011,
      "step": 2662
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4366188049316406,
      "learning_rate": 1.9025787072629197e-05,
      "loss": 0.5786,
      "step": 2663
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4068852663040161,
      "learning_rate": 1.902490992641818e-05,
      "loss": 0.6229,
      "step": 2664
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4072139263153076,
      "learning_rate": 1.902403240574762e-05,
      "loss": 0.6083,
      "step": 2665
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4758281707763672,
      "learning_rate": 1.902315451065393e-05,
      "loss": 0.5884,
      "step": 2666
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4436253309249878,
      "learning_rate": 1.902227624117353e-05,
      "loss": 0.6335,
      "step": 2667
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3597619533538818,
      "learning_rate": 1.9021397597342863e-05,
      "loss": 0.6158,
      "step": 2668
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.418069839477539,
      "learning_rate": 1.902051857919839e-05,
      "loss": 0.5822,
      "step": 2669
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.449960708618164,
      "learning_rate": 1.9019639186776576e-05,
      "loss": 0.6068,
      "step": 2670
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4758144617080688,
      "learning_rate": 1.9018759420113908e-05,
      "loss": 0.6023,
      "step": 2671
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7216898202896118,
      "learning_rate": 1.9017879279246897e-05,
      "loss": 1.0818,
      "step": 2672
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5500966310501099,
      "learning_rate": 1.9016998764212057e-05,
      "loss": 0.6006,
      "step": 2673
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.533624291419983,
      "learning_rate": 1.9016117875045918e-05,
      "loss": 0.6014,
      "step": 2674
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6887701749801636,
      "learning_rate": 1.901523661178504e-05,
      "loss": 0.6086,
      "step": 2675
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6935988664627075,
      "learning_rate": 1.9014354974465977e-05,
      "loss": 0.5968,
      "step": 2676
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4349968433380127,
      "learning_rate": 1.9013472963125318e-05,
      "loss": 0.5892,
      "step": 2677
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6180118322372437,
      "learning_rate": 1.9012590577799653e-05,
      "loss": 0.5919,
      "step": 2678
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5732225179672241,
      "learning_rate": 1.90117078185256e-05,
      "loss": 0.6049,
      "step": 2679
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4693093299865723,
      "learning_rate": 1.9010824685339782e-05,
      "loss": 0.9844,
      "step": 2680
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.633956789970398,
      "learning_rate": 1.9009941178278844e-05,
      "loss": 0.6061,
      "step": 2681
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6519396305084229,
      "learning_rate": 1.900905729737944e-05,
      "loss": 0.5854,
      "step": 2682
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4054416418075562,
      "learning_rate": 1.9008173042678248e-05,
      "loss": 0.5811,
      "step": 2683
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5381534099578857,
      "learning_rate": 1.900728841421196e-05,
      "loss": 0.6361,
      "step": 2684
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5539802312850952,
      "learning_rate": 1.9006403412017273e-05,
      "loss": 0.5665,
      "step": 2685
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4624415636062622,
      "learning_rate": 1.900551803613091e-05,
      "loss": 0.6166,
      "step": 2686
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4927735328674316,
      "learning_rate": 1.9004632286589612e-05,
      "loss": 0.5867,
      "step": 2687
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.502066731452942,
      "learning_rate": 1.9003746163430123e-05,
      "loss": 0.5861,
      "step": 2688
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5038952827453613,
      "learning_rate": 1.9002859666689216e-05,
      "loss": 0.5783,
      "step": 2689
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4086142778396606,
      "learning_rate": 1.900197279640367e-05,
      "loss": 0.6243,
      "step": 2690
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.457011342048645,
      "learning_rate": 1.900108555261028e-05,
      "loss": 0.5619,
      "step": 2691
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3788217306137085,
      "learning_rate": 1.9000197935345866e-05,
      "loss": 0.5796,
      "step": 2692
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.437461256980896,
      "learning_rate": 1.8999309944647254e-05,
      "loss": 0.5504,
      "step": 2693
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.368642807006836,
      "learning_rate": 1.8998421580551285e-05,
      "loss": 0.563,
      "step": 2694
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.575412392616272,
      "learning_rate": 1.8997532843094823e-05,
      "loss": 0.564,
      "step": 2695
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5781127214431763,
      "learning_rate": 1.8996643732314743e-05,
      "loss": 0.5679,
      "step": 2696
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3999598026275635,
      "learning_rate": 1.8995754248247935e-05,
      "loss": 0.5454,
      "step": 2697
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5881913900375366,
      "learning_rate": 1.8994864390931304e-05,
      "loss": 0.5796,
      "step": 2698
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5630104541778564,
      "learning_rate": 1.8993974160401772e-05,
      "loss": 0.6462,
      "step": 2699
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4324036836624146,
      "learning_rate": 1.8993083556696277e-05,
      "loss": 0.5961,
      "step": 2700
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.452718734741211,
      "learning_rate": 1.899219257985177e-05,
      "loss": 0.5734,
      "step": 2701
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4770864248275757,
      "learning_rate": 1.8991301229905225e-05,
      "loss": 0.5997,
      "step": 2702
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6577904224395752,
      "learning_rate": 1.8990409506893618e-05,
      "loss": 0.5465,
      "step": 2703
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5122541189193726,
      "learning_rate": 1.8989517410853956e-05,
      "loss": 0.5764,
      "step": 2704
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5301066637039185,
      "learning_rate": 1.8988624941823244e-05,
      "loss": 0.6005,
      "step": 2705
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5388706922531128,
      "learning_rate": 1.8987732099838522e-05,
      "loss": 0.589,
      "step": 2706
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4508253335952759,
      "learning_rate": 1.8986838884936828e-05,
      "loss": 0.5602,
      "step": 2707
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4978386163711548,
      "learning_rate": 1.898594529715523e-05,
      "loss": 0.631,
      "step": 2708
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6168835163116455,
      "learning_rate": 1.89850513365308e-05,
      "loss": 0.5876,
      "step": 2709
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4333293437957764,
      "learning_rate": 1.8984157003100628e-05,
      "loss": 0.6013,
      "step": 2710
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.440584659576416,
      "learning_rate": 1.8983262296901827e-05,
      "loss": 0.5809,
      "step": 2711
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.695663571357727,
      "learning_rate": 1.8982367217971514e-05,
      "loss": 0.6511,
      "step": 2712
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4008653163909912,
      "learning_rate": 1.8981471766346833e-05,
      "loss": 0.5864,
      "step": 2713
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5152497291564941,
      "learning_rate": 1.8980575942064935e-05,
      "loss": 0.9461,
      "step": 2714
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7294450998306274,
      "learning_rate": 1.8979679745162995e-05,
      "loss": 0.5889,
      "step": 2715
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6788650751113892,
      "learning_rate": 1.8978783175678183e-05,
      "loss": 0.6312,
      "step": 2716
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5126789808273315,
      "learning_rate": 1.8977886233647715e-05,
      "loss": 0.6499,
      "step": 2717
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6200640201568604,
      "learning_rate": 1.8976988919108802e-05,
      "loss": 0.5631,
      "step": 2718
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4826009273529053,
      "learning_rate": 1.897609123209867e-05,
      "loss": 0.5983,
      "step": 2719
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4862620830535889,
      "learning_rate": 1.8975193172654573e-05,
      "loss": 0.5796,
      "step": 2720
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5792546272277832,
      "learning_rate": 1.8974294740813765e-05,
      "loss": 0.5882,
      "step": 2721
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5882946252822876,
      "learning_rate": 1.897339593661353e-05,
      "loss": 0.9791,
      "step": 2722
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7254302501678467,
      "learning_rate": 1.897249676009116e-05,
      "loss": 0.5932,
      "step": 2723
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5402292013168335,
      "learning_rate": 1.897159721128396e-05,
      "loss": 0.6605,
      "step": 2724
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5618494749069214,
      "learning_rate": 1.897069729022926e-05,
      "loss": 0.6148,
      "step": 2725
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.424680233001709,
      "learning_rate": 1.896979699696439e-05,
      "loss": 0.5686,
      "step": 2726
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4649386405944824,
      "learning_rate": 1.8968896331526717e-05,
      "loss": 0.5958,
      "step": 2727
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4535595178604126,
      "learning_rate": 1.8967995293953604e-05,
      "loss": 0.5687,
      "step": 2728
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4892688989639282,
      "learning_rate": 1.896709388428243e-05,
      "loss": 0.5783,
      "step": 2729
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5848169326782227,
      "learning_rate": 1.896619210255061e-05,
      "loss": 0.6196,
      "step": 2730
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.48268461227417,
      "learning_rate": 1.8965289948795553e-05,
      "loss": 0.5975,
      "step": 2731
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3964824676513672,
      "learning_rate": 1.896438742305469e-05,
      "loss": 0.5771,
      "step": 2732
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5687810182571411,
      "learning_rate": 1.8963484525365476e-05,
      "loss": 0.5736,
      "step": 2733
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4655038118362427,
      "learning_rate": 1.896258125576536e-05,
      "loss": 0.6298,
      "step": 2734
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4461020231246948,
      "learning_rate": 1.8961677614291833e-05,
      "loss": 0.5819,
      "step": 2735
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4876999855041504,
      "learning_rate": 1.8960773600982384e-05,
      "loss": 0.5811,
      "step": 2736
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4765583276748657,
      "learning_rate": 1.895986921587452e-05,
      "loss": 0.5789,
      "step": 2737
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3884475231170654,
      "learning_rate": 1.8958964459005773e-05,
      "loss": 0.6027,
      "step": 2738
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4994429349899292,
      "learning_rate": 1.8958059330413673e-05,
      "loss": 0.6179,
      "step": 2739
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4552361965179443,
      "learning_rate": 1.895715383013578e-05,
      "loss": 0.5942,
      "step": 2740
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.484049677848816,
      "learning_rate": 1.8956247958209662e-05,
      "loss": 0.6133,
      "step": 2741
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4921467304229736,
      "learning_rate": 1.8955341714672916e-05,
      "loss": 0.5816,
      "step": 2742
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.255405306816101,
      "learning_rate": 1.895443509956313e-05,
      "loss": 0.9961,
      "step": 2743
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.567042350769043,
      "learning_rate": 1.895352811291793e-05,
      "loss": 0.6091,
      "step": 2744
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4943428039550781,
      "learning_rate": 1.8952620754774945e-05,
      "loss": 0.566,
      "step": 2745
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3979182243347168,
      "learning_rate": 1.895171302517182e-05,
      "loss": 0.58,
      "step": 2746
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5279783010482788,
      "learning_rate": 1.8950804924146223e-05,
      "loss": 0.6183,
      "step": 2747
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5629302263259888,
      "learning_rate": 1.8949896451735832e-05,
      "loss": 0.5903,
      "step": 2748
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4162020683288574,
      "learning_rate": 1.894898760797834e-05,
      "loss": 0.5953,
      "step": 2749
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3510191440582275,
      "learning_rate": 1.8948078392911458e-05,
      "loss": 0.5781,
      "step": 2750
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3622034788131714,
      "learning_rate": 1.8947168806572907e-05,
      "loss": 1.0348,
      "step": 2751
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.707818865776062,
      "learning_rate": 1.8946258849000436e-05,
      "loss": 0.6081,
      "step": 2752
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.648437261581421,
      "learning_rate": 1.894534852023179e-05,
      "loss": 0.5684,
      "step": 2753
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4131041765213013,
      "learning_rate": 1.8944437820304747e-05,
      "loss": 0.5946,
      "step": 2754
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.509548306465149,
      "learning_rate": 1.8943526749257092e-05,
      "loss": 0.6184,
      "step": 2755
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5148805379867554,
      "learning_rate": 1.8942615307126627e-05,
      "loss": 0.6375,
      "step": 2756
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.543106198310852,
      "learning_rate": 1.8941703493951163e-05,
      "loss": 0.5887,
      "step": 2757
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.46307373046875,
      "learning_rate": 1.8940791309768547e-05,
      "loss": 0.6067,
      "step": 2758
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5035806894302368,
      "learning_rate": 1.8939878754616616e-05,
      "loss": 0.5808,
      "step": 2759
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5047498941421509,
      "learning_rate": 1.8938965828533236e-05,
      "loss": 0.5899,
      "step": 2760
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4424293041229248,
      "learning_rate": 1.8938052531556285e-05,
      "loss": 0.5756,
      "step": 2761
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4793251752853394,
      "learning_rate": 1.8937138863723663e-05,
      "loss": 0.5791,
      "step": 2762
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4958919286727905,
      "learning_rate": 1.893622482507327e-05,
      "loss": 0.6221,
      "step": 2763
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4632169008255005,
      "learning_rate": 1.893531041564304e-05,
      "loss": 0.6446,
      "step": 2764
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.390894889831543,
      "learning_rate": 1.893439563547091e-05,
      "loss": 0.5601,
      "step": 2765
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.537914752960205,
      "learning_rate": 1.8933480484594837e-05,
      "loss": 0.6087,
      "step": 2766
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3862478733062744,
      "learning_rate": 1.8932564963052785e-05,
      "loss": 0.5795,
      "step": 2767
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.411312222480774,
      "learning_rate": 1.8931649070882754e-05,
      "loss": 0.6041,
      "step": 2768
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6315492391586304,
      "learning_rate": 1.8930732808122733e-05,
      "loss": 0.6018,
      "step": 2769
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.412427544593811,
      "learning_rate": 1.8929816174810748e-05,
      "loss": 0.5975,
      "step": 2770
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3993479013442993,
      "learning_rate": 1.892889917098483e-05,
      "loss": 0.6022,
      "step": 2771
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.642385482788086,
      "learning_rate": 1.8927981796683022e-05,
      "loss": 0.5861,
      "step": 2772
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.7574234008789062,
      "learning_rate": 1.8927064051943393e-05,
      "loss": 0.5638,
      "step": 2773
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.512546181678772,
      "learning_rate": 1.892614593680402e-05,
      "loss": 0.5526,
      "step": 2774
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5235576629638672,
      "learning_rate": 1.8925227451303e-05,
      "loss": 0.6008,
      "step": 2775
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.457882046699524,
      "learning_rate": 1.892430859547844e-05,
      "loss": 0.602,
      "step": 2776
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.468570590019226,
      "learning_rate": 1.892338936936846e-05,
      "loss": 0.6007,
      "step": 2777
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.457924246788025,
      "learning_rate": 1.8922469773011208e-05,
      "loss": 0.5803,
      "step": 2778
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4318922758102417,
      "learning_rate": 1.892154980644484e-05,
      "loss": 0.5914,
      "step": 2779
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.493300199508667,
      "learning_rate": 1.892062946970752e-05,
      "loss": 0.5882,
      "step": 2780
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4607733488082886,
      "learning_rate": 1.891970876283744e-05,
      "loss": 0.5956,
      "step": 2781
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4092562198638916,
      "learning_rate": 1.89187876858728e-05,
      "loss": 0.6038,
      "step": 2782
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4300868511199951,
      "learning_rate": 1.891786623885182e-05,
      "loss": 0.5864,
      "step": 2783
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4023692607879639,
      "learning_rate": 1.8916944421812727e-05,
      "loss": 0.6061,
      "step": 2784
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4157767295837402,
      "learning_rate": 1.8916022234793772e-05,
      "loss": 0.5565,
      "step": 2785
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5016810894012451,
      "learning_rate": 1.891509967783322e-05,
      "loss": 0.5664,
      "step": 2786
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4175186157226562,
      "learning_rate": 1.8914176750969342e-05,
      "loss": 0.5627,
      "step": 2787
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.395292043685913,
      "learning_rate": 1.891325345424044e-05,
      "loss": 0.6045,
      "step": 2788
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4339922666549683,
      "learning_rate": 1.8912329787684823e-05,
      "loss": 0.604,
      "step": 2789
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4258049726486206,
      "learning_rate": 1.891140575134081e-05,
      "loss": 0.5792,
      "step": 2790
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4751743078231812,
      "learning_rate": 1.8910481345246746e-05,
      "loss": 0.5844,
      "step": 2791
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5076370239257812,
      "learning_rate": 1.8909556569440984e-05,
      "loss": 0.5847,
      "step": 2792
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5150847434997559,
      "learning_rate": 1.8908631423961893e-05,
      "loss": 0.6039,
      "step": 2793
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4701486825942993,
      "learning_rate": 1.8907705908847862e-05,
      "loss": 0.5967,
      "step": 2794
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4956843852996826,
      "learning_rate": 1.890678002413729e-05,
      "loss": 0.5448,
      "step": 2795
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3462003469467163,
      "learning_rate": 1.8905853769868598e-05,
      "loss": 0.5994,
      "step": 2796
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4656132459640503,
      "learning_rate": 1.890492714608021e-05,
      "loss": 0.5983,
      "step": 2797
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.363145351409912,
      "learning_rate": 1.890400015281058e-05,
      "loss": 0.6085,
      "step": 2798
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3875911235809326,
      "learning_rate": 1.8903072790098166e-05,
      "loss": 0.5876,
      "step": 2799
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4405796527862549,
      "learning_rate": 1.890214505798145e-05,
      "loss": 0.5748,
      "step": 2800
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4388805627822876,
      "learning_rate": 1.8901216956498927e-05,
      "loss": 0.6072,
      "step": 2801
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4688191413879395,
      "learning_rate": 1.89002884856891e-05,
      "loss": 0.6219,
      "step": 2802
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.331599473953247,
      "learning_rate": 1.8899359645590493e-05,
      "loss": 0.6069,
      "step": 2803
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4686331748962402,
      "learning_rate": 1.8898430436241647e-05,
      "loss": 0.6157,
      "step": 2804
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.414666771888733,
      "learning_rate": 1.8897500857681117e-05,
      "loss": 0.608,
      "step": 2805
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3900099992752075,
      "learning_rate": 1.8896570909947477e-05,
      "loss": 0.6025,
      "step": 2806
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5080525875091553,
      "learning_rate": 1.8895640593079302e-05,
      "loss": 0.5707,
      "step": 2807
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4957443475723267,
      "learning_rate": 1.8894709907115202e-05,
      "loss": 0.5978,
      "step": 2808
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5137174129486084,
      "learning_rate": 1.8893778852093788e-05,
      "loss": 0.641,
      "step": 2809
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5391979217529297,
      "learning_rate": 1.8892847428053692e-05,
      "loss": 0.6415,
      "step": 2810
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4517310857772827,
      "learning_rate": 1.889191563503356e-05,
      "loss": 0.5815,
      "step": 2811
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3441331386566162,
      "learning_rate": 1.8890983473072054e-05,
      "loss": 0.5806,
      "step": 2812
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3896843194961548,
      "learning_rate": 1.8890050942207857e-05,
      "loss": 0.5823,
      "step": 2813
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.469037413597107,
      "learning_rate": 1.8889118042479652e-05,
      "loss": 0.5573,
      "step": 2814
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4152953624725342,
      "learning_rate": 1.888818477392615e-05,
      "loss": 0.5924,
      "step": 2815
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3795814514160156,
      "learning_rate": 1.8887251136586076e-05,
      "loss": 0.6257,
      "step": 2816
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3793669939041138,
      "learning_rate": 1.8886317130498164e-05,
      "loss": 0.564,
      "step": 2817
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3908051252365112,
      "learning_rate": 1.888538275570117e-05,
      "loss": 0.6068,
      "step": 2818
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3913109302520752,
      "learning_rate": 1.888444801223387e-05,
      "loss": 0.5828,
      "step": 2819
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5389907360076904,
      "learning_rate": 1.8883512900135038e-05,
      "loss": 0.6274,
      "step": 2820
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6002607345581055,
      "learning_rate": 1.8882577419443473e-05,
      "loss": 0.5917,
      "step": 2821
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5061556100845337,
      "learning_rate": 1.8881641570197997e-05,
      "loss": 0.6017,
      "step": 2822
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5430324077606201,
      "learning_rate": 1.8880705352437437e-05,
      "loss": 0.6002,
      "step": 2823
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.701945424079895,
      "learning_rate": 1.8879768766200638e-05,
      "loss": 0.5804,
      "step": 2824
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4810922145843506,
      "learning_rate": 1.887883181152646e-05,
      "loss": 0.6181,
      "step": 2825
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5121026039123535,
      "learning_rate": 1.8877894488453778e-05,
      "loss": 0.5889,
      "step": 2826
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6275583505630493,
      "learning_rate": 1.8876956797021488e-05,
      "loss": 0.6393,
      "step": 2827
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5854336023330688,
      "learning_rate": 1.8876018737268488e-05,
      "loss": 0.5874,
      "step": 2828
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.615003228187561,
      "learning_rate": 1.887508030923371e-05,
      "loss": 0.6047,
      "step": 2829
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6521271467208862,
      "learning_rate": 1.8874141512956087e-05,
      "loss": 0.5677,
      "step": 2830
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.540893793106079,
      "learning_rate": 1.8873202348474565e-05,
      "loss": 0.5718,
      "step": 2831
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5483040809631348,
      "learning_rate": 1.887226281582812e-05,
      "loss": 0.6222,
      "step": 2832
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4969583749771118,
      "learning_rate": 1.8871322915055734e-05,
      "loss": 0.6094,
      "step": 2833
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5704295635223389,
      "learning_rate": 1.8870382646196403e-05,
      "loss": 0.5676,
      "step": 2834
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5186645984649658,
      "learning_rate": 1.8869442009289133e-05,
      "loss": 0.6068,
      "step": 2835
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4911972284317017,
      "learning_rate": 1.8868501004372967e-05,
      "loss": 0.5737,
      "step": 2836
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4429354667663574,
      "learning_rate": 1.886755963148694e-05,
      "loss": 0.5949,
      "step": 2837
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.415108323097229,
      "learning_rate": 1.8866617890670116e-05,
      "loss": 0.5752,
      "step": 2838
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.476253867149353,
      "learning_rate": 1.8865675781961562e-05,
      "loss": 0.5815,
      "step": 2839
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.370142936706543,
      "learning_rate": 1.8864733305400378e-05,
      "loss": 0.5808,
      "step": 2840
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4416515827178955,
      "learning_rate": 1.886379046102566e-05,
      "loss": 0.594,
      "step": 2841
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4391509294509888,
      "learning_rate": 1.886284724887653e-05,
      "loss": 0.588,
      "step": 2842
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.954014539718628,
      "learning_rate": 1.886190366899213e-05,
      "loss": 0.5645,
      "step": 2843
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.391229510307312,
      "learning_rate": 1.8860959721411604e-05,
      "loss": 0.5832,
      "step": 2844
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4350532293319702,
      "learning_rate": 1.886001540617412e-05,
      "loss": 0.5583,
      "step": 2845
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.405492901802063,
      "learning_rate": 1.885907072331886e-05,
      "loss": 0.6205,
      "step": 2846
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3656632900238037,
      "learning_rate": 1.8858125672885023e-05,
      "loss": 0.568,
      "step": 2847
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4814492464065552,
      "learning_rate": 1.8857180254911816e-05,
      "loss": 0.5879,
      "step": 2848
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.391994833946228,
      "learning_rate": 1.8856234469438467e-05,
      "loss": 0.5549,
      "step": 2849
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4102767705917358,
      "learning_rate": 1.8855288316504218e-05,
      "loss": 0.6141,
      "step": 2850
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4219602346420288,
      "learning_rate": 1.8854341796148333e-05,
      "loss": 0.5958,
      "step": 2851
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4419243335723877,
      "learning_rate": 1.8853394908410078e-05,
      "loss": 0.5992,
      "step": 2852
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4004042148590088,
      "learning_rate": 1.885244765332874e-05,
      "loss": 0.6239,
      "step": 2853
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4756108522415161,
      "learning_rate": 1.8851500030943625e-05,
      "loss": 0.6013,
      "step": 2854
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.498624563217163,
      "learning_rate": 1.8850552041294053e-05,
      "loss": 0.555,
      "step": 2855
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.450300931930542,
      "learning_rate": 1.884960368441936e-05,
      "loss": 0.5971,
      "step": 2856
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4802653789520264,
      "learning_rate": 1.8848654960358887e-05,
      "loss": 0.601,
      "step": 2857
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.612383484840393,
      "learning_rate": 1.8847705869152006e-05,
      "loss": 1.0368,
      "step": 2858
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6186778545379639,
      "learning_rate": 1.8846756410838088e-05,
      "loss": 0.572,
      "step": 2859
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4469170570373535,
      "learning_rate": 1.884580658545654e-05,
      "loss": 0.621,
      "step": 2860
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6425336599349976,
      "learning_rate": 1.8844856393046763e-05,
      "loss": 0.6106,
      "step": 2861
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4935388565063477,
      "learning_rate": 1.884390583364818e-05,
      "loss": 0.5859,
      "step": 2862
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3871313333511353,
      "learning_rate": 1.8842954907300236e-05,
      "loss": 0.5874,
      "step": 2863
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5642285346984863,
      "learning_rate": 1.884200361404239e-05,
      "loss": 0.5733,
      "step": 2864
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.399235725402832,
      "learning_rate": 1.8841051953914107e-05,
      "loss": 0.5508,
      "step": 2865
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4665220975875854,
      "learning_rate": 1.8840099926954873e-05,
      "loss": 1.0458,
      "step": 2866
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5696377754211426,
      "learning_rate": 1.8839147533204194e-05,
      "loss": 0.613,
      "step": 2867
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6838574409484863,
      "learning_rate": 1.883819477270158e-05,
      "loss": 0.5979,
      "step": 2868
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4238959550857544,
      "learning_rate": 1.883724164548657e-05,
      "loss": 0.6086,
      "step": 2869
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4234658479690552,
      "learning_rate": 1.8836288151598706e-05,
      "loss": 0.5822,
      "step": 2870
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.436424732208252,
      "learning_rate": 1.8835334291077555e-05,
      "loss": 0.6097,
      "step": 2871
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3790899515151978,
      "learning_rate": 1.8834380063962686e-05,
      "loss": 0.5641,
      "step": 2872
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3747611045837402,
      "learning_rate": 1.88334254702937e-05,
      "loss": 0.6167,
      "step": 2873
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5851631164550781,
      "learning_rate": 1.8832470510110198e-05,
      "loss": 0.6314,
      "step": 2874
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4468451738357544,
      "learning_rate": 1.883151518345181e-05,
      "loss": 0.6302,
      "step": 2875
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.395129680633545,
      "learning_rate": 1.8830559490358168e-05,
      "loss": 0.5872,
      "step": 2876
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4539141654968262,
      "learning_rate": 1.8829603430868928e-05,
      "loss": 0.5946,
      "step": 2877
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4883267879486084,
      "learning_rate": 1.882864700502376e-05,
      "loss": 0.632,
      "step": 2878
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5104913711547852,
      "learning_rate": 1.8827690212862343e-05,
      "loss": 0.5846,
      "step": 2879
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5117045640945435,
      "learning_rate": 1.882673305442438e-05,
      "loss": 0.597,
      "step": 2880
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3605990409851074,
      "learning_rate": 1.8825775529749588e-05,
      "loss": 0.5915,
      "step": 2881
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4463727474212646,
      "learning_rate": 1.8824817638877693e-05,
      "loss": 0.5977,
      "step": 2882
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3830626010894775,
      "learning_rate": 1.882385938184844e-05,
      "loss": 0.5879,
      "step": 2883
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4646847248077393,
      "learning_rate": 1.8822900758701582e-05,
      "loss": 0.5732,
      "step": 2884
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.556897759437561,
      "learning_rate": 1.8821941769476905e-05,
      "loss": 0.6164,
      "step": 2885
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.468887448310852,
      "learning_rate": 1.8820982414214194e-05,
      "loss": 0.5747,
      "step": 2886
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4055606126785278,
      "learning_rate": 1.8820022692953256e-05,
      "loss": 0.5511,
      "step": 2887
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4435597658157349,
      "learning_rate": 1.8819062605733908e-05,
      "loss": 0.5635,
      "step": 2888
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4742064476013184,
      "learning_rate": 1.8818102152595988e-05,
      "loss": 0.5967,
      "step": 2889
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4692213535308838,
      "learning_rate": 1.8817141333579347e-05,
      "loss": 0.5726,
      "step": 2890
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4025237560272217,
      "learning_rate": 1.8816180148723854e-05,
      "loss": 0.5668,
      "step": 2891
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.458909034729004,
      "learning_rate": 1.8815218598069386e-05,
      "loss": 0.587,
      "step": 2892
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4971436262130737,
      "learning_rate": 1.881425668165584e-05,
      "loss": 0.6098,
      "step": 2893
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5002174377441406,
      "learning_rate": 1.881329439952313e-05,
      "loss": 0.5657,
      "step": 2894
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.520529866218567,
      "learning_rate": 1.8812331751711177e-05,
      "loss": 0.5862,
      "step": 2895
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.666106939315796,
      "learning_rate": 1.881136873825993e-05,
      "loss": 1.0468,
      "step": 2896
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6485412120819092,
      "learning_rate": 1.881040535920934e-05,
      "loss": 0.6101,
      "step": 2897
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.573036551475525,
      "learning_rate": 1.8809441614599387e-05,
      "loss": 0.5846,
      "step": 2898
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5080887079238892,
      "learning_rate": 1.8808477504470057e-05,
      "loss": 0.596,
      "step": 2899
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6035377979278564,
      "learning_rate": 1.880751302886134e-05,
      "loss": 0.583,
      "step": 2900
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6075729131698608,
      "learning_rate": 1.880654818781327e-05,
      "loss": 0.5857,
      "step": 2901
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4204256534576416,
      "learning_rate": 1.8805582981365877e-05,
      "loss": 0.5581,
      "step": 2902
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5322391986846924,
      "learning_rate": 1.88046174095592e-05,
      "loss": 0.5905,
      "step": 2903
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6116596460342407,
      "learning_rate": 1.8803651472433308e-05,
      "loss": 0.6101,
      "step": 2904
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5252680778503418,
      "learning_rate": 1.8802685170028282e-05,
      "loss": 0.5507,
      "step": 2905
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4353132247924805,
      "learning_rate": 1.880171850238421e-05,
      "loss": 0.5878,
      "step": 2906
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6663414239883423,
      "learning_rate": 1.8800751469541205e-05,
      "loss": 0.6168,
      "step": 2907
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5381532907485962,
      "learning_rate": 1.8799784071539392e-05,
      "loss": 0.5689,
      "step": 2908
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4377071857452393,
      "learning_rate": 1.8798816308418903e-05,
      "loss": 0.58,
      "step": 2909
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4625436067581177,
      "learning_rate": 1.87978481802199e-05,
      "loss": 0.647,
      "step": 2910
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.480860710144043,
      "learning_rate": 1.8796879686982547e-05,
      "loss": 0.5876,
      "step": 2911
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3979650735855103,
      "learning_rate": 1.8795910828747035e-05,
      "loss": 0.5759,
      "step": 2912
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5042881965637207,
      "learning_rate": 1.8794941605553555e-05,
      "loss": 0.5973,
      "step": 2913
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.472637414932251,
      "learning_rate": 1.879397201744233e-05,
      "loss": 0.5836,
      "step": 2914
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.557440161705017,
      "learning_rate": 1.8793002064453586e-05,
      "loss": 0.6002,
      "step": 2915
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.475391149520874,
      "learning_rate": 1.8792031746627563e-05,
      "loss": 0.6444,
      "step": 2916
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.446536898612976,
      "learning_rate": 1.879106106400453e-05,
      "loss": 0.6402,
      "step": 2917
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.521857738494873,
      "learning_rate": 1.879009001662476e-05,
      "loss": 0.5773,
      "step": 2918
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4363738298416138,
      "learning_rate": 1.878911860452854e-05,
      "loss": 0.5765,
      "step": 2919
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5111351013183594,
      "learning_rate": 1.8788146827756178e-05,
      "loss": 0.5726,
      "step": 2920
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4527709484100342,
      "learning_rate": 1.8787174686347996e-05,
      "loss": 0.6163,
      "step": 2921
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3309900760650635,
      "learning_rate": 1.8786202180344328e-05,
      "loss": 0.5783,
      "step": 2922
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5192902088165283,
      "learning_rate": 1.8785229309785523e-05,
      "loss": 0.5488,
      "step": 2923
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.39121413230896,
      "learning_rate": 1.8784256074711954e-05,
      "loss": 0.5679,
      "step": 2924
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5094727277755737,
      "learning_rate": 1.8783282475163997e-05,
      "loss": 0.6003,
      "step": 2925
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.404447317123413,
      "learning_rate": 1.878230851118205e-05,
      "loss": 0.5554,
      "step": 2926
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3795958757400513,
      "learning_rate": 1.878133418280652e-05,
      "loss": 0.6056,
      "step": 2927
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4641469717025757,
      "learning_rate": 1.878035949007784e-05,
      "loss": 0.5949,
      "step": 2928
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.380627155303955,
      "learning_rate": 1.8779384433036448e-05,
      "loss": 0.5943,
      "step": 2929
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4634854793548584,
      "learning_rate": 1.87784090117228e-05,
      "loss": 0.6037,
      "step": 2930
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4466478824615479,
      "learning_rate": 1.877743322617737e-05,
      "loss": 0.5603,
      "step": 2931
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3864800930023193,
      "learning_rate": 1.8776457076440652e-05,
      "loss": 0.5641,
      "step": 2932
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4228886365890503,
      "learning_rate": 1.8775480562553136e-05,
      "loss": 0.6335,
      "step": 2933
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4812748432159424,
      "learning_rate": 1.8774503684555344e-05,
      "loss": 0.5571,
      "step": 2934
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3475285768508911,
      "learning_rate": 1.877352644248781e-05,
      "loss": 0.5898,
      "step": 2935
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5155490636825562,
      "learning_rate": 1.877254883639108e-05,
      "loss": 0.6038,
      "step": 2936
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7589415311813354,
      "learning_rate": 1.8771570866305714e-05,
      "loss": 1.0273,
      "step": 2937
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.564674973487854,
      "learning_rate": 1.8770592532272296e-05,
      "loss": 0.619,
      "step": 2938
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5216808319091797,
      "learning_rate": 1.8769613834331413e-05,
      "loss": 0.6452,
      "step": 2939
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4858628511428833,
      "learning_rate": 1.8768634772523682e-05,
      "loss": 0.5503,
      "step": 2940
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4481964111328125,
      "learning_rate": 1.8767655346889713e-05,
      "loss": 0.5876,
      "step": 2941
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.461822509765625,
      "learning_rate": 1.8766675557470153e-05,
      "loss": 0.5628,
      "step": 2942
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5009303092956543,
      "learning_rate": 1.8765695404305653e-05,
      "loss": 0.5759,
      "step": 2943
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3968383073806763,
      "learning_rate": 1.8764714887436878e-05,
      "loss": 0.5835,
      "step": 2944
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6111161708831787,
      "learning_rate": 1.8763734006904515e-05,
      "loss": 0.5754,
      "step": 2945
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3639906644821167,
      "learning_rate": 1.8762752762749267e-05,
      "loss": 0.5891,
      "step": 2946
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4527472257614136,
      "learning_rate": 1.8761771155011838e-05,
      "loss": 0.6072,
      "step": 2947
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4202121496200562,
      "learning_rate": 1.8760789183732964e-05,
      "loss": 0.6377,
      "step": 2948
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3556159734725952,
      "learning_rate": 1.8759806848953384e-05,
      "loss": 0.5908,
      "step": 2949
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3853901624679565,
      "learning_rate": 1.8758824150713856e-05,
      "loss": 0.5718,
      "step": 2950
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5237056016921997,
      "learning_rate": 1.875784108905516e-05,
      "loss": 0.5763,
      "step": 2951
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4407633543014526,
      "learning_rate": 1.875685766401808e-05,
      "loss": 0.6011,
      "step": 2952
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3553041219711304,
      "learning_rate": 1.8755873875643424e-05,
      "loss": 0.577,
      "step": 2953
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3556371927261353,
      "learning_rate": 1.8754889723972008e-05,
      "loss": 0.5915,
      "step": 2954
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.454453706741333,
      "learning_rate": 1.8753905209044666e-05,
      "loss": 0.5531,
      "step": 2955
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4630941152572632,
      "learning_rate": 1.875292033090225e-05,
      "loss": 0.5731,
      "step": 2956
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.527828574180603,
      "learning_rate": 1.875193508958562e-05,
      "loss": 0.6076,
      "step": 2957
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4449779987335205,
      "learning_rate": 1.875094948513566e-05,
      "loss": 0.6078,
      "step": 2958
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3419374227523804,
      "learning_rate": 1.874996351759326e-05,
      "loss": 0.5782,
      "step": 2959
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4446860551834106,
      "learning_rate": 1.874897718699933e-05,
      "loss": 0.6466,
      "step": 2960
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4762914180755615,
      "learning_rate": 1.87479904933948e-05,
      "loss": 0.5771,
      "step": 2961
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4690254926681519,
      "learning_rate": 1.8747003436820607e-05,
      "loss": 0.5604,
      "step": 2962
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5156041383743286,
      "learning_rate": 1.8746016017317704e-05,
      "loss": 0.6012,
      "step": 2963
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.556293249130249,
      "learning_rate": 1.8745028234927062e-05,
      "loss": 0.6301,
      "step": 2964
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3913450241088867,
      "learning_rate": 1.8744040089689664e-05,
      "loss": 0.5984,
      "step": 2965
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5069531202316284,
      "learning_rate": 1.8743051581646513e-05,
      "loss": 0.5217,
      "step": 2966
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6002635955810547,
      "learning_rate": 1.8742062710838622e-05,
      "loss": 0.5979,
      "step": 2967
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4933247566223145,
      "learning_rate": 1.8741073477307025e-05,
      "loss": 0.5885,
      "step": 2968
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.450362205505371,
      "learning_rate": 1.874008388109276e-05,
      "loss": 0.6025,
      "step": 2969
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5832256078720093,
      "learning_rate": 1.8739093922236892e-05,
      "loss": 0.5486,
      "step": 2970
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.458146333694458,
      "learning_rate": 1.873810360078049e-05,
      "loss": 0.587,
      "step": 2971
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4816715717315674,
      "learning_rate": 1.8737112916764653e-05,
      "loss": 0.5345,
      "step": 2972
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4587608575820923,
      "learning_rate": 1.8736121870230485e-05,
      "loss": 0.5871,
      "step": 2973
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3681056499481201,
      "learning_rate": 1.87351304612191e-05,
      "loss": 0.5985,
      "step": 2974
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3836642503738403,
      "learning_rate": 1.8734138689771636e-05,
      "loss": 0.5818,
      "step": 2975
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4537466764450073,
      "learning_rate": 1.8733146555929246e-05,
      "loss": 0.6128,
      "step": 2976
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5145599842071533,
      "learning_rate": 1.8732154059733093e-05,
      "loss": 0.5865,
      "step": 2977
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.474785566329956,
      "learning_rate": 1.8731161201224357e-05,
      "loss": 0.6096,
      "step": 2978
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4906564950942993,
      "learning_rate": 1.8730167980444237e-05,
      "loss": 0.5913,
      "step": 2979
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4016013145446777,
      "learning_rate": 1.872917439743394e-05,
      "loss": 0.5739,
      "step": 2980
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.494085431098938,
      "learning_rate": 1.8728180452234692e-05,
      "loss": 0.6076,
      "step": 2981
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.548069715499878,
      "learning_rate": 1.872718614488773e-05,
      "loss": 0.6397,
      "step": 2982
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.450762152671814,
      "learning_rate": 1.872619147543432e-05,
      "loss": 0.5832,
      "step": 2983
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7585077285766602,
      "learning_rate": 1.8725196443915725e-05,
      "loss": 0.9829,
      "step": 2984
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7519336938858032,
      "learning_rate": 1.872420105037323e-05,
      "loss": 0.6105,
      "step": 2985
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6608879566192627,
      "learning_rate": 1.8723205294848138e-05,
      "loss": 0.5706,
      "step": 2986
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4498106241226196,
      "learning_rate": 1.8722209177381763e-05,
      "loss": 0.5709,
      "step": 2987
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4934988021850586,
      "learning_rate": 1.872121269801544e-05,
      "loss": 0.5813,
      "step": 2988
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5457170009613037,
      "learning_rate": 1.8720215856790508e-05,
      "loss": 0.5902,
      "step": 2989
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3576322793960571,
      "learning_rate": 1.8719218653748338e-05,
      "loss": 0.5781,
      "step": 2990
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5467414855957031,
      "learning_rate": 1.871822108893029e-05,
      "loss": 0.5549,
      "step": 2991
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4220728874206543,
      "learning_rate": 1.8717223162377768e-05,
      "loss": 0.5953,
      "step": 2992
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4027515649795532,
      "learning_rate": 1.8716224874132175e-05,
      "loss": 0.656,
      "step": 2993
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4180636405944824,
      "learning_rate": 1.8715226224234928e-05,
      "loss": 0.5802,
      "step": 2994
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4935543537139893,
      "learning_rate": 1.8714227212727464e-05,
      "loss": 1.0023,
      "step": 2995
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.603169322013855,
      "learning_rate": 1.8713227839651238e-05,
      "loss": 0.6144,
      "step": 2996
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5564048290252686,
      "learning_rate": 1.8712228105047708e-05,
      "loss": 0.5859,
      "step": 2997
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4539369344711304,
      "learning_rate": 1.8711228008958364e-05,
      "loss": 0.573,
      "step": 2998
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4615081548690796,
      "learning_rate": 1.8710227551424696e-05,
      "loss": 0.5962,
      "step": 2999
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5064613819122314,
      "learning_rate": 1.8709226732488216e-05,
      "loss": 0.5852,
      "step": 3000
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9486846923828125,
      "learning_rate": 1.9783700701096643e-05,
      "loss": 0.5701,
      "step": 3001
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8415573835372925,
      "learning_rate": 1.9783489979639053e-05,
      "loss": 0.5017,
      "step": 3002
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.039788007736206,
      "learning_rate": 1.9783279156711022e-05,
      "loss": 0.5783,
      "step": 3003
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9317278861999512,
      "learning_rate": 1.9783068232314747e-05,
      "loss": 0.4449,
      "step": 3004
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0086398124694824,
      "learning_rate": 1.9782857206452404e-05,
      "loss": 0.6001,
      "step": 3005
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1085941791534424,
      "learning_rate": 1.9782646079126194e-05,
      "loss": 0.5434,
      "step": 3006
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0043272972106934,
      "learning_rate": 1.9782434850338295e-05,
      "loss": 0.5185,
      "step": 3007
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9883149862289429,
      "learning_rate": 1.978222352009091e-05,
      "loss": 0.4492,
      "step": 3008
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.024111270904541,
      "learning_rate": 1.978201208838622e-05,
      "loss": 0.5931,
      "step": 3009
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9496465921401978,
      "learning_rate": 1.9781800555226422e-05,
      "loss": 0.4792,
      "step": 3010
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9437496662139893,
      "learning_rate": 1.9781588920613714e-05,
      "loss": 0.5777,
      "step": 3011
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9621328115463257,
      "learning_rate": 1.978137718455029e-05,
      "loss": 0.5018,
      "step": 3012
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0271778106689453,
      "learning_rate": 1.9781165347038334e-05,
      "loss": 0.8265,
      "step": 3013
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8976458311080933,
      "learning_rate": 1.9780953408080058e-05,
      "loss": 0.8978,
      "step": 3014
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5630037784576416,
      "learning_rate": 1.9780741367677657e-05,
      "loss": 0.5499,
      "step": 3015
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4783365726470947,
      "learning_rate": 1.9780529225833323e-05,
      "loss": 0.5151,
      "step": 3016
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0338199138641357,
      "learning_rate": 1.9780316982549262e-05,
      "loss": 0.5917,
      "step": 3017
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1676025390625,
      "learning_rate": 1.9780104637827676e-05,
      "loss": 0.5209,
      "step": 3018
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.052696943283081,
      "learning_rate": 1.9779892191670767e-05,
      "loss": 0.5343,
      "step": 3019
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1881208419799805,
      "learning_rate": 1.9779679644080735e-05,
      "loss": 0.6158,
      "step": 3020
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.09500789642334,
      "learning_rate": 1.9779466995059788e-05,
      "loss": 0.6505,
      "step": 3021
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9423387050628662,
      "learning_rate": 1.9779254244610128e-05,
      "loss": 0.4716,
      "step": 3022
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2151150703430176,
      "learning_rate": 1.9779041392733962e-05,
      "loss": 0.604,
      "step": 3023
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.050572633743286,
      "learning_rate": 1.97788284394335e-05,
      "loss": 0.4647,
      "step": 3024
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.0202603340148926,
      "learning_rate": 1.9778615384710955e-05,
      "loss": 0.7503,
      "step": 3025
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.9447968006134033,
      "learning_rate": 1.9778402228568526e-05,
      "loss": 0.8879,
      "step": 3026
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.4162116050720215,
      "learning_rate": 1.9778188971008434e-05,
      "loss": 0.6358,
      "step": 3027
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3106257915496826,
      "learning_rate": 1.9777975612032886e-05,
      "loss": 0.6134,
      "step": 3028
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.149956703186035,
      "learning_rate": 1.977776215164409e-05,
      "loss": 0.6264,
      "step": 3029
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.05026912689209,
      "learning_rate": 1.977754858984427e-05,
      "loss": 0.5548,
      "step": 3030
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0810461044311523,
      "learning_rate": 1.9777334926635637e-05,
      "loss": 0.5739,
      "step": 3031
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8861199617385864,
      "learning_rate": 1.97771211620204e-05,
      "loss": 0.5293,
      "step": 3032
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9472615718841553,
      "learning_rate": 1.9776907296000786e-05,
      "loss": 0.6441,
      "step": 3033
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9022804498672485,
      "learning_rate": 1.9776693328579012e-05,
      "loss": 0.4996,
      "step": 3034
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9976351261138916,
      "learning_rate": 1.977647925975729e-05,
      "loss": 0.6315,
      "step": 3035
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.968045711517334,
      "learning_rate": 1.9776265089537844e-05,
      "loss": 0.4812,
      "step": 3036
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.536545515060425,
      "learning_rate": 1.97760508179229e-05,
      "loss": 0.8204,
      "step": 3037
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.9894649982452393,
      "learning_rate": 1.9775836444914674e-05,
      "loss": 0.7644,
      "step": 3038
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.9178929328918457,
      "learning_rate": 1.977562197051539e-05,
      "loss": 0.5886,
      "step": 3039
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.6241791248321533,
      "learning_rate": 1.977540739472728e-05,
      "loss": 0.5087,
      "step": 3040
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.448291301727295,
      "learning_rate": 1.9775192717552558e-05,
      "loss": 0.6944,
      "step": 3041
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0410053730010986,
      "learning_rate": 1.9774977938993458e-05,
      "loss": 0.4875,
      "step": 3042
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.155322551727295,
      "learning_rate": 1.9774763059052208e-05,
      "loss": 0.6252,
      "step": 3043
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.222175121307373,
      "learning_rate": 1.977454807773103e-05,
      "loss": 0.4871,
      "step": 3044
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.26008939743042,
      "learning_rate": 1.977433299503216e-05,
      "loss": 0.6783,
      "step": 3045
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.023695945739746,
      "learning_rate": 1.977411781095783e-05,
      "loss": 0.5024,
      "step": 3046
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.029694080352783,
      "learning_rate": 1.977390252551026e-05,
      "loss": 0.565,
      "step": 3047
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.022886037826538,
      "learning_rate": 1.9773687138691702e-05,
      "loss": 0.5662,
      "step": 3048
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0591177940368652,
      "learning_rate": 1.9773471650504373e-05,
      "loss": 0.5789,
      "step": 3049
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.24352765083313,
      "learning_rate": 1.9773256060950518e-05,
      "loss": 0.6371,
      "step": 3050
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0890283584594727,
      "learning_rate": 1.977304037003237e-05,
      "loss": 0.6221,
      "step": 3051
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1701440811157227,
      "learning_rate": 1.977282457775216e-05,
      "loss": 0.4904,
      "step": 3052
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0757358074188232,
      "learning_rate": 1.9772608684112136e-05,
      "loss": 0.4905,
      "step": 3053
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.882857084274292,
      "learning_rate": 1.9772392689114532e-05,
      "loss": 0.5935,
      "step": 3054
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0416126251220703,
      "learning_rate": 1.9772176592761586e-05,
      "loss": 0.5836,
      "step": 3055
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9908473491668701,
      "learning_rate": 1.9771960395055547e-05,
      "loss": 0.5127,
      "step": 3056
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.028080940246582,
      "learning_rate": 1.977174409599865e-05,
      "loss": 0.5523,
      "step": 3057
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1780402660369873,
      "learning_rate": 1.9771527695593142e-05,
      "loss": 0.626,
      "step": 3058
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.259219169616699,
      "learning_rate": 1.9771311193841266e-05,
      "loss": 0.6747,
      "step": 3059
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.052616596221924,
      "learning_rate": 1.9771094590745265e-05,
      "loss": 0.4909,
      "step": 3060
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0267131328582764,
      "learning_rate": 1.9770877886307393e-05,
      "loss": 0.6314,
      "step": 3061
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.101393938064575,
      "learning_rate": 1.9770661080529893e-05,
      "loss": 0.551,
      "step": 3062
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9301049709320068,
      "learning_rate": 1.977044417341501e-05,
      "loss": 0.5361,
      "step": 3063
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9783865213394165,
      "learning_rate": 1.9770227164964996e-05,
      "loss": 0.5521,
      "step": 3064
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.032505750656128,
      "learning_rate": 1.9770010055182105e-05,
      "loss": 0.556,
      "step": 3065
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1889500617980957,
      "learning_rate": 1.9769792844068586e-05,
      "loss": 0.5386,
      "step": 3066
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0877084732055664,
      "learning_rate": 1.9769575531626697e-05,
      "loss": 0.6381,
      "step": 3067
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.960108995437622,
      "learning_rate": 1.976935811785868e-05,
      "loss": 0.4992,
      "step": 3068
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0623819828033447,
      "learning_rate": 1.9769140602766805e-05,
      "loss": 0.5939,
      "step": 3069
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1530582904815674,
      "learning_rate": 1.9768922986353316e-05,
      "loss": 0.5301,
      "step": 3070
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9341779947280884,
      "learning_rate": 1.9768705268620473e-05,
      "loss": 0.5247,
      "step": 3071
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9057590961456299,
      "learning_rate": 1.976848744957054e-05,
      "loss": 0.5152,
      "step": 3072
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2188916206359863,
      "learning_rate": 1.976826952920577e-05,
      "loss": 0.4948,
      "step": 3073
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.032231330871582,
      "learning_rate": 1.976805150752842e-05,
      "loss": 0.4907,
      "step": 3074
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0079731941223145,
      "learning_rate": 1.976783338454076e-05,
      "loss": 0.5895,
      "step": 3075
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.019284963607788,
      "learning_rate": 1.976761516024505e-05,
      "loss": 0.4841,
      "step": 3076
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0893468856811523,
      "learning_rate": 1.9767396834643548e-05,
      "loss": 0.6349,
      "step": 3077
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9190330505371094,
      "learning_rate": 1.9767178407738522e-05,
      "loss": 0.5245,
      "step": 3078
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8968101739883423,
      "learning_rate": 1.976695987953224e-05,
      "loss": 0.5779,
      "step": 3079
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8494163751602173,
      "learning_rate": 1.9766741250026966e-05,
      "loss": 0.4835,
      "step": 3080
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.044034481048584,
      "learning_rate": 1.9766522519224964e-05,
      "loss": 0.5553,
      "step": 3081
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.965625286102295,
      "learning_rate": 1.9766303687128508e-05,
      "loss": 0.5229,
      "step": 3082
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.561919689178467,
      "learning_rate": 1.9766084753739868e-05,
      "loss": 0.8432,
      "step": 3083
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.238224744796753,
      "learning_rate": 1.9765865719061306e-05,
      "loss": 0.9261,
      "step": 3084
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3292274475097656,
      "learning_rate": 1.9765646583095108e-05,
      "loss": 0.9358,
      "step": 3085
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6828519105911255,
      "learning_rate": 1.976542734584353e-05,
      "loss": 0.8399,
      "step": 3086
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.2838377952575684,
      "learning_rate": 1.976520800730886e-05,
      "loss": 0.638,
      "step": 3087
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.955641269683838,
      "learning_rate": 1.9764988567493367e-05,
      "loss": 0.5131,
      "step": 3088
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.262708902359009,
      "learning_rate": 1.976476902639933e-05,
      "loss": 0.6408,
      "step": 3089
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3442745208740234,
      "learning_rate": 1.9764549384029022e-05,
      "loss": 0.5963,
      "step": 3090
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.273639678955078,
      "learning_rate": 1.9764329640384724e-05,
      "loss": 0.6474,
      "step": 3091
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.154961347579956,
      "learning_rate": 1.976410979546871e-05,
      "loss": 0.5348,
      "step": 3092
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2107763290405273,
      "learning_rate": 1.9763889849283265e-05,
      "loss": 0.6641,
      "step": 3093
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.143245220184326,
      "learning_rate": 1.976366980183067e-05,
      "loss": 0.4833,
      "step": 3094
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.47251558303833,
      "learning_rate": 1.9763449653113206e-05,
      "loss": 0.6834,
      "step": 3095
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.119821071624756,
      "learning_rate": 1.9763229403133154e-05,
      "loss": 0.5328,
      "step": 3096
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.049665689468384,
      "learning_rate": 1.9763009051892804e-05,
      "loss": 0.6098,
      "step": 3097
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.155862808227539,
      "learning_rate": 1.976278859939444e-05,
      "loss": 0.5706,
      "step": 3098
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2245266437530518,
      "learning_rate": 1.9762568045640343e-05,
      "loss": 0.6449,
      "step": 3099
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9731483459472656,
      "learning_rate": 1.9762347390632802e-05,
      "loss": 0.4888,
      "step": 3100
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.016911506652832,
      "learning_rate": 1.9762126634374114e-05,
      "loss": 0.5595,
      "step": 3101
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0926828384399414,
      "learning_rate": 1.976190577686656e-05,
      "loss": 0.5599,
      "step": 3102
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0480008125305176,
      "learning_rate": 1.976168481811243e-05,
      "loss": 0.5509,
      "step": 3103
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9016286134719849,
      "learning_rate": 1.9761463758114023e-05,
      "loss": 0.5236,
      "step": 3104
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9315251111984253,
      "learning_rate": 1.9761242596873623e-05,
      "loss": 0.6526,
      "step": 3105
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9839003086090088,
      "learning_rate": 1.9761021334393534e-05,
      "loss": 0.5173,
      "step": 3106
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.011167526245117,
      "learning_rate": 1.976079997067604e-05,
      "loss": 0.6234,
      "step": 3107
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9074289798736572,
      "learning_rate": 1.9760578505723445e-05,
      "loss": 0.4948,
      "step": 3108
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9964786767959595,
      "learning_rate": 1.9760356939538043e-05,
      "loss": 0.6233,
      "step": 3109
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9969182014465332,
      "learning_rate": 1.976013527212213e-05,
      "loss": 0.49,
      "step": 3110
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1730926036834717,
      "learning_rate": 1.9759913503478006e-05,
      "loss": 0.6342,
      "step": 3111
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9516793489456177,
      "learning_rate": 1.9759691633607975e-05,
      "loss": 0.4881,
      "step": 3112
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9561833143234253,
      "learning_rate": 1.9759469662514336e-05,
      "loss": 0.5324,
      "step": 3113
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.924721598625183,
      "learning_rate": 1.975924759019939e-05,
      "loss": 0.5588,
      "step": 3114
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9833886623382568,
      "learning_rate": 1.9759025416665437e-05,
      "loss": 0.6321,
      "step": 3115
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9120454788208008,
      "learning_rate": 1.975880314191479e-05,
      "loss": 0.4793,
      "step": 3116
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.867318034172058,
      "learning_rate": 1.9758580765949745e-05,
      "loss": 0.5612,
      "step": 3117
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0777475833892822,
      "learning_rate": 1.9758358288772616e-05,
      "loss": 0.5427,
      "step": 3118
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.040300130844116,
      "learning_rate": 1.9758135710385705e-05,
      "loss": 0.6407,
      "step": 3119
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.13492751121521,
      "learning_rate": 1.9757913030791323e-05,
      "loss": 0.4737,
      "step": 3120
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9300183057785034,
      "learning_rate": 1.9757690249991778e-05,
      "loss": 0.625,
      "step": 3121
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0722286701202393,
      "learning_rate": 1.9757467367989386e-05,
      "loss": 0.5095,
      "step": 3122
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0526652336120605,
      "learning_rate": 1.9757244384786446e-05,
      "loss": 0.5613,
      "step": 3123
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.033536911010742,
      "learning_rate": 1.9757021300385288e-05,
      "loss": 0.6345,
      "step": 3124
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.105868339538574,
      "learning_rate": 1.9756798114788212e-05,
      "loss": 0.6307,
      "step": 3125
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.967597484588623,
      "learning_rate": 1.9756574827997537e-05,
      "loss": 0.5364,
      "step": 3126
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0382144451141357,
      "learning_rate": 1.9756351440015583e-05,
      "loss": 0.5683,
      "step": 3127
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8639848232269287,
      "learning_rate": 1.9756127950844664e-05,
      "loss": 0.506,
      "step": 3128
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9285972118377686,
      "learning_rate": 1.975590436048709e-05,
      "loss": 0.5773,
      "step": 3129
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0570051670074463,
      "learning_rate": 1.9755680668945195e-05,
      "loss": 0.516,
      "step": 3130
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9181872606277466,
      "learning_rate": 1.975545687622129e-05,
      "loss": 0.5934,
      "step": 3131
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9350491762161255,
      "learning_rate": 1.9755232982317696e-05,
      "loss": 0.5097,
      "step": 3132
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.043841600418091,
      "learning_rate": 1.9755008987236736e-05,
      "loss": 0.6054,
      "step": 3133
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.912093162536621,
      "learning_rate": 1.9754784890980736e-05,
      "loss": 0.5259,
      "step": 3134
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1258184909820557,
      "learning_rate": 1.9754560693552017e-05,
      "loss": 0.6824,
      "step": 3135
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8917001485824585,
      "learning_rate": 1.9754336394952903e-05,
      "loss": 0.486,
      "step": 3136
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.919644832611084,
      "learning_rate": 1.9754111995185726e-05,
      "loss": 0.5921,
      "step": 3137
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9889439344406128,
      "learning_rate": 1.9753887494252807e-05,
      "loss": 0.515,
      "step": 3138
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1391549110412598,
      "learning_rate": 1.975366289215648e-05,
      "loss": 0.6267,
      "step": 3139
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8956754207611084,
      "learning_rate": 1.975343818889907e-05,
      "loss": 0.5051,
      "step": 3140
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.007573127746582,
      "learning_rate": 1.975321338448291e-05,
      "loss": 0.6456,
      "step": 3141
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9000214338302612,
      "learning_rate": 1.975298847891033e-05,
      "loss": 0.4751,
      "step": 3142
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.088585376739502,
      "learning_rate": 1.9752763472183667e-05,
      "loss": 0.6138,
      "step": 3143
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9884523153305054,
      "learning_rate": 1.975253836430525e-05,
      "loss": 0.4848,
      "step": 3144
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.94326651096344,
      "learning_rate": 1.9752313155277415e-05,
      "loss": 0.5548,
      "step": 3145
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9895414113998413,
      "learning_rate": 1.9752087845102496e-05,
      "loss": 0.5512,
      "step": 3146
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.065871000289917,
      "learning_rate": 1.9751862433782834e-05,
      "loss": 0.6656,
      "step": 3147
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.970832347869873,
      "learning_rate": 1.975163692132076e-05,
      "loss": 0.4957,
      "step": 3148
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.049541473388672,
      "learning_rate": 1.9751411307718622e-05,
      "loss": 0.5489,
      "step": 3149
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9981498718261719,
      "learning_rate": 1.9751185592978754e-05,
      "loss": 0.564,
      "step": 3150
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.141230821609497,
      "learning_rate": 1.97509597771035e-05,
      "loss": 0.6841,
      "step": 3151
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9387648105621338,
      "learning_rate": 1.9750733860095196e-05,
      "loss": 0.4757,
      "step": 3152
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9700908660888672,
      "learning_rate": 1.9750507841956197e-05,
      "loss": 0.5316,
      "step": 3153
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1162054538726807,
      "learning_rate": 1.9750281722688833e-05,
      "loss": 0.5137,
      "step": 3154
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0120534896850586,
      "learning_rate": 1.975005550229546e-05,
      "loss": 0.6896,
      "step": 3155
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0517213344573975,
      "learning_rate": 1.9749829180778418e-05,
      "loss": 0.4973,
      "step": 3156
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.052084445953369,
      "learning_rate": 1.974960275814006e-05,
      "loss": 0.5726,
      "step": 3157
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.073915958404541,
      "learning_rate": 1.974937623438273e-05,
      "loss": 0.6087,
      "step": 3158
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0715348720550537,
      "learning_rate": 1.9749149609508772e-05,
      "loss": 0.5883,
      "step": 3159
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0734333992004395,
      "learning_rate": 1.974892288352055e-05,
      "loss": 0.6072,
      "step": 3160
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0059447288513184,
      "learning_rate": 1.9748696056420406e-05,
      "loss": 0.583,
      "step": 3161
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1606574058532715,
      "learning_rate": 1.9748469128210696e-05,
      "loss": 0.5634,
      "step": 3162
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.910947322845459,
      "learning_rate": 1.974824209889377e-05,
      "loss": 0.5944,
      "step": 3163
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0244686603546143,
      "learning_rate": 1.974801496847199e-05,
      "loss": 0.4996,
      "step": 3164
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1784355640411377,
      "learning_rate": 1.9747787736947702e-05,
      "loss": 0.5515,
      "step": 3165
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9957174062728882,
      "learning_rate": 1.974756040432327e-05,
      "loss": 0.5403,
      "step": 3166
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.071347236633301,
      "learning_rate": 1.974733297060105e-05,
      "loss": 0.5897,
      "step": 3167
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.021423101425171,
      "learning_rate": 1.9747105435783398e-05,
      "loss": 0.5032,
      "step": 3168
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0430400371551514,
      "learning_rate": 1.974687779987268e-05,
      "loss": 0.6216,
      "step": 3169
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.039310932159424,
      "learning_rate": 1.974665006287125e-05,
      "loss": 0.4892,
      "step": 3170
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0124099254608154,
      "learning_rate": 1.9746422224781475e-05,
      "loss": 0.5832,
      "step": 3171
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9965643882751465,
      "learning_rate": 1.9746194285605714e-05,
      "loss": 0.5605,
      "step": 3172
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1062514781951904,
      "learning_rate": 1.9745966245346336e-05,
      "loss": 0.6143,
      "step": 3173
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.98531973361969,
      "learning_rate": 1.9745738104005702e-05,
      "loss": 0.5386,
      "step": 3174
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0554120540618896,
      "learning_rate": 1.974550986158618e-05,
      "loss": 0.6162,
      "step": 3175
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.059623956680298,
      "learning_rate": 1.9745281518090135e-05,
      "loss": 0.5907,
      "step": 3176
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0806775093078613,
      "learning_rate": 1.974505307351994e-05,
      "loss": 0.6048,
      "step": 3177
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.078282117843628,
      "learning_rate": 1.974482452787796e-05,
      "loss": 0.4836,
      "step": 3178
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0403895378112793,
      "learning_rate": 1.9744595881166563e-05,
      "loss": 0.5898,
      "step": 3179
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9889564514160156,
      "learning_rate": 1.974436713338813e-05,
      "loss": 0.4796,
      "step": 3180
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.218933343887329,
      "learning_rate": 1.974413828454503e-05,
      "loss": 0.5721,
      "step": 3181
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0353801250457764,
      "learning_rate": 1.9743909334639628e-05,
      "loss": 0.5679,
      "step": 3182
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1262500286102295,
      "learning_rate": 1.9743680283674306e-05,
      "loss": 0.7359,
      "step": 3183
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.921964168548584,
      "learning_rate": 1.9743451131651442e-05,
      "loss": 0.5011,
      "step": 3184
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9440196752548218,
      "learning_rate": 1.9743221878573403e-05,
      "loss": 0.6193,
      "step": 3185
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8619022369384766,
      "learning_rate": 1.974299252444258e-05,
      "loss": 0.4673,
      "step": 3186
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.021090269088745,
      "learning_rate": 1.974276306926134e-05,
      "loss": 0.708,
      "step": 3187
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0602877140045166,
      "learning_rate": 1.974253351303207e-05,
      "loss": 0.5874,
      "step": 3188
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.045295476913452,
      "learning_rate": 1.9742303855757147e-05,
      "loss": 0.5685,
      "step": 3189
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.896843433380127,
      "learning_rate": 1.9742074097438954e-05,
      "loss": 0.499,
      "step": 3190
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9152251482009888,
      "learning_rate": 1.9741844238079875e-05,
      "loss": 0.5944,
      "step": 3191
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.956850528717041,
      "learning_rate": 1.9741614277682295e-05,
      "loss": 0.5425,
      "step": 3192
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8893026113510132,
      "learning_rate": 1.9741384216248596e-05,
      "loss": 0.5699,
      "step": 3193
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.993003010749817,
      "learning_rate": 1.9741154053781165e-05,
      "loss": 0.5054,
      "step": 3194
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.687153339385986,
      "learning_rate": 1.974092379028239e-05,
      "loss": 1.0553,
      "step": 3195
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.140742301940918,
      "learning_rate": 1.9740693425754656e-05,
      "loss": 1.0236,
      "step": 3196
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2729103565216064,
      "learning_rate": 1.974046296020036e-05,
      "loss": 0.6605,
      "step": 3197
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.14296555519104,
      "learning_rate": 1.9740232393621884e-05,
      "loss": 0.4815,
      "step": 3198
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9734432697296143,
      "learning_rate": 1.9740001726021623e-05,
      "loss": 0.5833,
      "step": 3199
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.026477575302124,
      "learning_rate": 1.973977095740197e-05,
      "loss": 0.6012,
      "step": 3200
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1004486083984375,
      "learning_rate": 1.9739540087765318e-05,
      "loss": 0.5604,
      "step": 3201
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9796149730682373,
      "learning_rate": 1.973930911711406e-05,
      "loss": 0.5553,
      "step": 3202
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8882805109024048,
      "learning_rate": 1.9739078045450587e-05,
      "loss": 0.6279,
      "step": 3203
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8899712562561035,
      "learning_rate": 1.9738846872777308e-05,
      "loss": 0.4611,
      "step": 3204
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0861432552337646,
      "learning_rate": 1.9738615599096614e-05,
      "loss": 0.6209,
      "step": 3205
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1023404598236084,
      "learning_rate": 1.97383842244109e-05,
      "loss": 0.617,
      "step": 3206
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9749491214752197,
      "learning_rate": 1.9738152748722572e-05,
      "loss": 0.5643,
      "step": 3207
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1768367290496826,
      "learning_rate": 1.973792117203403e-05,
      "loss": 0.5512,
      "step": 3208
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.066460609436035,
      "learning_rate": 1.9737689494347668e-05,
      "loss": 0.5692,
      "step": 3209
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9051676988601685,
      "learning_rate": 1.9737457715665894e-05,
      "loss": 0.5001,
      "step": 3210
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8994163274765015,
      "learning_rate": 1.9737225835991117e-05,
      "loss": 0.5766,
      "step": 3211
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0153157711029053,
      "learning_rate": 1.9736993855325734e-05,
      "loss": 0.4753,
      "step": 3212
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0866665840148926,
      "learning_rate": 1.973676177367216e-05,
      "loss": 0.6027,
      "step": 3213
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9824594259262085,
      "learning_rate": 1.9736529591032793e-05,
      "loss": 0.4817,
      "step": 3214
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.976349949836731,
      "learning_rate": 1.973629730741004e-05,
      "loss": 0.5807,
      "step": 3215
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9901670217514038,
      "learning_rate": 1.9736064922806323e-05,
      "loss": 0.517,
      "step": 3216
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8931759595870972,
      "learning_rate": 1.9735832437224037e-05,
      "loss": 0.5584,
      "step": 3217
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.902725100517273,
      "learning_rate": 1.9735599850665606e-05,
      "loss": 0.5241,
      "step": 3218
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9363903999328613,
      "learning_rate": 1.9735367163133433e-05,
      "loss": 0.5367,
      "step": 3219
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.001575231552124,
      "learning_rate": 1.9735134374629936e-05,
      "loss": 0.5425,
      "step": 3220
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.921150803565979,
      "learning_rate": 1.973490148515753e-05,
      "loss": 0.5862,
      "step": 3221
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9989943504333496,
      "learning_rate": 1.9734668494718624e-05,
      "loss": 0.4684,
      "step": 3222
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0393292903900146,
      "learning_rate": 1.9734435403315644e-05,
      "loss": 0.6976,
      "step": 3223
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8620727062225342,
      "learning_rate": 1.9734202210951e-05,
      "loss": 0.5014,
      "step": 3224
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0923898220062256,
      "learning_rate": 1.9733968917627116e-05,
      "loss": 0.5841,
      "step": 3225
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9424000978469849,
      "learning_rate": 1.9733735523346408e-05,
      "loss": 0.4881,
      "step": 3226
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9742484092712402,
      "learning_rate": 1.9733502028111294e-05,
      "loss": 0.5604,
      "step": 3227
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9663575887680054,
      "learning_rate": 1.9733268431924203e-05,
      "loss": 0.4877,
      "step": 3228
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9713321924209595,
      "learning_rate": 1.973303473478755e-05,
      "loss": 0.5913,
      "step": 3229
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9402337074279785,
      "learning_rate": 1.973280093670377e-05,
      "loss": 0.5001,
      "step": 3230
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1072211265563965,
      "learning_rate": 1.9732567037675278e-05,
      "loss": 0.5671,
      "step": 3231
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.981092095375061,
      "learning_rate": 1.9732333037704497e-05,
      "loss": 0.4571,
      "step": 3232
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.116938829421997,
      "learning_rate": 1.9732098936793864e-05,
      "loss": 0.5837,
      "step": 3233
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1600594520568848,
      "learning_rate": 1.9731864734945804e-05,
      "loss": 0.6041,
      "step": 3234
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0937435626983643,
      "learning_rate": 1.9731630432162743e-05,
      "loss": 0.5892,
      "step": 3235
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8990387916564941,
      "learning_rate": 1.9731396028447115e-05,
      "loss": 0.4807,
      "step": 3236
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.195044994354248,
      "learning_rate": 1.9731161523801344e-05,
      "loss": 0.6267,
      "step": 3237
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.100938558578491,
      "learning_rate": 1.973092691822787e-05,
      "loss": 0.6401,
      "step": 3238
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.608461618423462,
      "learning_rate": 1.9730692211729125e-05,
      "loss": 0.8761,
      "step": 3239
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2074928283691406,
      "learning_rate": 1.9730457404307538e-05,
      "loss": 0.8555,
      "step": 3240
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3060245513916016,
      "learning_rate": 1.9730222495965553e-05,
      "loss": 0.6546,
      "step": 3241
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.19736909866333,
      "learning_rate": 1.9729987486705596e-05,
      "loss": 0.4844,
      "step": 3242
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.212905168533325,
      "learning_rate": 1.972975237653011e-05,
      "loss": 0.7086,
      "step": 3243
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.05190110206604,
      "learning_rate": 1.9729517165441535e-05,
      "loss": 0.5579,
      "step": 3244
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3073153495788574,
      "learning_rate": 1.9729281853442312e-05,
      "loss": 0.6171,
      "step": 3245
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.01464581489563,
      "learning_rate": 1.9729046440534873e-05,
      "loss": 0.4791,
      "step": 3246
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1348025798797607,
      "learning_rate": 1.9728810926721665e-05,
      "loss": 0.7145,
      "step": 3247
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9551759958267212,
      "learning_rate": 1.9728575312005133e-05,
      "loss": 0.552,
      "step": 3248
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.123979330062866,
      "learning_rate": 1.972833959638772e-05,
      "loss": 0.5344,
      "step": 3249
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8291796445846558,
      "learning_rate": 1.9728103779871864e-05,
      "loss": 0.4724,
      "step": 3250
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.8873162269592285,
      "learning_rate": 1.972786786246002e-05,
      "loss": 0.5939,
      "step": 3251
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9723751544952393,
      "learning_rate": 1.972763184415463e-05,
      "loss": 0.4717,
      "step": 3252
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9343458414077759,
      "learning_rate": 1.9727395724958142e-05,
      "loss": 0.6118,
      "step": 3253
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0148632526397705,
      "learning_rate": 1.9727159504873007e-05,
      "loss": 0.5114,
      "step": 3254
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.992466688156128,
      "learning_rate": 1.9726923183901666e-05,
      "loss": 0.6611,
      "step": 3255
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9310030937194824,
      "learning_rate": 1.9726686762046585e-05,
      "loss": 0.545,
      "step": 3256
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9290529489517212,
      "learning_rate": 1.9726450239310208e-05,
      "loss": 0.6177,
      "step": 3257
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0077381134033203,
      "learning_rate": 1.9726213615694988e-05,
      "loss": 0.5496,
      "step": 3258
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0073916912078857,
      "learning_rate": 1.972597689120338e-05,
      "loss": 0.5884,
      "step": 3259
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9342612028121948,
      "learning_rate": 1.9725740065837834e-05,
      "loss": 0.4785,
      "step": 3260
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9731746912002563,
      "learning_rate": 1.9725503139600816e-05,
      "loss": 0.6299,
      "step": 3261
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9130996465682983,
      "learning_rate": 1.9725266112494777e-05,
      "loss": 0.4873,
      "step": 3262
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9505351781845093,
      "learning_rate": 1.9725028984522178e-05,
      "loss": 0.6765,
      "step": 3263
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.880428433418274,
      "learning_rate": 1.972479175568548e-05,
      "loss": 0.4685,
      "step": 3264
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0127272605895996,
      "learning_rate": 1.9724554425987135e-05,
      "loss": 0.5803,
      "step": 3265
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0168612003326416,
      "learning_rate": 1.9724316995429614e-05,
      "loss": 0.5619,
      "step": 3266
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.140887975692749,
      "learning_rate": 1.9724079464015374e-05,
      "loss": 0.7063,
      "step": 3267
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.081256151199341,
      "learning_rate": 1.972384183174688e-05,
      "loss": 0.4942,
      "step": 3268
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.959761619567871,
      "learning_rate": 1.9723604098626595e-05,
      "loss": 0.6113,
      "step": 3269
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8763900995254517,
      "learning_rate": 1.9723366264656995e-05,
      "loss": 0.4861,
      "step": 3270
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.959806203842163,
      "learning_rate": 1.972312832984053e-05,
      "loss": 0.5583,
      "step": 3271
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.141889810562134,
      "learning_rate": 1.972289029417968e-05,
      "loss": 0.5123,
      "step": 3272
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9511362314224243,
      "learning_rate": 1.972265215767691e-05,
      "loss": 0.5435,
      "step": 3273
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.317656993865967,
      "learning_rate": 1.972241392033469e-05,
      "loss": 0.4994,
      "step": 3274
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0234601497650146,
      "learning_rate": 1.9722175582155486e-05,
      "loss": 0.6068,
      "step": 3275
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9499385356903076,
      "learning_rate": 1.972193714314178e-05,
      "loss": 0.5504,
      "step": 3276
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9721012115478516,
      "learning_rate": 1.9721698603296038e-05,
      "loss": 0.594,
      "step": 3277
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.892053246498108,
      "learning_rate": 1.9721459962620736e-05,
      "loss": 0.4822,
      "step": 3278
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.5696632862091064,
      "learning_rate": 1.9721221221118348e-05,
      "loss": 0.5904,
      "step": 3279
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9558016061782837,
      "learning_rate": 1.9720982378791352e-05,
      "loss": 0.4786,
      "step": 3280
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.888043761253357,
      "learning_rate": 1.972074343564222e-05,
      "loss": 0.5811,
      "step": 3281
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.925316572189331,
      "learning_rate": 1.9720504391673442e-05,
      "loss": 0.5533,
      "step": 3282
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.052177667617798,
      "learning_rate": 1.9720265246887487e-05,
      "loss": 0.5451,
      "step": 3283
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8650739192962646,
      "learning_rate": 1.9720026001286835e-05,
      "loss": 0.5662,
      "step": 3284
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.845059633255005,
      "learning_rate": 1.971978665487397e-05,
      "loss": 0.9286,
      "step": 3285
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5528390407562256,
      "learning_rate": 1.9719547207651376e-05,
      "loss": 0.9079,
      "step": 3286
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.242771863937378,
      "learning_rate": 1.971930765962154e-05,
      "loss": 0.5746,
      "step": 3287
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2899329662323,
      "learning_rate": 1.9719068010786936e-05,
      "loss": 0.5919,
      "step": 3288
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0834484100341797,
      "learning_rate": 1.9718828261150054e-05,
      "loss": 0.5948,
      "step": 3289
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.923039197921753,
      "learning_rate": 1.9718588410713383e-05,
      "loss": 0.5259,
      "step": 3290
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9258517026901245,
      "learning_rate": 1.971834845947941e-05,
      "loss": 0.5531,
      "step": 3291
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.062692880630493,
      "learning_rate": 1.9718108407450626e-05,
      "loss": 0.6277,
      "step": 3292
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0450592041015625,
      "learning_rate": 1.9717868254629514e-05,
      "loss": 0.5874,
      "step": 3293
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0532314777374268,
      "learning_rate": 1.971762800101857e-05,
      "loss": 0.5312,
      "step": 3294
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.127838373184204,
      "learning_rate": 1.9717387646620283e-05,
      "loss": 0.6119,
      "step": 3295
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8995745182037354,
      "learning_rate": 1.971714719143715e-05,
      "loss": 0.4835,
      "step": 3296
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.130401134490967,
      "learning_rate": 1.971690663547166e-05,
      "loss": 0.5939,
      "step": 3297
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.112684488296509,
      "learning_rate": 1.971666597872631e-05,
      "loss": 0.604,
      "step": 3298
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8965733051300049,
      "learning_rate": 1.9716425221203598e-05,
      "loss": 0.5386,
      "step": 3299
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.020155191421509,
      "learning_rate": 1.9716184362906014e-05,
      "loss": 0.5916,
      "step": 3300
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0019521713256836,
      "learning_rate": 1.9715943403836063e-05,
      "loss": 0.5803,
      "step": 3301
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9695719480514526,
      "learning_rate": 1.9715702343996245e-05,
      "loss": 0.512,
      "step": 3302
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9328060150146484,
      "learning_rate": 1.9715461183389058e-05,
      "loss": 0.5506,
      "step": 3303
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8646650314331055,
      "learning_rate": 1.9715219922017002e-05,
      "loss": 0.4804,
      "step": 3304
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.09447979927063,
      "learning_rate": 1.9714978559882578e-05,
      "loss": 0.6059,
      "step": 3305
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0849344730377197,
      "learning_rate": 1.9714737096988292e-05,
      "loss": 0.494,
      "step": 3306
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9800459146499634,
      "learning_rate": 1.9714495533336652e-05,
      "loss": 0.6624,
      "step": 3307
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9856503009796143,
      "learning_rate": 1.971425386893015e-05,
      "loss": 0.5244,
      "step": 3308
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9929920434951782,
      "learning_rate": 1.9714012103771308e-05,
      "loss": 0.5978,
      "step": 3309
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8402103185653687,
      "learning_rate": 1.971377023786263e-05,
      "loss": 0.4981,
      "step": 3310
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9972115755081177,
      "learning_rate": 1.9713528271206614e-05,
      "loss": 0.5346,
      "step": 3311
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8369039297103882,
      "learning_rate": 1.9713286203805778e-05,
      "loss": 0.4708,
      "step": 3312
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1000123023986816,
      "learning_rate": 1.9713044035662638e-05,
      "loss": 0.6567,
      "step": 3313
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9790536165237427,
      "learning_rate": 1.9712801766779694e-05,
      "loss": 0.5372,
      "step": 3314
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8562794923782349,
      "learning_rate": 1.9712559397159467e-05,
      "loss": 0.5082,
      "step": 3315
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8774744272232056,
      "learning_rate": 1.9712316926804465e-05,
      "loss": 0.5166,
      "step": 3316
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9113540649414062,
      "learning_rate": 1.9712074355717208e-05,
      "loss": 0.5595,
      "step": 3317
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.018913507461548,
      "learning_rate": 1.971183168390021e-05,
      "loss": 0.6227,
      "step": 3318
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9593697786331177,
      "learning_rate": 1.9711588911355982e-05,
      "loss": 0.6063,
      "step": 3319
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9375789165496826,
      "learning_rate": 1.9711346038087055e-05,
      "loss": 0.6026,
      "step": 3320
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9431573152542114,
      "learning_rate": 1.971110306409594e-05,
      "loss": 0.5771,
      "step": 3321
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.079366683959961,
      "learning_rate": 1.971085998938515e-05,
      "loss": 0.6014,
      "step": 3322
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.00927734375,
      "learning_rate": 1.971061681395722e-05,
      "loss": 0.6054,
      "step": 3323
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9053122997283936,
      "learning_rate": 1.9710373537814662e-05,
      "loss": 0.5973,
      "step": 3324
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9889461994171143,
      "learning_rate": 1.9710130160960002e-05,
      "loss": 0.5621,
      "step": 3325
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.238046884536743,
      "learning_rate": 1.9709886683395768e-05,
      "loss": 0.5798,
      "step": 3326
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.135744094848633,
      "learning_rate": 1.9709643105124483e-05,
      "loss": 0.6003,
      "step": 3327
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8616716861724854,
      "learning_rate": 1.970939942614867e-05,
      "loss": 0.4874,
      "step": 3328
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9889836311340332,
      "learning_rate": 1.9709155646470858e-05,
      "loss": 0.6121,
      "step": 3329
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0414750576019287,
      "learning_rate": 1.970891176609358e-05,
      "loss": 0.549,
      "step": 3330
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.060370683670044,
      "learning_rate": 1.970866778501936e-05,
      "loss": 0.5782,
      "step": 3331
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0370683670043945,
      "learning_rate": 1.9708423703250726e-05,
      "loss": 0.5898,
      "step": 3332
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9582985639572144,
      "learning_rate": 1.970817952079022e-05,
      "loss": 0.5702,
      "step": 3333
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9024266004562378,
      "learning_rate": 1.9707935237640362e-05,
      "loss": 0.5045,
      "step": 3334
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.017181873321533,
      "learning_rate": 1.97076908538037e-05,
      "loss": 0.6346,
      "step": 3335
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.003878355026245,
      "learning_rate": 1.9707446369282756e-05,
      "loss": 0.5551,
      "step": 3336
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9213016033172607,
      "learning_rate": 1.970720178408007e-05,
      "loss": 0.5755,
      "step": 3337
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8884198665618896,
      "learning_rate": 1.9706957098198175e-05,
      "loss": 0.4893,
      "step": 3338
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.772874355316162,
      "learning_rate": 1.970671231163962e-05,
      "loss": 0.9593,
      "step": 3339
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1793699264526367,
      "learning_rate": 1.970646742440693e-05,
      "loss": 0.9334,
      "step": 3340
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2501959800720215,
      "learning_rate": 1.9706222436502656e-05,
      "loss": 0.6597,
      "step": 3341
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.270705461502075,
      "learning_rate": 1.970597734792933e-05,
      "loss": 0.4851,
      "step": 3342
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.105064630508423,
      "learning_rate": 1.9705732158689503e-05,
      "loss": 0.6075,
      "step": 3343
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9812688827514648,
      "learning_rate": 1.970548686878571e-05,
      "loss": 0.5399,
      "step": 3344
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0239505767822266,
      "learning_rate": 1.9705241478220502e-05,
      "loss": 0.5746,
      "step": 3345
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.063350200653076,
      "learning_rate": 1.9704995986996416e-05,
      "loss": 0.473,
      "step": 3346
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.076598644256592,
      "learning_rate": 1.9704750395116002e-05,
      "loss": 0.6,
      "step": 3347
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.835410237312317,
      "learning_rate": 1.970450470258181e-05,
      "loss": 0.5502,
      "step": 3348
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.011986017227173,
      "learning_rate": 1.9704258909396387e-05,
      "loss": 0.5859,
      "step": 3349
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.020167589187622,
      "learning_rate": 1.970401301556228e-05,
      "loss": 0.5185,
      "step": 3350
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.023442268371582,
      "learning_rate": 1.970376702108204e-05,
      "loss": 0.5666,
      "step": 3351
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9946315288543701,
      "learning_rate": 1.970352092595822e-05,
      "loss": 0.5601,
      "step": 3352
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0623700618743896,
      "learning_rate": 1.9703274730193372e-05,
      "loss": 0.5413,
      "step": 3353
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9387356042861938,
      "learning_rate": 1.970302843379005e-05,
      "loss": 0.563,
      "step": 3354
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.04628324508667,
      "learning_rate": 1.9702782036750802e-05,
      "loss": 0.5973,
      "step": 3355
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9575313329696655,
      "learning_rate": 1.9702535539078194e-05,
      "loss": 0.4822,
      "step": 3356
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.939095139503479,
      "learning_rate": 1.970228894077477e-05,
      "loss": 0.5676,
      "step": 3357
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9970612525939941,
      "learning_rate": 1.9702042241843103e-05,
      "loss": 0.5535,
      "step": 3358
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9848638772964478,
      "learning_rate": 1.970179544228574e-05,
      "loss": 0.6055,
      "step": 3359
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0140562057495117,
      "learning_rate": 1.9701548542105246e-05,
      "loss": 0.5801,
      "step": 3360
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9841820001602173,
      "learning_rate": 1.9701301541304177e-05,
      "loss": 0.6323,
      "step": 3361
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8747596740722656,
      "learning_rate": 1.9701054439885102e-05,
      "loss": 0.4619,
      "step": 3362
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8239002227783203,
      "learning_rate": 1.9700807237850577e-05,
      "loss": 0.5098,
      "step": 3363
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8283717632293701,
      "learning_rate": 1.970055993520317e-05,
      "loss": 0.5772,
      "step": 3364
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0121777057647705,
      "learning_rate": 1.9700312531945444e-05,
      "loss": 0.6,
      "step": 3365
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8091506958007812,
      "learning_rate": 1.970006502807996e-05,
      "loss": 0.4618,
      "step": 3366
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.004254102706909,
      "learning_rate": 1.96998174236093e-05,
      "loss": 0.6653,
      "step": 3367
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9084304571151733,
      "learning_rate": 1.9699569718536015e-05,
      "loss": 0.5293,
      "step": 3368
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9965786933898926,
      "learning_rate": 1.9699321912862685e-05,
      "loss": 0.6702,
      "step": 3369
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0214807987213135,
      "learning_rate": 1.9699074006591876e-05,
      "loss": 0.5002,
      "step": 3370
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8922103643417358,
      "learning_rate": 1.969882599972616e-05,
      "loss": 0.6109,
      "step": 3371
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.908345341682434,
      "learning_rate": 1.969857789226811e-05,
      "loss": 0.5587,
      "step": 3372
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0148868560791016,
      "learning_rate": 1.9698329684220296e-05,
      "loss": 0.5512,
      "step": 3373
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9337934255599976,
      "learning_rate": 1.96980813755853e-05,
      "loss": 0.5971,
      "step": 3374
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9385786056518555,
      "learning_rate": 1.9697832966365687e-05,
      "loss": 0.5435,
      "step": 3375
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8962656259536743,
      "learning_rate": 1.9697584456564042e-05,
      "loss": 0.4946,
      "step": 3376
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9239524602890015,
      "learning_rate": 1.9697335846182937e-05,
      "loss": 0.5508,
      "step": 3377
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1380863189697266,
      "learning_rate": 1.969708713522495e-05,
      "loss": 0.5216,
      "step": 3378
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.0930936336517334,
      "learning_rate": 1.9696838323692667e-05,
      "loss": 0.8841,
      "step": 3379
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.482057571411133,
      "learning_rate": 1.969658941158866e-05,
      "loss": 0.8858,
      "step": 3380
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3299286365509033,
      "learning_rate": 1.9696340398915522e-05,
      "loss": 0.5164,
      "step": 3381
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.73349666595459,
      "learning_rate": 1.9696091285675825e-05,
      "loss": 0.6037,
      "step": 3382
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0792765617370605,
      "learning_rate": 1.9695842071872155e-05,
      "loss": 0.6085,
      "step": 3383
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9414771795272827,
      "learning_rate": 1.96955927575071e-05,
      "loss": 0.5428,
      "step": 3384
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1403441429138184,
      "learning_rate": 1.9695343342583247e-05,
      "loss": 0.578,
      "step": 3385
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0067825317382812,
      "learning_rate": 1.9695093827103175e-05,
      "loss": 0.5342,
      "step": 3386
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0510478019714355,
      "learning_rate": 1.9694844211069477e-05,
      "loss": 0.6995,
      "step": 3387
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0342438220977783,
      "learning_rate": 1.9694594494484746e-05,
      "loss": 0.6196,
      "step": 3388
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.137932777404785,
      "learning_rate": 1.9694344677351563e-05,
      "loss": 0.6229,
      "step": 3389
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0582244396209717,
      "learning_rate": 1.9694094759672526e-05,
      "loss": 0.5335,
      "step": 3390
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9878313541412354,
      "learning_rate": 1.9693844741450224e-05,
      "loss": 0.5255,
      "step": 3391
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.067176103591919,
      "learning_rate": 1.969359462268725e-05,
      "loss": 0.6282,
      "step": 3392
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.272578001022339,
      "learning_rate": 1.9693344403386204e-05,
      "loss": 0.67,
      "step": 3393
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.077770471572876,
      "learning_rate": 1.969309408354967e-05,
      "loss": 0.5008,
      "step": 3394
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2489283084869385,
      "learning_rate": 1.9692843663180255e-05,
      "loss": 0.6249,
      "step": 3395
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1154427528381348,
      "learning_rate": 1.969259314228055e-05,
      "loss": 0.5677,
      "step": 3396
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0028367042541504,
      "learning_rate": 1.9692342520853152e-05,
      "loss": 0.5971,
      "step": 3397
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9325191974639893,
      "learning_rate": 1.9692091798900662e-05,
      "loss": 0.5159,
      "step": 3398
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.622724771499634,
      "learning_rate": 1.9691840976425686e-05,
      "loss": 0.9056,
      "step": 3399
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2164411544799805,
      "learning_rate": 1.969159005343082e-05,
      "loss": 0.8844,
      "step": 3400
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.6024584770202637,
      "learning_rate": 1.9691339029918666e-05,
      "loss": 0.6135,
      "step": 3401
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4310879707336426,
      "learning_rate": 1.9691087905891833e-05,
      "loss": 0.5463,
      "step": 3402
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0155365467071533,
      "learning_rate": 1.969083668135292e-05,
      "loss": 0.5722,
      "step": 3403
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9728822708129883,
      "learning_rate": 1.969058535630453e-05,
      "loss": 0.5105,
      "step": 3404
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.129049062728882,
      "learning_rate": 1.969033393074928e-05,
      "loss": 0.5325,
      "step": 3405
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.034395217895508,
      "learning_rate": 1.9690082404689764e-05,
      "loss": 0.4778,
      "step": 3406
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9464199542999268,
      "learning_rate": 1.9689830778128604e-05,
      "loss": 0.5721,
      "step": 3407
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9561737775802612,
      "learning_rate": 1.9689579051068404e-05,
      "loss": 0.5474,
      "step": 3408
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2778642177581787,
      "learning_rate": 1.9689327223511773e-05,
      "loss": 0.579,
      "step": 3409
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0048017501831055,
      "learning_rate": 1.9689075295461327e-05,
      "loss": 0.553,
      "step": 3410
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0466854572296143,
      "learning_rate": 1.9688823266919672e-05,
      "loss": 0.5935,
      "step": 3411
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9961285591125488,
      "learning_rate": 1.9688571137889433e-05,
      "loss": 0.4869,
      "step": 3412
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9413810968399048,
      "learning_rate": 1.968831890837321e-05,
      "loss": 0.5588,
      "step": 3413
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9260772466659546,
      "learning_rate": 1.9688066578373635e-05,
      "loss": 0.493,
      "step": 3414
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0860812664031982,
      "learning_rate": 1.9687814147893314e-05,
      "loss": 0.5538,
      "step": 3415
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.210702419281006,
      "learning_rate": 1.968756161693487e-05,
      "loss": 0.599,
      "step": 3416
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.896286129951477,
      "learning_rate": 1.9687308985500922e-05,
      "loss": 0.6213,
      "step": 3417
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.948264241218567,
      "learning_rate": 1.9687056253594085e-05,
      "loss": 0.4967,
      "step": 3418
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.687241315841675,
      "learning_rate": 1.9686803421216987e-05,
      "loss": 0.644,
      "step": 3419
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8468608856201172,
      "learning_rate": 1.9686550488372247e-05,
      "loss": 0.4881,
      "step": 3420
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8038520812988281,
      "learning_rate": 1.9686297455062488e-05,
      "loss": 0.4933,
      "step": 3421
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9869807958602905,
      "learning_rate": 1.9686044321290335e-05,
      "loss": 0.5845,
      "step": 3422
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.7156288623809814,
      "learning_rate": 1.9685791087058418e-05,
      "loss": 0.8426,
      "step": 3423
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.5518875122070312,
      "learning_rate": 1.9685537752369356e-05,
      "loss": 0.8721,
      "step": 3424
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.285343647003174,
      "learning_rate": 1.9685284317225777e-05,
      "loss": 0.5834,
      "step": 3425
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.114319324493408,
      "learning_rate": 1.9685030781630315e-05,
      "loss": 0.5071,
      "step": 3426
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.932546615600586,
      "learning_rate": 1.9684777145585596e-05,
      "loss": 0.5278,
      "step": 3427
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.049022912979126,
      "learning_rate": 1.9684523409094254e-05,
      "loss": 0.5837,
      "step": 3428
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.922922134399414,
      "learning_rate": 1.9684269572158913e-05,
      "loss": 0.5722,
      "step": 3429
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9759148359298706,
      "learning_rate": 1.9684015634782213e-05,
      "loss": 0.512,
      "step": 3430
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2293665409088135,
      "learning_rate": 1.9683761596966788e-05,
      "loss": 0.8515,
      "step": 3431
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.052597999572754,
      "learning_rate": 1.9683507458715265e-05,
      "loss": 0.8047,
      "step": 3432
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1810266971588135,
      "learning_rate": 1.968325322003029e-05,
      "loss": 0.5262,
      "step": 3433
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.322509288787842,
      "learning_rate": 1.968299888091449e-05,
      "loss": 0.6944,
      "step": 3434
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9757858514785767,
      "learning_rate": 1.968274444137051e-05,
      "loss": 0.5674,
      "step": 3435
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9313360452651978,
      "learning_rate": 1.9682489901400987e-05,
      "loss": 0.5505,
      "step": 3436
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2236926555633545,
      "learning_rate": 1.968223526100856e-05,
      "loss": 0.5789,
      "step": 3437
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8626269102096558,
      "learning_rate": 1.968198052019587e-05,
      "loss": 0.4941,
      "step": 3438
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9108057022094727,
      "learning_rate": 1.968172567896556e-05,
      "loss": 0.5815,
      "step": 3439
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9509518146514893,
      "learning_rate": 1.9681470737320275e-05,
      "loss": 0.5179,
      "step": 3440
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0396475791931152,
      "learning_rate": 1.9681215695262653e-05,
      "loss": 0.5422,
      "step": 3441
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9102667570114136,
      "learning_rate": 1.968096055279535e-05,
      "loss": 0.4925,
      "step": 3442
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.107306480407715,
      "learning_rate": 1.9680705309920997e-05,
      "loss": 0.6109,
      "step": 3443
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0422961711883545,
      "learning_rate": 1.9680449966642256e-05,
      "loss": 0.4981,
      "step": 3444
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9288002252578735,
      "learning_rate": 1.9680194522961766e-05,
      "loss": 0.6615,
      "step": 3445
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9727163314819336,
      "learning_rate": 1.967993897888218e-05,
      "loss": 0.5419,
      "step": 3446
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.041627883911133,
      "learning_rate": 1.967968333440615e-05,
      "loss": 0.5371,
      "step": 3447
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.028963804244995,
      "learning_rate": 1.967942758953632e-05,
      "loss": 0.5522,
      "step": 3448
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.980185866355896,
      "learning_rate": 1.9679171744275352e-05,
      "loss": 0.5323,
      "step": 3449
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0181784629821777,
      "learning_rate": 1.9678915798625895e-05,
      "loss": 0.6025,
      "step": 3450
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.049088954925537,
      "learning_rate": 1.96786597525906e-05,
      "loss": 0.6516,
      "step": 3451
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.083725690841675,
      "learning_rate": 1.967840360617213e-05,
      "loss": 0.5069,
      "step": 3452
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9857069253921509,
      "learning_rate": 1.9678147359373138e-05,
      "loss": 0.5867,
      "step": 3453
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9636297225952148,
      "learning_rate": 1.967789101219628e-05,
      "loss": 0.5342,
      "step": 3454
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1167078018188477,
      "learning_rate": 1.967763456464422e-05,
      "loss": 0.652,
      "step": 3455
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.125089168548584,
      "learning_rate": 1.967737801671961e-05,
      "loss": 0.5057,
      "step": 3456
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8493165969848633,
      "learning_rate": 1.9677121368425118e-05,
      "loss": 0.6111,
      "step": 3457
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8593051433563232,
      "learning_rate": 1.9676864619763402e-05,
      "loss": 0.5092,
      "step": 3458
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.193974256515503,
      "learning_rate": 1.9676607770737125e-05,
      "loss": 0.9153,
      "step": 3459
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4096035957336426,
      "learning_rate": 1.9676350821348953e-05,
      "loss": 0.896,
      "step": 3460
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.472208261489868,
      "learning_rate": 1.9676093771601547e-05,
      "loss": 0.6946,
      "step": 3461
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3759958744049072,
      "learning_rate": 1.9675836621497578e-05,
      "loss": 0.492,
      "step": 3462
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1883463859558105,
      "learning_rate": 1.9675579371039714e-05,
      "loss": 0.6274,
      "step": 3463
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8326054811477661,
      "learning_rate": 1.9675322020230614e-05,
      "loss": 0.4849,
      "step": 3464
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.073497772216797,
      "learning_rate": 1.9675064569072955e-05,
      "loss": 0.6627,
      "step": 3465
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0229225158691406,
      "learning_rate": 1.9674807017569407e-05,
      "loss": 0.4802,
      "step": 3466
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0521132946014404,
      "learning_rate": 1.9674549365722636e-05,
      "loss": 0.5966,
      "step": 3467
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9387627840042114,
      "learning_rate": 1.9674291613535323e-05,
      "loss": 0.4732,
      "step": 3468
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8308660984039307,
      "learning_rate": 1.9674033761010133e-05,
      "loss": 0.5539,
      "step": 3469
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9252632856369019,
      "learning_rate": 1.967377580814974e-05,
      "loss": 0.5146,
      "step": 3470
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4574127197265625,
      "learning_rate": 1.9673517754956827e-05,
      "loss": 0.9024,
      "step": 3471
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.368472099304199,
      "learning_rate": 1.9673259601434065e-05,
      "loss": 0.7887,
      "step": 3472
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3304927349090576,
      "learning_rate": 1.9673001347584133e-05,
      "loss": 0.722,
      "step": 3473
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2933454513549805,
      "learning_rate": 1.9672742993409707e-05,
      "loss": 0.5859,
      "step": 3474
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.293687105178833,
      "learning_rate": 1.967248453891347e-05,
      "loss": 0.6877,
      "step": 3475
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.208024501800537,
      "learning_rate": 1.9672225984098104e-05,
      "loss": 0.4823,
      "step": 3476
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.162714958190918,
      "learning_rate": 1.9671967328966282e-05,
      "loss": 0.5702,
      "step": 3477
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.97700035572052,
      "learning_rate": 1.9671708573520698e-05,
      "loss": 0.4956,
      "step": 3478
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.012300729751587,
      "learning_rate": 1.9671449717764027e-05,
      "loss": 0.5437,
      "step": 3479
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0984885692596436,
      "learning_rate": 1.967119076169896e-05,
      "loss": 0.5746,
      "step": 3480
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.066087007522583,
      "learning_rate": 1.9670931705328177e-05,
      "loss": 0.6595,
      "step": 3481
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0066614151000977,
      "learning_rate": 1.967067254865437e-05,
      "loss": 0.4843,
      "step": 3482
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0051958560943604,
      "learning_rate": 1.9670413291680223e-05,
      "loss": 0.5963,
      "step": 3483
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9051837921142578,
      "learning_rate": 1.9670153934408426e-05,
      "loss": 0.5745,
      "step": 3484
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.005959987640381,
      "learning_rate": 1.966989447684167e-05,
      "loss": 0.616,
      "step": 3485
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.078482151031494,
      "learning_rate": 1.9669634918982645e-05,
      "loss": 0.4999,
      "step": 3486
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0293076038360596,
      "learning_rate": 1.9669375260834042e-05,
      "loss": 0.5477,
      "step": 3487
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9912538528442383,
      "learning_rate": 1.9669115502398557e-05,
      "loss": 0.5567,
      "step": 3488
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0429883003234863,
      "learning_rate": 1.9668855643678884e-05,
      "loss": 0.5154,
      "step": 3489
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.021449327468872,
      "learning_rate": 1.9668595684677717e-05,
      "loss": 0.5143,
      "step": 3490
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.992899775505066,
      "learning_rate": 1.966833562539775e-05,
      "loss": 0.5443,
      "step": 3491
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.12870717048645,
      "learning_rate": 1.966807546584168e-05,
      "loss": 0.6315,
      "step": 3492
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.3158321380615234,
      "learning_rate": 1.9667815206012212e-05,
      "loss": 0.8793,
      "step": 3493
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.234635591506958,
      "learning_rate": 1.9667554845912037e-05,
      "loss": 0.8876,
      "step": 3494
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1860785484313965,
      "learning_rate": 1.966729438554386e-05,
      "loss": 0.6024,
      "step": 3495
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0308167934417725,
      "learning_rate": 1.966703382491038e-05,
      "loss": 0.5012,
      "step": 3496
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0606489181518555,
      "learning_rate": 1.9666773164014302e-05,
      "loss": 0.5904,
      "step": 3497
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8545279502868652,
      "learning_rate": 1.966651240285833e-05,
      "loss": 0.4795,
      "step": 3498
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9769015312194824,
      "learning_rate": 1.9666251541445165e-05,
      "loss": 0.6289,
      "step": 3499
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0465428829193115,
      "learning_rate": 1.9665990579777515e-05,
      "loss": 0.5418,
      "step": 3500
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.089188575744629,
      "learning_rate": 1.9665729517858088e-05,
      "loss": 0.6529,
      "step": 3501
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.910556674003601,
      "learning_rate": 1.9665468355689588e-05,
      "loss": 0.4466,
      "step": 3502
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9219577312469482,
      "learning_rate": 1.9665207093274725e-05,
      "loss": 0.5749,
      "step": 3503
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9627426862716675,
      "learning_rate": 1.966494573061621e-05,
      "loss": 0.5198,
      "step": 3504
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.966321587562561,
      "learning_rate": 1.9664684267716748e-05,
      "loss": 0.6056,
      "step": 3505
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8614258766174316,
      "learning_rate": 1.9664422704579063e-05,
      "loss": 0.4454,
      "step": 3506
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.010737180709839,
      "learning_rate": 1.9664161041205856e-05,
      "loss": 0.6468,
      "step": 3507
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.94419527053833,
      "learning_rate": 1.9663899277599846e-05,
      "loss": 0.5062,
      "step": 3508
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2271411418914795,
      "learning_rate": 1.9663637413763747e-05,
      "loss": 0.6366,
      "step": 3509
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9914648532867432,
      "learning_rate": 1.9663375449700275e-05,
      "loss": 0.4764,
      "step": 3510
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.008544445037842,
      "learning_rate": 1.9663113385412148e-05,
      "loss": 0.6079,
      "step": 3511
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0345940589904785,
      "learning_rate": 1.966285122090208e-05,
      "loss": 0.4966,
      "step": 3512
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9322832822799683,
      "learning_rate": 1.9662588956172798e-05,
      "loss": 0.5619,
      "step": 3513
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.929491400718689,
      "learning_rate": 1.9662326591227013e-05,
      "loss": 0.5169,
      "step": 3514
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9137061834335327,
      "learning_rate": 1.966206412606745e-05,
      "loss": 0.6109,
      "step": 3515
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0233309268951416,
      "learning_rate": 1.9661801560696835e-05,
      "loss": 0.4913,
      "step": 3516
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.9608569145202637,
      "learning_rate": 1.9661538895117885e-05,
      "loss": 0.9108,
      "step": 3517
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.5156335830688477,
      "learning_rate": 1.966127612933333e-05,
      "loss": 0.917,
      "step": 3518
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1634790897369385,
      "learning_rate": 1.9661013263345893e-05,
      "loss": 0.5718,
      "step": 3519
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0454702377319336,
      "learning_rate": 1.9660750297158294e-05,
      "loss": 0.492,
      "step": 3520
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0268280506134033,
      "learning_rate": 1.9660487230773273e-05,
      "loss": 0.5415,
      "step": 3521
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9378156661987305,
      "learning_rate": 1.9660224064193548e-05,
      "loss": 0.5193,
      "step": 3522
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.046097993850708,
      "learning_rate": 1.965996079742185e-05,
      "loss": 0.6183,
      "step": 3523
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.044984817504883,
      "learning_rate": 1.9659697430460917e-05,
      "loss": 0.4608,
      "step": 3524
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9041393995285034,
      "learning_rate": 1.965943396331347e-05,
      "loss": 0.613,
      "step": 3525
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9425950050354004,
      "learning_rate": 1.965917039598225e-05,
      "loss": 0.5334,
      "step": 3526
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.073453664779663,
      "learning_rate": 1.9658906728469984e-05,
      "loss": 0.629,
      "step": 3527
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.958126425743103,
      "learning_rate": 1.9658642960779413e-05,
      "loss": 0.498,
      "step": 3528
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0628297328948975,
      "learning_rate": 1.9658379092913267e-05,
      "loss": 0.6482,
      "step": 3529
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.913489580154419,
      "learning_rate": 1.9658115124874285e-05,
      "loss": 0.5083,
      "step": 3530
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9329952001571655,
      "learning_rate": 1.9657851056665206e-05,
      "loss": 0.5307,
      "step": 3531
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9955588579177856,
      "learning_rate": 1.965758688828877e-05,
      "loss": 0.6031,
      "step": 3532
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9280604124069214,
      "learning_rate": 1.965732261974771e-05,
      "loss": 0.5743,
      "step": 3533
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8697320222854614,
      "learning_rate": 1.9657058251044777e-05,
      "loss": 0.5138,
      "step": 3534
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.05340313911438,
      "learning_rate": 1.9656793782182704e-05,
      "loss": 0.6146,
      "step": 3535
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.814979076385498,
      "learning_rate": 1.9656529213164237e-05,
      "loss": 0.4745,
      "step": 3536
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9243580102920532,
      "learning_rate": 1.9656264543992123e-05,
      "loss": 0.5314,
      "step": 3537
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.096036911010742,
      "learning_rate": 1.9655999774669103e-05,
      "loss": 0.5464,
      "step": 3538
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9662317037582397,
      "learning_rate": 1.9655734905197926e-05,
      "loss": 0.5312,
      "step": 3539
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9152038097381592,
      "learning_rate": 1.9655469935581335e-05,
      "loss": 0.5829,
      "step": 3540
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.027747392654419,
      "learning_rate": 1.9655204865822085e-05,
      "loss": 0.6687,
      "step": 3541
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.041733980178833,
      "learning_rate": 1.965493969592292e-05,
      "loss": 0.507,
      "step": 3542
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9520190954208374,
      "learning_rate": 1.9654674425886586e-05,
      "loss": 0.5588,
      "step": 3543
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8593686819076538,
      "learning_rate": 1.9654409055715846e-05,
      "loss": 0.4914,
      "step": 3544
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9642235040664673,
      "learning_rate": 1.9654143585413445e-05,
      "loss": 0.6068,
      "step": 3545
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9681981801986694,
      "learning_rate": 1.9653878014982134e-05,
      "loss": 0.4715,
      "step": 3546
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9775952100753784,
      "learning_rate": 1.965361234442467e-05,
      "loss": 0.5932,
      "step": 3547
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8309191465377808,
      "learning_rate": 1.965334657374381e-05,
      "loss": 0.515,
      "step": 3548
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0494742393493652,
      "learning_rate": 1.9653080702942314e-05,
      "loss": 0.557,
      "step": 3549
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9771888256072998,
      "learning_rate": 1.9652814732022933e-05,
      "loss": 0.5605,
      "step": 3550
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7810146808624268,
      "learning_rate": 1.9652548660988425e-05,
      "loss": 0.5069,
      "step": 3551
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8497684001922607,
      "learning_rate": 1.9652282489841555e-05,
      "loss": 0.5789,
      "step": 3552
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9052411317825317,
      "learning_rate": 1.965201621858508e-05,
      "loss": 0.5461,
      "step": 3553
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9765242338180542,
      "learning_rate": 1.9651749847221762e-05,
      "loss": 0.5303,
      "step": 3554
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.353900909423828,
      "learning_rate": 1.9651483375754363e-05,
      "loss": 0.6068,
      "step": 3555
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7986613512039185,
      "learning_rate": 1.965121680418565e-05,
      "loss": 0.4708,
      "step": 3556
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.3080854415893555,
      "learning_rate": 1.9650950132518388e-05,
      "loss": 0.982,
      "step": 3557
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.0515897274017334,
      "learning_rate": 1.9650683360755335e-05,
      "loss": 1.0141,
      "step": 3558
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.453050374984741,
      "learning_rate": 1.9650416488899267e-05,
      "loss": 0.6273,
      "step": 3559
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.115875720977783,
      "learning_rate": 1.9650149516952947e-05,
      "loss": 0.5329,
      "step": 3560
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1284306049346924,
      "learning_rate": 1.9649882444919143e-05,
      "loss": 0.6039,
      "step": 3561
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9569575786590576,
      "learning_rate": 1.9649615272800632e-05,
      "loss": 0.5255,
      "step": 3562
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9746359586715698,
      "learning_rate": 1.9649348000600173e-05,
      "loss": 0.6084,
      "step": 3563
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9945354461669922,
      "learning_rate": 1.9649080628320553e-05,
      "loss": 0.5119,
      "step": 3564
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1268789768218994,
      "learning_rate": 1.9648813155964532e-05,
      "loss": 0.6317,
      "step": 3565
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.001471996307373,
      "learning_rate": 1.9648545583534892e-05,
      "loss": 0.5003,
      "step": 3566
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.870749831199646,
      "learning_rate": 1.9648277911034404e-05,
      "loss": 0.5382,
      "step": 3567
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9077657461166382,
      "learning_rate": 1.9648010138465844e-05,
      "loss": 0.526,
      "step": 3568
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.7149438858032227,
      "learning_rate": 1.9647742265831993e-05,
      "loss": 0.604,
      "step": 3569
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0740582942962646,
      "learning_rate": 1.964747429313563e-05,
      "loss": 0.5257,
      "step": 3570
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.088408946990967,
      "learning_rate": 1.9647206220379528e-05,
      "loss": 0.653,
      "step": 3571
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9469298124313354,
      "learning_rate": 1.964693804756647e-05,
      "loss": 0.5355,
      "step": 3572
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9029076099395752,
      "learning_rate": 1.9646669774699242e-05,
      "loss": 0.5971,
      "step": 3573
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9086753129959106,
      "learning_rate": 1.964640140178062e-05,
      "loss": 0.5221,
      "step": 3574
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9274194240570068,
      "learning_rate": 1.9646132928813388e-05,
      "loss": 0.5588,
      "step": 3575
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8491551876068115,
      "learning_rate": 1.9645864355800336e-05,
      "loss": 0.4744,
      "step": 3576
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.072655200958252,
      "learning_rate": 1.9645595682744246e-05,
      "loss": 0.6219,
      "step": 3577
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9584569931030273,
      "learning_rate": 1.9645326909647904e-05,
      "loss": 0.4815,
      "step": 3578
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8948111534118652,
      "learning_rate": 1.9645058036514097e-05,
      "loss": 0.6509,
      "step": 3579
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9139870405197144,
      "learning_rate": 1.9644789063345615e-05,
      "loss": 0.4659,
      "step": 3580
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9059709310531616,
      "learning_rate": 1.964451999014525e-05,
      "loss": 0.4969,
      "step": 3581
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8558440208435059,
      "learning_rate": 1.9644250816915787e-05,
      "loss": 0.5514,
      "step": 3582
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9136477708816528,
      "learning_rate": 1.9643981543660025e-05,
      "loss": 0.6115,
      "step": 3583
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0134706497192383,
      "learning_rate": 1.964371217038075e-05,
      "loss": 0.5058,
      "step": 3584
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.907696008682251,
      "learning_rate": 1.964344269708076e-05,
      "loss": 0.5934,
      "step": 3585
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.908410906791687,
      "learning_rate": 1.9643173123762848e-05,
      "loss": 0.461,
      "step": 3586
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0136892795562744,
      "learning_rate": 1.964290345042981e-05,
      "loss": 0.5541,
      "step": 3587
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9106115102767944,
      "learning_rate": 1.9642633677084447e-05,
      "loss": 0.4944,
      "step": 3588
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.954862356185913,
      "learning_rate": 1.964236380372955e-05,
      "loss": 0.5662,
      "step": 3589
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8665039539337158,
      "learning_rate": 1.964209383036792e-05,
      "loss": 0.51,
      "step": 3590
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.973280668258667,
      "learning_rate": 1.964182375700236e-05,
      "loss": 0.6483,
      "step": 3591
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8068153858184814,
      "learning_rate": 1.9641553583635672e-05,
      "loss": 0.4671,
      "step": 3592
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9447708129882812,
      "learning_rate": 1.9641283310270657e-05,
      "loss": 0.624,
      "step": 3593
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7437238693237305,
      "learning_rate": 1.9641012936910112e-05,
      "loss": 0.5147,
      "step": 3594
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.782769799232483,
      "learning_rate": 1.9640742463556844e-05,
      "loss": 0.5218,
      "step": 3595
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.913712978363037,
      "learning_rate": 1.9640471890213667e-05,
      "loss": 0.5521,
      "step": 3596
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9197231531143188,
      "learning_rate": 1.9640201216883377e-05,
      "loss": 0.5249,
      "step": 3597
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.822367548942566,
      "learning_rate": 1.9639930443568786e-05,
      "loss": 0.5303,
      "step": 3598
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0404751300811768,
      "learning_rate": 1.96396595702727e-05,
      "loss": 0.6561,
      "step": 3599
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9088597297668457,
      "learning_rate": 1.963938859699793e-05,
      "loss": 0.4581,
      "step": 3600
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.841267704963684,
      "learning_rate": 1.9639117523747288e-05,
      "loss": 0.6311,
      "step": 3601
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7884801626205444,
      "learning_rate": 1.963884635052358e-05,
      "loss": 0.4819,
      "step": 3602
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.948339819908142,
      "learning_rate": 1.9638575077329626e-05,
      "loss": 0.6004,
      "step": 3603
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8627976179122925,
      "learning_rate": 1.9638303704168233e-05,
      "loss": 0.4692,
      "step": 3604
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0403971672058105,
      "learning_rate": 1.9638032231042217e-05,
      "loss": 0.579,
      "step": 3605
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.840428113937378,
      "learning_rate": 1.96377606579544e-05,
      "loss": 0.5147,
      "step": 3606
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9885557889938354,
      "learning_rate": 1.963748898490759e-05,
      "loss": 0.5978,
      "step": 3607
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9550697803497314,
      "learning_rate": 1.963721721190461e-05,
      "loss": 0.514,
      "step": 3608
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0045955181121826,
      "learning_rate": 1.9636945338948275e-05,
      "loss": 0.6014,
      "step": 3609
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9578734636306763,
      "learning_rate": 1.9636673366041406e-05,
      "loss": 0.6096,
      "step": 3610
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0401289463043213,
      "learning_rate": 1.9636401293186824e-05,
      "loss": 0.6088,
      "step": 3611
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.009080410003662,
      "learning_rate": 1.9636129120387352e-05,
      "loss": 0.5316,
      "step": 3612
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9837238788604736,
      "learning_rate": 1.9635856847645812e-05,
      "loss": 0.6246,
      "step": 3613
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7988015413284302,
      "learning_rate": 1.963558447496503e-05,
      "loss": 0.4653,
      "step": 3614
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9651588201522827,
      "learning_rate": 1.9635312002347827e-05,
      "loss": 0.5469,
      "step": 3615
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.936478853225708,
      "learning_rate": 1.963503942979703e-05,
      "loss": 0.5594,
      "step": 3616
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8066489696502686,
      "learning_rate": 1.963476675731547e-05,
      "loss": 0.548,
      "step": 3617
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.88052499294281,
      "learning_rate": 1.9634493984905968e-05,
      "loss": 0.4562,
      "step": 3618
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9555045366287231,
      "learning_rate": 1.963422111257136e-05,
      "loss": 0.5421,
      "step": 3619
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9716624021530151,
      "learning_rate": 1.963394814031447e-05,
      "loss": 0.5292,
      "step": 3620
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0591394901275635,
      "learning_rate": 1.9633675068138135e-05,
      "loss": 0.6034,
      "step": 3621
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0161499977111816,
      "learning_rate": 1.9633401896045184e-05,
      "loss": 0.5653,
      "step": 3622
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2307088375091553,
      "learning_rate": 1.9633128624038453e-05,
      "loss": 0.8547,
      "step": 3623
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.887170672416687,
      "learning_rate": 1.963285525212077e-05,
      "loss": 0.9207,
      "step": 3624
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1878085136413574,
      "learning_rate": 1.9632581780294978e-05,
      "loss": 0.6276,
      "step": 3625
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9837020635604858,
      "learning_rate": 1.963230820856391e-05,
      "loss": 0.4872,
      "step": 3626
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.926742672920227,
      "learning_rate": 1.9632034536930398e-05,
      "loss": 0.587,
      "step": 3627
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9169743061065674,
      "learning_rate": 1.963176076539729e-05,
      "loss": 0.528,
      "step": 3628
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9780030250549316,
      "learning_rate": 1.9631486893967414e-05,
      "loss": 0.5683,
      "step": 3629
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9155170917510986,
      "learning_rate": 1.9631212922643627e-05,
      "loss": 0.6093,
      "step": 3630
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8801990747451782,
      "learning_rate": 1.9630938851428753e-05,
      "loss": 0.5615,
      "step": 3631
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8840959072113037,
      "learning_rate": 1.963066468032565e-05,
      "loss": 0.5868,
      "step": 3632
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8347574472427368,
      "learning_rate": 1.963039040933715e-05,
      "loss": 0.5192,
      "step": 3633
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.103066921234131,
      "learning_rate": 1.96301160384661e-05,
      "loss": 0.6373,
      "step": 3634
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0017502307891846,
      "learning_rate": 1.9629841567715347e-05,
      "loss": 0.568,
      "step": 3635
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9061907529830933,
      "learning_rate": 1.962956699708774e-05,
      "loss": 0.5619,
      "step": 3636
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.903071403503418,
      "learning_rate": 1.9629292326586126e-05,
      "loss": 0.5376,
      "step": 3637
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.014578104019165,
      "learning_rate": 1.962901755621335e-05,
      "loss": 0.6263,
      "step": 3638
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.989715576171875,
      "learning_rate": 1.9628742685972267e-05,
      "loss": 0.6091,
      "step": 3639
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8538533449172974,
      "learning_rate": 1.9628467715865725e-05,
      "loss": 0.4836,
      "step": 3640
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8823596239089966,
      "learning_rate": 1.9628192645896572e-05,
      "loss": 0.5879,
      "step": 3641
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9970442056655884,
      "learning_rate": 1.962791747606767e-05,
      "loss": 0.5833,
      "step": 3642
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1952970027923584,
      "learning_rate": 1.9627642206381864e-05,
      "loss": 0.6097,
      "step": 3643
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8514467477798462,
      "learning_rate": 1.9627366836842018e-05,
      "loss": 0.534,
      "step": 3644
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8769948482513428,
      "learning_rate": 1.962709136745098e-05,
      "loss": 0.5467,
      "step": 3645
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0955569744110107,
      "learning_rate": 1.962681579821161e-05,
      "loss": 0.5978,
      "step": 3646
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9396708011627197,
      "learning_rate": 1.9626540129126766e-05,
      "loss": 0.5617,
      "step": 3647
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8635520935058594,
      "learning_rate": 1.9626264360199304e-05,
      "loss": 0.5086,
      "step": 3648
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9900753498077393,
      "learning_rate": 1.9625988491432092e-05,
      "loss": 0.6311,
      "step": 3649
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.928536057472229,
      "learning_rate": 1.9625712522827983e-05,
      "loss": 0.5042,
      "step": 3650
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.8214898109436035,
      "learning_rate": 1.9625436454389847e-05,
      "loss": 0.9492,
      "step": 3651
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.5023813247680664,
      "learning_rate": 1.962516028612054e-05,
      "loss": 0.9535,
      "step": 3652
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.123873710632324,
      "learning_rate": 1.9624884018022932e-05,
      "loss": 0.532,
      "step": 3653
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0796947479248047,
      "learning_rate": 1.9624607650099885e-05,
      "loss": 0.5465,
      "step": 3654
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.100795269012451,
      "learning_rate": 1.9624331182354263e-05,
      "loss": 0.5961,
      "step": 3655
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.837954044342041,
      "learning_rate": 1.9624054614788942e-05,
      "loss": 0.4532,
      "step": 3656
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.913655400276184,
      "learning_rate": 1.9623777947406782e-05,
      "loss": 0.6371,
      "step": 3657
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.910586953163147,
      "learning_rate": 1.9623501180210653e-05,
      "loss": 0.4715,
      "step": 3658
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8888623714447021,
      "learning_rate": 1.9623224313203434e-05,
      "loss": 0.6135,
      "step": 3659
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0085606575012207,
      "learning_rate": 1.9622947346387984e-05,
      "loss": 0.5547,
      "step": 3660
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0072383880615234,
      "learning_rate": 1.9622670279767185e-05,
      "loss": 0.6535,
      "step": 3661
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9104145765304565,
      "learning_rate": 1.962239311334391e-05,
      "loss": 0.5135,
      "step": 3662
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8282573223114014,
      "learning_rate": 1.962211584712103e-05,
      "loss": 0.5603,
      "step": 3663
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.844398021697998,
      "learning_rate": 1.9621838481101422e-05,
      "loss": 0.4751,
      "step": 3664
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9271999597549438,
      "learning_rate": 1.962156101528796e-05,
      "loss": 0.6043,
      "step": 3665
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.886346697807312,
      "learning_rate": 1.962128344968353e-05,
      "loss": 0.5418,
      "step": 3666
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8855600357055664,
      "learning_rate": 1.9621005784291002e-05,
      "loss": 0.5268,
      "step": 3667
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.981999397277832,
      "learning_rate": 1.962072801911326e-05,
      "loss": 0.5784,
      "step": 3668
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8387376070022583,
      "learning_rate": 1.9620450154153182e-05,
      "loss": 0.6039,
      "step": 3669
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8637102842330933,
      "learning_rate": 1.9620172189413656e-05,
      "loss": 0.5027,
      "step": 3670
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9080661535263062,
      "learning_rate": 1.9619894124897558e-05,
      "loss": 0.6251,
      "step": 3671
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8739711046218872,
      "learning_rate": 1.9619615960607775e-05,
      "loss": 0.4948,
      "step": 3672
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9071835279464722,
      "learning_rate": 1.9619337696547195e-05,
      "loss": 0.52,
      "step": 3673
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8773552179336548,
      "learning_rate": 1.9619059332718697e-05,
      "loss": 0.5464,
      "step": 3674
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8581881523132324,
      "learning_rate": 1.9618780869125174e-05,
      "loss": 0.5939,
      "step": 3675
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9574058055877686,
      "learning_rate": 1.961850230576951e-05,
      "loss": 0.4675,
      "step": 3676
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.917250394821167,
      "learning_rate": 1.9618223642654598e-05,
      "loss": 0.6069,
      "step": 3677
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.83281672000885,
      "learning_rate": 1.9617944879783328e-05,
      "loss": 0.4715,
      "step": 3678
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8712011575698853,
      "learning_rate": 1.9617666017158588e-05,
      "loss": 0.6005,
      "step": 3679
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8258287906646729,
      "learning_rate": 1.9617387054783273e-05,
      "loss": 0.5226,
      "step": 3680
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7885934114456177,
      "learning_rate": 1.9617107992660274e-05,
      "loss": 0.5407,
      "step": 3681
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9674819707870483,
      "learning_rate": 1.9616828830792488e-05,
      "loss": 0.5146,
      "step": 3682
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9262330532073975,
      "learning_rate": 1.9616549569182808e-05,
      "loss": 0.5689,
      "step": 3683
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.918624758720398,
      "learning_rate": 1.961627020783413e-05,
      "loss": 0.5388,
      "step": 3684
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8313897848129272,
      "learning_rate": 1.9615990746749355e-05,
      "loss": 0.5456,
      "step": 3685
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9179985523223877,
      "learning_rate": 1.9615711185931378e-05,
      "loss": 0.5382,
      "step": 3686
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.845149278640747,
      "learning_rate": 1.96154315253831e-05,
      "loss": 0.5605,
      "step": 3687
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8172235488891602,
      "learning_rate": 1.9615151765107423e-05,
      "loss": 0.504,
      "step": 3688
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9986599683761597,
      "learning_rate": 1.9614871905107245e-05,
      "loss": 0.6296,
      "step": 3689
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7759969234466553,
      "learning_rate": 1.9614591945385472e-05,
      "loss": 0.4765,
      "step": 3690
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.9247634410858154,
      "learning_rate": 1.9614311885945004e-05,
      "loss": 0.8818,
      "step": 3691
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.572571277618408,
      "learning_rate": 1.9614031726788747e-05,
      "loss": 0.9376,
      "step": 3692
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9045342206954956,
      "learning_rate": 1.961375146791961e-05,
      "loss": 0.5688,
      "step": 3693
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.221043109893799,
      "learning_rate": 1.9613471109340497e-05,
      "loss": 0.6879,
      "step": 3694
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0167195796966553,
      "learning_rate": 1.9613190651054314e-05,
      "loss": 0.5537,
      "step": 3695
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9247621297836304,
      "learning_rate": 1.961291009306397e-05,
      "loss": 0.5207,
      "step": 3696
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1168885231018066,
      "learning_rate": 1.961262943537238e-05,
      "loss": 0.6761,
      "step": 3697
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0533504486083984,
      "learning_rate": 1.961234867798245e-05,
      "loss": 0.5079,
      "step": 3698
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0226035118103027,
      "learning_rate": 1.9612067820897093e-05,
      "loss": 0.5686,
      "step": 3699
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8993934392929077,
      "learning_rate": 1.9611786864119224e-05,
      "loss": 0.4915,
      "step": 3700
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8926255702972412,
      "learning_rate": 1.961150580765175e-05,
      "loss": 0.5271,
      "step": 3701
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9053514003753662,
      "learning_rate": 1.96112246514976e-05,
      "loss": 0.5826,
      "step": 3702
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8444117307662964,
      "learning_rate": 1.9610943395659677e-05,
      "loss": 0.5199,
      "step": 3703
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.937068223953247,
      "learning_rate": 1.96106620401409e-05,
      "loss": 0.4928,
      "step": 3704
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0579211711883545,
      "learning_rate": 1.9610380584944192e-05,
      "loss": 0.5845,
      "step": 3705
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8244069814682007,
      "learning_rate": 1.9610099030072466e-05,
      "loss": 0.4969,
      "step": 3706
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8461639881134033,
      "learning_rate": 1.9609817375528652e-05,
      "loss": 0.5895,
      "step": 3707
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8124754428863525,
      "learning_rate": 1.960953562131566e-05,
      "loss": 0.4708,
      "step": 3708
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9717267751693726,
      "learning_rate": 1.9609253767436418e-05,
      "loss": 0.5714,
      "step": 3709
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0306437015533447,
      "learning_rate": 1.9608971813893846e-05,
      "loss": 0.5683,
      "step": 3710
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9653675556182861,
      "learning_rate": 1.9608689760690875e-05,
      "loss": 0.6104,
      "step": 3711
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9636318683624268,
      "learning_rate": 1.9608407607830426e-05,
      "loss": 0.5115,
      "step": 3712
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.008660316467285,
      "learning_rate": 1.9608125355315425e-05,
      "loss": 0.5924,
      "step": 3713
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9464620351791382,
      "learning_rate": 1.9607843003148797e-05,
      "loss": 0.549,
      "step": 3714
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9411555528640747,
      "learning_rate": 1.9607560551333477e-05,
      "loss": 0.6223,
      "step": 3715
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.819130778312683,
      "learning_rate": 1.960727799987239e-05,
      "loss": 0.506,
      "step": 3716
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.739246129989624,
      "learning_rate": 1.9606995348768466e-05,
      "loss": 0.9061,
      "step": 3717
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.6016814708709717,
      "learning_rate": 1.9606712598024637e-05,
      "loss": 0.9515,
      "step": 3718
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1536223888397217,
      "learning_rate": 1.960642974764384e-05,
      "loss": 0.5496,
      "step": 3719
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3260693550109863,
      "learning_rate": 1.9606146797629003e-05,
      "loss": 0.6303,
      "step": 3720
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.002493381500244,
      "learning_rate": 1.9605863747983063e-05,
      "loss": 0.6164,
      "step": 3721
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9702283143997192,
      "learning_rate": 1.9605580598708955e-05,
      "loss": 0.5405,
      "step": 3722
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.063782215118408,
      "learning_rate": 1.9605297349809617e-05,
      "loss": 0.808,
      "step": 3723
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8686013221740723,
      "learning_rate": 1.9605014001287986e-05,
      "loss": 0.8672,
      "step": 3724
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2264797687530518,
      "learning_rate": 1.9604730553146997e-05,
      "loss": 0.5396,
      "step": 3725
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4012975692749023,
      "learning_rate": 1.9604447005389596e-05,
      "loss": 0.525,
      "step": 3726
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.15545916557312,
      "learning_rate": 1.9604163358018723e-05,
      "loss": 0.6203,
      "step": 3727
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9639898538589478,
      "learning_rate": 1.9603879611037316e-05,
      "loss": 0.5408,
      "step": 3728
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8975776433944702,
      "learning_rate": 1.960359576444832e-05,
      "loss": 0.5443,
      "step": 3729
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.136817455291748,
      "learning_rate": 1.960331181825468e-05,
      "loss": 0.4583,
      "step": 3730
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0836260318756104,
      "learning_rate": 1.960302777245934e-05,
      "loss": 0.6642,
      "step": 3731
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.818389654159546,
      "learning_rate": 1.9602743627065246e-05,
      "loss": 0.4638,
      "step": 3732
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8822194337844849,
      "learning_rate": 1.960245938207534e-05,
      "loss": 0.5694,
      "step": 3733
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.017407178878784,
      "learning_rate": 1.960217503749258e-05,
      "loss": 0.5319,
      "step": 3734
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1248278617858887,
      "learning_rate": 1.9601890593319914e-05,
      "loss": 0.6553,
      "step": 3735
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8204983472824097,
      "learning_rate": 1.9601606049560283e-05,
      "loss": 0.4799,
      "step": 3736
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9079381227493286,
      "learning_rate": 1.9601321406216645e-05,
      "loss": 0.6113,
      "step": 3737
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.98751962184906,
      "learning_rate": 1.960103666329195e-05,
      "loss": 0.4565,
      "step": 3738
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.070664405822754,
      "learning_rate": 1.9600751820789152e-05,
      "loss": 0.6387,
      "step": 3739
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9497990608215332,
      "learning_rate": 1.9600466878711207e-05,
      "loss": 0.467,
      "step": 3740
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.623175621032715,
      "learning_rate": 1.9600181837061066e-05,
      "loss": 0.991,
      "step": 3741
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2118279933929443,
      "learning_rate": 1.959989669584169e-05,
      "loss": 0.9815,
      "step": 3742
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0341947078704834,
      "learning_rate": 1.959961145505603e-05,
      "loss": 0.5394,
      "step": 3743
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0723137855529785,
      "learning_rate": 1.9599326114707057e-05,
      "loss": 0.4926,
      "step": 3744
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9341628551483154,
      "learning_rate": 1.9599040674797714e-05,
      "loss": 0.5814,
      "step": 3745
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9412453174591064,
      "learning_rate": 1.9598755135330972e-05,
      "loss": 0.5448,
      "step": 3746
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.152130365371704,
      "learning_rate": 1.959846949630979e-05,
      "loss": 0.5375,
      "step": 3747
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.952209711074829,
      "learning_rate": 1.9598183757737133e-05,
      "loss": 0.5219,
      "step": 3748
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.010481834411621,
      "learning_rate": 1.959789791961596e-05,
      "loss": 0.511,
      "step": 3749
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8983986377716064,
      "learning_rate": 1.959761198194924e-05,
      "loss": 0.6264,
      "step": 3750
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8075023889541626,
      "learning_rate": 1.9597325944739934e-05,
      "loss": 0.5243,
      "step": 3751
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8989427089691162,
      "learning_rate": 1.959703980799101e-05,
      "loss": 0.5279,
      "step": 3752
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.147968053817749,
      "learning_rate": 1.9596753571705437e-05,
      "loss": 0.5425,
      "step": 3753
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.009361982345581,
      "learning_rate": 1.9596467235886188e-05,
      "loss": 0.5599,
      "step": 3754
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9209345579147339,
      "learning_rate": 1.9596180800536225e-05,
      "loss": 0.5641,
      "step": 3755
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8934026956558228,
      "learning_rate": 1.959589426565852e-05,
      "loss": 0.5286,
      "step": 3756
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8885831832885742,
      "learning_rate": 1.959560763125605e-05,
      "loss": 0.5778,
      "step": 3757
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8625489473342896,
      "learning_rate": 1.9595320897331783e-05,
      "loss": 0.4878,
      "step": 3758
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9837943315505981,
      "learning_rate": 1.9595034063888698e-05,
      "loss": 0.6912,
      "step": 3759
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8829591274261475,
      "learning_rate": 1.959474713092976e-05,
      "loss": 0.4859,
      "step": 3760
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9657381772994995,
      "learning_rate": 1.959446009845796e-05,
      "loss": 0.6312,
      "step": 3761
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.804093360900879,
      "learning_rate": 1.9594172966476258e-05,
      "loss": 0.4774,
      "step": 3762
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8759372234344482,
      "learning_rate": 1.9593885734987647e-05,
      "loss": 0.5686,
      "step": 3763
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.905678153038025,
      "learning_rate": 1.9593598403995094e-05,
      "loss": 0.6496,
      "step": 3764
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.909092903137207,
      "learning_rate": 1.959331097350159e-05,
      "loss": 0.5065,
      "step": 3765
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9174096584320068,
      "learning_rate": 1.9593023443510107e-05,
      "loss": 0.5516,
      "step": 3766
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.06195068359375,
      "learning_rate": 1.959273581402363e-05,
      "loss": 0.5784,
      "step": 3767
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9084028005599976,
      "learning_rate": 1.9592448085045147e-05,
      "loss": 0.5355,
      "step": 3768
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9244128465652466,
      "learning_rate": 1.9592160256577636e-05,
      "loss": 0.5342,
      "step": 3769
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8629350662231445,
      "learning_rate": 1.9591872328624085e-05,
      "loss": 0.4889,
      "step": 3770
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.087432861328125,
      "learning_rate": 1.9591584301187477e-05,
      "loss": 0.5409,
      "step": 3771
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0008773803710938,
      "learning_rate": 1.9591296174270807e-05,
      "loss": 0.5311,
      "step": 3772
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.011533498764038,
      "learning_rate": 1.9591007947877053e-05,
      "loss": 0.5253,
      "step": 3773
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9381788969039917,
      "learning_rate": 1.9590719622009216e-05,
      "loss": 0.5433,
      "step": 3774
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9496726989746094,
      "learning_rate": 1.9590431196670276e-05,
      "loss": 0.5758,
      "step": 3775
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9442787170410156,
      "learning_rate": 1.959014267186323e-05,
      "loss": 0.5148,
      "step": 3776
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8702341318130493,
      "learning_rate": 1.958985404759107e-05,
      "loss": 0.5878,
      "step": 3777
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8150337934494019,
      "learning_rate": 1.9589565323856787e-05,
      "loss": 0.4753,
      "step": 3778
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.842096209526062,
      "learning_rate": 1.9589276500663374e-05,
      "loss": 0.5682,
      "step": 3779
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.871877670288086,
      "learning_rate": 1.9588987578013836e-05,
      "loss": 0.4797,
      "step": 3780
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9562889337539673,
      "learning_rate": 1.9588698555911158e-05,
      "loss": 0.5904,
      "step": 3781
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9622915983200073,
      "learning_rate": 1.9588409434358345e-05,
      "loss": 0.4679,
      "step": 3782
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4696462154388428,
      "learning_rate": 1.9588120213358395e-05,
      "loss": 0.9764,
      "step": 3783
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.289210081100464,
      "learning_rate": 1.9587830892914304e-05,
      "loss": 0.935,
      "step": 3784
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2150089740753174,
      "learning_rate": 1.9587541473029077e-05,
      "loss": 0.689,
      "step": 3785
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0153348445892334,
      "learning_rate": 1.9587251953705712e-05,
      "loss": 0.4768,
      "step": 3786
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9838978052139282,
      "learning_rate": 1.9586962334947212e-05,
      "loss": 0.5831,
      "step": 3787
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9997788667678833,
      "learning_rate": 1.9586672616756585e-05,
      "loss": 0.4919,
      "step": 3788
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9125803709030151,
      "learning_rate": 1.9586382799136832e-05,
      "loss": 0.5797,
      "step": 3789
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.950254201889038,
      "learning_rate": 1.958609288209096e-05,
      "loss": 0.5904,
      "step": 3790
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9535374641418457,
      "learning_rate": 1.9585802865621977e-05,
      "loss": 0.6403,
      "step": 3791
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.853293776512146,
      "learning_rate": 1.958551274973289e-05,
      "loss": 0.5092,
      "step": 3792
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.931304931640625,
      "learning_rate": 1.9585222534426704e-05,
      "loss": 0.6282,
      "step": 3793
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8043893575668335,
      "learning_rate": 1.9584932219706435e-05,
      "loss": 0.5422,
      "step": 3794
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4607417583465576,
      "learning_rate": 1.9584641805575093e-05,
      "loss": 0.9108,
      "step": 3795
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.277395725250244,
      "learning_rate": 1.958435129203569e-05,
      "loss": 0.8997,
      "step": 3796
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0514237880706787,
      "learning_rate": 1.9584060679091235e-05,
      "loss": 0.6455,
      "step": 3797
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.054694414138794,
      "learning_rate": 1.9583769966744746e-05,
      "loss": 0.5485,
      "step": 3798
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8869198560714722,
      "learning_rate": 1.958347915499924e-05,
      "loss": 0.5653,
      "step": 3799
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8359017372131348,
      "learning_rate": 1.9583188243857726e-05,
      "loss": 0.5277,
      "step": 3800
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.773429274559021,
      "learning_rate": 1.9582897233323225e-05,
      "loss": 0.4908,
      "step": 3801
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9274989366531372,
      "learning_rate": 1.958260612339876e-05,
      "loss": 0.5614,
      "step": 3802
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9321260452270508,
      "learning_rate": 1.9582314914087344e-05,
      "loss": 0.5863,
      "step": 3803
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.86699378490448,
      "learning_rate": 1.9582023605392e-05,
      "loss": 0.4901,
      "step": 3804
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9185889959335327,
      "learning_rate": 1.958173219731575e-05,
      "loss": 0.5946,
      "step": 3805
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8457525968551636,
      "learning_rate": 1.958144068986161e-05,
      "loss": 0.4856,
      "step": 3806
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8776696920394897,
      "learning_rate": 1.9581149083032612e-05,
      "loss": 0.6171,
      "step": 3807
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.858508825302124,
      "learning_rate": 1.9580857376831778e-05,
      "loss": 0.4882,
      "step": 3808
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8322787284851074,
      "learning_rate": 1.9580565571262135e-05,
      "loss": 0.5908,
      "step": 3809
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8692080974578857,
      "learning_rate": 1.95802736663267e-05,
      "loss": 0.5405,
      "step": 3810
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0189523696899414,
      "learning_rate": 1.957998166202851e-05,
      "loss": 0.654,
      "step": 3811
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9118375778198242,
      "learning_rate": 1.9579689558370595e-05,
      "loss": 0.5361,
      "step": 3812
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8217592239379883,
      "learning_rate": 1.9579397355355976e-05,
      "loss": 0.5798,
      "step": 3813
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8076825141906738,
      "learning_rate": 1.957910505298769e-05,
      "loss": 0.5089,
      "step": 3814
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9675260782241821,
      "learning_rate": 1.9578812651268765e-05,
      "loss": 0.6185,
      "step": 3815
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.929643988609314,
      "learning_rate": 1.957852015020224e-05,
      "loss": 0.5481,
      "step": 3816
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.939846396446228,
      "learning_rate": 1.957822754979114e-05,
      "loss": 0.6427,
      "step": 3817
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9351643323898315,
      "learning_rate": 1.9577934850038505e-05,
      "loss": 0.4945,
      "step": 3818
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0698161125183105,
      "learning_rate": 1.957764205094737e-05,
      "loss": 0.5599,
      "step": 3819
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8762227296829224,
      "learning_rate": 1.957734915252077e-05,
      "loss": 0.5679,
      "step": 3820
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8232635259628296,
      "learning_rate": 1.9577056154761748e-05,
      "loss": 0.5229,
      "step": 3821
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0877981185913086,
      "learning_rate": 1.9576763057673336e-05,
      "loss": 0.6303,
      "step": 3822
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9747437238693237,
      "learning_rate": 1.957646986125858e-05,
      "loss": 0.6467,
      "step": 3823
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8203619718551636,
      "learning_rate": 1.9576176565520516e-05,
      "loss": 0.5241,
      "step": 3824
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8665086030960083,
      "learning_rate": 1.9575883170462186e-05,
      "loss": 0.5872,
      "step": 3825
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9062374830245972,
      "learning_rate": 1.9575589676086638e-05,
      "loss": 0.6085,
      "step": 3826
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8230785131454468,
      "learning_rate": 1.9575296082396913e-05,
      "loss": 0.6078,
      "step": 3827
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8932050466537476,
      "learning_rate": 1.9575002389396056e-05,
      "loss": 0.5376,
      "step": 3828
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.906282901763916,
      "learning_rate": 1.957470859708711e-05,
      "loss": 0.5444,
      "step": 3829
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8574320077896118,
      "learning_rate": 1.9574414705473127e-05,
      "loss": 0.5796,
      "step": 3830
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1525256633758545,
      "learning_rate": 1.9574120714557154e-05,
      "loss": 0.5643,
      "step": 3831
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8578165769577026,
      "learning_rate": 1.9573826624342236e-05,
      "loss": 0.5289,
      "step": 3832
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9182016849517822,
      "learning_rate": 1.957353243483143e-05,
      "loss": 0.6181,
      "step": 3833
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7664110660552979,
      "learning_rate": 1.9573238146027784e-05,
      "loss": 0.4775,
      "step": 3834
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0073390007019043,
      "learning_rate": 1.957294375793435e-05,
      "loss": 0.6638,
      "step": 3835
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8763731718063354,
      "learning_rate": 1.957264927055418e-05,
      "loss": 0.4779,
      "step": 3836
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.83838951587677,
      "learning_rate": 1.957235468389033e-05,
      "loss": 0.6267,
      "step": 3837
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7738450765609741,
      "learning_rate": 1.9572059997945853e-05,
      "loss": 0.472,
      "step": 3838
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.8006913661956787,
      "learning_rate": 1.957176521272381e-05,
      "loss": 0.912,
      "step": 3839
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.228377342224121,
      "learning_rate": 1.9571470328227254e-05,
      "loss": 0.9253,
      "step": 3840
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.22159743309021,
      "learning_rate": 1.9571175344459248e-05,
      "loss": 0.6339,
      "step": 3841
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1850931644439697,
      "learning_rate": 1.9570880261422846e-05,
      "loss": 0.5026,
      "step": 3842
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9349864721298218,
      "learning_rate": 1.957058507912111e-05,
      "loss": 0.5881,
      "step": 3843
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.830106258392334,
      "learning_rate": 1.9570289797557104e-05,
      "loss": 0.5384,
      "step": 3844
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9192044734954834,
      "learning_rate": 1.9569994416733892e-05,
      "loss": 0.5614,
      "step": 3845
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.981472134590149,
      "learning_rate": 1.956969893665453e-05,
      "loss": 0.4862,
      "step": 3846
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9801552295684814,
      "learning_rate": 1.956940335732209e-05,
      "loss": 0.5817,
      "step": 3847
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.897931456565857,
      "learning_rate": 1.9569107678739635e-05,
      "loss": 0.5515,
      "step": 3848
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9373915195465088,
      "learning_rate": 1.956881190091023e-05,
      "loss": 0.598,
      "step": 3849
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8541908264160156,
      "learning_rate": 1.9568516023836945e-05,
      "loss": 0.5061,
      "step": 3850
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.886778473854065,
      "learning_rate": 1.956822004752285e-05,
      "loss": 0.5742,
      "step": 3851
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8677978515625,
      "learning_rate": 1.956792397197101e-05,
      "loss": 0.4877,
      "step": 3852
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.003192901611328,
      "learning_rate": 1.95676277971845e-05,
      "loss": 0.6571,
      "step": 3853
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8637328147888184,
      "learning_rate": 1.956733152316639e-05,
      "loss": 0.4951,
      "step": 3854
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8739882707595825,
      "learning_rate": 1.956703514991975e-05,
      "loss": 0.5563,
      "step": 3855
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9153698682785034,
      "learning_rate": 1.9566738677447662e-05,
      "loss": 0.5418,
      "step": 3856
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9109656810760498,
      "learning_rate": 1.9566442105753193e-05,
      "loss": 0.622,
      "step": 3857
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8302069902420044,
      "learning_rate": 1.9566145434839426e-05,
      "loss": 0.5301,
      "step": 3858
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.032013177871704,
      "learning_rate": 1.956584866470943e-05,
      "loss": 0.5794,
      "step": 3859
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9268012046813965,
      "learning_rate": 1.9565551795366286e-05,
      "loss": 0.53,
      "step": 3860
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8675469160079956,
      "learning_rate": 1.9565254826813076e-05,
      "loss": 0.5064,
      "step": 3861
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8904006481170654,
      "learning_rate": 1.9564957759052877e-05,
      "loss": 0.5075,
      "step": 3862
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9481168985366821,
      "learning_rate": 1.956466059208877e-05,
      "loss": 0.6434,
      "step": 3863
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8922746181488037,
      "learning_rate": 1.9564363325923842e-05,
      "loss": 0.5499,
      "step": 3864
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9126032590866089,
      "learning_rate": 1.9564065960561165e-05,
      "loss": 0.5873,
      "step": 3865
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8777587413787842,
      "learning_rate": 1.9563768496003835e-05,
      "loss": 0.4812,
      "step": 3866
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9181262254714966,
      "learning_rate": 1.9563470932254932e-05,
      "loss": 0.5178,
      "step": 3867
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8753776550292969,
      "learning_rate": 1.9563173269317543e-05,
      "loss": 0.5702,
      "step": 3868
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.931808352470398,
      "learning_rate": 1.9562875507194754e-05,
      "loss": 0.5929,
      "step": 3869
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8975634574890137,
      "learning_rate": 1.9562577645889655e-05,
      "loss": 0.463,
      "step": 3870
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1212127208709717,
      "learning_rate": 1.9562279685405336e-05,
      "loss": 0.6399,
      "step": 3871
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9738719463348389,
      "learning_rate": 1.956198162574488e-05,
      "loss": 0.5242,
      "step": 3872
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.823659062385559,
      "learning_rate": 1.956168346691139e-05,
      "loss": 0.5429,
      "step": 3873
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0150368213653564,
      "learning_rate": 1.9561385208907952e-05,
      "loss": 0.5529,
      "step": 3874
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0405938625335693,
      "learning_rate": 1.9561086851737656e-05,
      "loss": 0.5873,
      "step": 3875
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9643762111663818,
      "learning_rate": 1.9560788395403604e-05,
      "loss": 0.5354,
      "step": 3876
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0128257274627686,
      "learning_rate": 1.9560489839908885e-05,
      "loss": 0.6105,
      "step": 3877
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9166885614395142,
      "learning_rate": 1.95601911852566e-05,
      "loss": 0.4928,
      "step": 3878
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.928942322731018,
      "learning_rate": 1.9559892431449846e-05,
      "loss": 0.5521,
      "step": 3879
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8174275159835815,
      "learning_rate": 1.9559593578491718e-05,
      "loss": 0.5387,
      "step": 3880
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.01135516166687,
      "learning_rate": 1.9559294626385318e-05,
      "loss": 0.6141,
      "step": 3881
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.892021894454956,
      "learning_rate": 1.955899557513375e-05,
      "loss": 0.485,
      "step": 3882
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7995225191116333,
      "learning_rate": 1.955869642474011e-05,
      "loss": 0.5493,
      "step": 3883
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.829280138015747,
      "learning_rate": 1.9558397175207506e-05,
      "loss": 0.4948,
      "step": 3884
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9434926509857178,
      "learning_rate": 1.9558097826539036e-05,
      "loss": 0.5874,
      "step": 3885
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9962177276611328,
      "learning_rate": 1.9557798378737805e-05,
      "loss": 0.5409,
      "step": 3886
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9047497510910034,
      "learning_rate": 1.9557498831806922e-05,
      "loss": 0.5834,
      "step": 3887
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.907099723815918,
      "learning_rate": 1.9557199185749494e-05,
      "loss": 0.533,
      "step": 3888
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9250227212905884,
      "learning_rate": 1.9556899440568632e-05,
      "loss": 0.6335,
      "step": 3889
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.899619460105896,
      "learning_rate": 1.9556599596267433e-05,
      "loss": 0.5119,
      "step": 3890
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9406695365905762,
      "learning_rate": 1.955629965284902e-05,
      "loss": 0.6257,
      "step": 3891
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8076837062835693,
      "learning_rate": 1.9555999610316496e-05,
      "loss": 0.4917,
      "step": 3892
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.881759762763977,
      "learning_rate": 1.9555699468672976e-05,
      "loss": 0.5811,
      "step": 3893
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8380794525146484,
      "learning_rate": 1.955539922792157e-05,
      "loss": 0.4456,
      "step": 3894
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8210200071334839,
      "learning_rate": 1.9555098888065398e-05,
      "loss": 0.5295,
      "step": 3895
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.930196762084961,
      "learning_rate": 1.9554798449107567e-05,
      "loss": 0.5624,
      "step": 3896
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0577657222747803,
      "learning_rate": 1.9554497911051202e-05,
      "loss": 0.6307,
      "step": 3897
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8258202075958252,
      "learning_rate": 1.9554197273899408e-05,
      "loss": 0.458,
      "step": 3898
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0505881309509277,
      "learning_rate": 1.9553896537655317e-05,
      "loss": 0.6471,
      "step": 3899
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7585076093673706,
      "learning_rate": 1.955359570232204e-05,
      "loss": 0.4423,
      "step": 3900
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.922337293624878,
      "learning_rate": 1.9553294767902697e-05,
      "loss": 0.574,
      "step": 3901
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8775968551635742,
      "learning_rate": 1.9552993734400413e-05,
      "loss": 0.5457,
      "step": 3902
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9824366569519043,
      "learning_rate": 1.9552692601818303e-05,
      "loss": 0.5984,
      "step": 3903
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8484649658203125,
      "learning_rate": 1.95523913701595e-05,
      "loss": 0.5026,
      "step": 3904
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2990682125091553,
      "learning_rate": 1.9552090039427122e-05,
      "loss": 0.9122,
      "step": 3905
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.048741102218628,
      "learning_rate": 1.9551788609624295e-05,
      "loss": 0.9095,
      "step": 3906
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9556232690811157,
      "learning_rate": 1.9551487080754148e-05,
      "loss": 0.5776,
      "step": 3907
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.004913330078125,
      "learning_rate": 1.9551185452819804e-05,
      "loss": 0.5034,
      "step": 3908
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9964015483856201,
      "learning_rate": 1.9550883725824396e-05,
      "loss": 0.6554,
      "step": 3909
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8212666511535645,
      "learning_rate": 1.9550581899771052e-05,
      "loss": 0.4618,
      "step": 3910
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9051845073699951,
      "learning_rate": 1.95502799746629e-05,
      "loss": 0.5603,
      "step": 3911
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8343278169631958,
      "learning_rate": 1.954997795050307e-05,
      "loss": 0.532,
      "step": 3912
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9429235458374023,
      "learning_rate": 1.95496758272947e-05,
      "loss": 0.6246,
      "step": 3913
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9454232454299927,
      "learning_rate": 1.9549373605040922e-05,
      "loss": 0.4817,
      "step": 3914
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9641863107681274,
      "learning_rate": 1.954907128374487e-05,
      "loss": 0.5788,
      "step": 3915
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.833397626876831,
      "learning_rate": 1.9548768863409676e-05,
      "loss": 0.5574,
      "step": 3916
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9019560813903809,
      "learning_rate": 1.9548466344038482e-05,
      "loss": 0.6405,
      "step": 3917
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9754997491836548,
      "learning_rate": 1.9548163725634422e-05,
      "loss": 0.4686,
      "step": 3918
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0989737510681152,
      "learning_rate": 1.9547861008200637e-05,
      "loss": 0.6639,
      "step": 3919
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8686846494674683,
      "learning_rate": 1.9547558191740267e-05,
      "loss": 0.4829,
      "step": 3920
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.979864478111267,
      "learning_rate": 1.9547255276256447e-05,
      "loss": 0.595,
      "step": 3921
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.960647702217102,
      "learning_rate": 1.9546952261752327e-05,
      "loss": 0.5373,
      "step": 3922
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2036490440368652,
      "learning_rate": 1.954664914823104e-05,
      "loss": 0.6909,
      "step": 3923
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9491482973098755,
      "learning_rate": 1.954634593569574e-05,
      "loss": 0.4899,
      "step": 3924
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8657118082046509,
      "learning_rate": 1.9546042624149567e-05,
      "loss": 0.5226,
      "step": 3925
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1708014011383057,
      "learning_rate": 1.9545739213595666e-05,
      "loss": 0.5549,
      "step": 3926
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.962980031967163,
      "learning_rate": 1.9545435704037187e-05,
      "loss": 0.6204,
      "step": 3927
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.860318899154663,
      "learning_rate": 1.9545132095477274e-05,
      "loss": 0.4888,
      "step": 3928
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8621764183044434,
      "learning_rate": 1.9544828387919077e-05,
      "loss": 0.5235,
      "step": 3929
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8814749717712402,
      "learning_rate": 1.954452458136575e-05,
      "loss": 0.6019,
      "step": 3930
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8681199550628662,
      "learning_rate": 1.954422067582044e-05,
      "loss": 0.5903,
      "step": 3931
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8473962545394897,
      "learning_rate": 1.9543916671286297e-05,
      "loss": 0.5094,
      "step": 3932
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7923551797866821,
      "learning_rate": 1.9543612567766476e-05,
      "loss": 0.5564,
      "step": 3933
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.82506263256073,
      "learning_rate": 1.9543308365264132e-05,
      "loss": 0.502,
      "step": 3934
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8965874910354614,
      "learning_rate": 1.954300406378242e-05,
      "loss": 0.5822,
      "step": 3935
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8817074298858643,
      "learning_rate": 1.9542699663324497e-05,
      "loss": 0.4979,
      "step": 3936
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9773422479629517,
      "learning_rate": 1.954239516389352e-05,
      "loss": 0.5931,
      "step": 3937
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8182153701782227,
      "learning_rate": 1.9542090565492646e-05,
      "loss": 0.5385,
      "step": 3938
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.989030361175537,
      "learning_rate": 1.9541785868125035e-05,
      "loss": 0.6929,
      "step": 3939
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7768428325653076,
      "learning_rate": 1.9541481071793842e-05,
      "loss": 0.4613,
      "step": 3940
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9763761758804321,
      "learning_rate": 1.9541176176502238e-05,
      "loss": 0.5674,
      "step": 3941
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9390512704849243,
      "learning_rate": 1.9540871182253377e-05,
      "loss": 0.5198,
      "step": 3942
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8214209079742432,
      "learning_rate": 1.9540566089050425e-05,
      "loss": 0.5434,
      "step": 3943
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.855008602142334,
      "learning_rate": 1.9540260896896544e-05,
      "loss": 0.5687,
      "step": 3944
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9170483350753784,
      "learning_rate": 1.9539955605794904e-05,
      "loss": 0.5859,
      "step": 3945
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7974270582199097,
      "learning_rate": 1.9539650215748672e-05,
      "loss": 0.4849,
      "step": 3946
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.953454613685608,
      "learning_rate": 1.953934472676101e-05,
      "loss": 0.6465,
      "step": 3947
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8290765285491943,
      "learning_rate": 1.953903913883509e-05,
      "loss": 0.4739,
      "step": 3948
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8236668109893799,
      "learning_rate": 1.953873345197408e-05,
      "loss": 0.5688,
      "step": 3949
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.760252594947815,
      "learning_rate": 1.9538427666181153e-05,
      "loss": 0.5121,
      "step": 3950
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7719447612762451,
      "learning_rate": 1.9538121781459478e-05,
      "loss": 0.512,
      "step": 3951
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9258919954299927,
      "learning_rate": 1.9537815797812224e-05,
      "loss": 0.5945,
      "step": 3952
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.876568078994751,
      "learning_rate": 1.953750971524257e-05,
      "loss": 0.5615,
      "step": 3953
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8317939043045044,
      "learning_rate": 1.9537203533753686e-05,
      "loss": 0.5646,
      "step": 3954
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8769102096557617,
      "learning_rate": 1.9536897253348755e-05,
      "loss": 0.5592,
      "step": 3955
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7277756929397583,
      "learning_rate": 1.953659087403095e-05,
      "loss": 0.4621,
      "step": 3956
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9555504322052002,
      "learning_rate": 1.9536284395803447e-05,
      "loss": 0.5745,
      "step": 3957
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0154073238372803,
      "learning_rate": 1.9535977818669423e-05,
      "loss": 0.5151,
      "step": 3958
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8486307859420776,
      "learning_rate": 1.9535671142632065e-05,
      "loss": 0.5903,
      "step": 3959
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9069702625274658,
      "learning_rate": 1.9535364367694544e-05,
      "loss": 0.4809,
      "step": 3960
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1796884536743164,
      "learning_rate": 1.953505749386005e-05,
      "loss": 0.6267,
      "step": 3961
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9990993738174438,
      "learning_rate": 1.9534750521131764e-05,
      "loss": 0.4463,
      "step": 3962
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.871727466583252,
      "learning_rate": 1.953444344951286e-05,
      "loss": 0.5331,
      "step": 3963
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1488466262817383,
      "learning_rate": 1.953413627900654e-05,
      "loss": 0.5611,
      "step": 3964
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.572208881378174,
      "learning_rate": 1.9533829009615974e-05,
      "loss": 0.5962,
      "step": 3965
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.134826183319092,
      "learning_rate": 1.953352164134436e-05,
      "loss": 0.5239,
      "step": 3966
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.083042860031128,
      "learning_rate": 1.9533214174194883e-05,
      "loss": 0.6332,
      "step": 3967
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8396748304367065,
      "learning_rate": 1.9532906608170727e-05,
      "loss": 0.4721,
      "step": 3968
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.020517110824585,
      "learning_rate": 1.953259894327509e-05,
      "loss": 0.5794,
      "step": 3969
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1216301918029785,
      "learning_rate": 1.9532291179511155e-05,
      "loss": 0.5228,
      "step": 3970
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9385709762573242,
      "learning_rate": 1.9531983316882116e-05,
      "loss": 0.5628,
      "step": 3971
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8014603853225708,
      "learning_rate": 1.9531675355391172e-05,
      "loss": 0.4522,
      "step": 3972
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9467196464538574,
      "learning_rate": 1.953136729504151e-05,
      "loss": 0.5946,
      "step": 3973
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0124857425689697,
      "learning_rate": 1.9531059135836327e-05,
      "loss": 0.5399,
      "step": 3974
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9418113231658936,
      "learning_rate": 1.9530750877778823e-05,
      "loss": 0.5725,
      "step": 3975
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8431074619293213,
      "learning_rate": 1.9530442520872188e-05,
      "loss": 0.4886,
      "step": 3976
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.868424654006958,
      "learning_rate": 1.9530134065119626e-05,
      "loss": 0.5449,
      "step": 3977
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.938189148902893,
      "learning_rate": 1.9529825510524337e-05,
      "loss": 0.5309,
      "step": 3978
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.079427480697632,
      "learning_rate": 1.9529516857089512e-05,
      "loss": 0.542,
      "step": 3979
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0507357120513916,
      "learning_rate": 1.9529208104818363e-05,
      "loss": 0.4528,
      "step": 3980
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9298008680343628,
      "learning_rate": 1.952889925371409e-05,
      "loss": 0.5817,
      "step": 3981
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0833680629730225,
      "learning_rate": 1.952859030377989e-05,
      "loss": 0.5505,
      "step": 3982
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0488693714141846,
      "learning_rate": 1.9528281255018973e-05,
      "loss": 0.5746,
      "step": 3983
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7557203769683838,
      "learning_rate": 1.9527972107434542e-05,
      "loss": 0.4622,
      "step": 3984
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.949953556060791,
      "learning_rate": 1.9527662861029806e-05,
      "loss": 0.6327,
      "step": 3985
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.922270655632019,
      "learning_rate": 1.952735351580797e-05,
      "loss": 0.4884,
      "step": 3986
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.165374755859375,
      "learning_rate": 1.9527044071772244e-05,
      "loss": 0.6191,
      "step": 3987
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8598136901855469,
      "learning_rate": 1.9526734528925838e-05,
      "loss": 0.5065,
      "step": 3988
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9557539224624634,
      "learning_rate": 1.9526424887271958e-05,
      "loss": 0.5899,
      "step": 3989
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.043344736099243,
      "learning_rate": 1.952611514681382e-05,
      "loss": 0.4776,
      "step": 3990
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9152897596359253,
      "learning_rate": 1.952580530755463e-05,
      "loss": 0.5175,
      "step": 3991
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8715858459472656,
      "learning_rate": 1.9525495369497612e-05,
      "loss": 0.5424,
      "step": 3992
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9228465557098389,
      "learning_rate": 1.9525185332645976e-05,
      "loss": 0.5624,
      "step": 3993
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.002700090408325,
      "learning_rate": 1.952487519700293e-05,
      "loss": 0.4845,
      "step": 3994
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.95978844165802,
      "learning_rate": 1.9524564962571705e-05,
      "loss": 0.5353,
      "step": 3995
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9388771057128906,
      "learning_rate": 1.9524254629355505e-05,
      "loss": 0.5093,
      "step": 3996
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8840296268463135,
      "learning_rate": 1.9523944197357557e-05,
      "loss": 0.5259,
      "step": 3997
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.913177251815796,
      "learning_rate": 1.9523633666581076e-05,
      "loss": 0.5093,
      "step": 3998
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8727645874023438,
      "learning_rate": 1.9523323037029285e-05,
      "loss": 0.6247,
      "step": 3999
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7956979274749756,
      "learning_rate": 1.952301230870541e-05,
      "loss": 0.4379,
      "step": 4000
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.354975938796997,
      "learning_rate": 1.9522701481612666e-05,
      "loss": 0.883,
      "step": 4001
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2368063926696777,
      "learning_rate": 1.9522390555754275e-05,
      "loss": 0.9324,
      "step": 4002
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0989809036254883,
      "learning_rate": 1.9522079531133477e-05,
      "loss": 0.6138,
      "step": 4003
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.309490442276001,
      "learning_rate": 1.9521768407753478e-05,
      "loss": 0.5212,
      "step": 4004
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9061745405197144,
      "learning_rate": 1.952145718561752e-05,
      "loss": 0.8681,
      "step": 4005
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.641380786895752,
      "learning_rate": 1.9521145864728826e-05,
      "loss": 0.867,
      "step": 4006
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1871464252471924,
      "learning_rate": 1.952083444509062e-05,
      "loss": 0.6539,
      "step": 4007
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.007891893386841,
      "learning_rate": 1.952052292670614e-05,
      "loss": 0.4942,
      "step": 4008
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9536666870117188,
      "learning_rate": 1.9520211309578613e-05,
      "loss": 0.6037,
      "step": 4009
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9353960752487183,
      "learning_rate": 1.951989959371127e-05,
      "loss": 0.4949,
      "step": 4010
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.900618076324463,
      "learning_rate": 1.9519587779107345e-05,
      "loss": 0.5612,
      "step": 4011
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8933004140853882,
      "learning_rate": 1.9519275865770075e-05,
      "loss": 0.5282,
      "step": 4012
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8781416416168213,
      "learning_rate": 1.9518963853702687e-05,
      "loss": 0.6246,
      "step": 4013
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9570740461349487,
      "learning_rate": 1.9518651742908426e-05,
      "loss": 0.5077,
      "step": 4014
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9345451593399048,
      "learning_rate": 1.9518339533390524e-05,
      "loss": 0.5978,
      "step": 4015
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7832207679748535,
      "learning_rate": 1.9518027225152222e-05,
      "loss": 0.4779,
      "step": 4016
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9568438529968262,
      "learning_rate": 1.9517714818196757e-05,
      "loss": 0.5539,
      "step": 4017
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0022618770599365,
      "learning_rate": 1.9517402312527367e-05,
      "loss": 0.556,
      "step": 4018
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8693898916244507,
      "learning_rate": 1.95170897081473e-05,
      "loss": 0.5644,
      "step": 4019
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7715526819229126,
      "learning_rate": 1.951677700505979e-05,
      "loss": 0.4583,
      "step": 4020
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7698472738265991,
      "learning_rate": 1.9516464203268087e-05,
      "loss": 0.5543,
      "step": 4021
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8848068714141846,
      "learning_rate": 1.9516151302775434e-05,
      "loss": 0.5261,
      "step": 4022
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9421788454055786,
      "learning_rate": 1.9515838303585073e-05,
      "loss": 0.6286,
      "step": 4023
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9400317668914795,
      "learning_rate": 1.951552520570025e-05,
      "loss": 0.5353,
      "step": 4024
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.935871958732605,
      "learning_rate": 1.9515212009124217e-05,
      "loss": 0.5496,
      "step": 4025
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8231655359268188,
      "learning_rate": 1.951489871386022e-05,
      "loss": 0.498,
      "step": 4026
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.8773114681243896,
      "learning_rate": 1.951458531991151e-05,
      "loss": 0.8986,
      "step": 4027
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.9658570289611816,
      "learning_rate": 1.951427182728133e-05,
      "loss": 1.0446,
      "step": 4028
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.148512125015259,
      "learning_rate": 1.9513958235972942e-05,
      "loss": 0.6133,
      "step": 4029
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0262022018432617,
      "learning_rate": 1.951364454598959e-05,
      "loss": 0.5061,
      "step": 4030
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1724965572357178,
      "learning_rate": 1.9513330757334534e-05,
      "loss": 0.6158,
      "step": 4031
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8652065992355347,
      "learning_rate": 1.9513016870011025e-05,
      "loss": 0.5193,
      "step": 4032
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8578978776931763,
      "learning_rate": 1.951270288402232e-05,
      "loss": 0.5678,
      "step": 4033
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8756628036499023,
      "learning_rate": 1.9512388799371674e-05,
      "loss": 0.5154,
      "step": 4034
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.867260456085205,
      "learning_rate": 1.951207461606234e-05,
      "loss": 0.6019,
      "step": 4035
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8542728424072266,
      "learning_rate": 1.951176033409759e-05,
      "loss": 0.57,
      "step": 4036
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.043715000152588,
      "learning_rate": 1.951144595348067e-05,
      "loss": 0.6275,
      "step": 4037
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9268391132354736,
      "learning_rate": 1.9511131474214846e-05,
      "loss": 0.57,
      "step": 4038
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9682592153549194,
      "learning_rate": 1.951081689630338e-05,
      "loss": 0.557,
      "step": 4039
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.90066397190094,
      "learning_rate": 1.9510502219749533e-05,
      "loss": 0.5415,
      "step": 4040
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.109206199645996,
      "learning_rate": 1.951018744455657e-05,
      "loss": 0.6717,
      "step": 4041
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.052248954772949,
      "learning_rate": 1.9509872570727762e-05,
      "loss": 0.5765,
      "step": 4042
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0012869834899902,
      "learning_rate": 1.9509557598266363e-05,
      "loss": 0.6226,
      "step": 4043
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8047258853912354,
      "learning_rate": 1.9509242527175645e-05,
      "loss": 0.4822,
      "step": 4044
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0709598064422607,
      "learning_rate": 1.9508927357458874e-05,
      "loss": 0.6017,
      "step": 4045
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.061821699142456,
      "learning_rate": 1.9508612089119326e-05,
      "loss": 0.5073,
      "step": 4046
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9726699590682983,
      "learning_rate": 1.9508296722160264e-05,
      "loss": 0.59,
      "step": 4047
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8455817699432373,
      "learning_rate": 1.9507981256584957e-05,
      "loss": 0.5087,
      "step": 4048
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8626372814178467,
      "learning_rate": 1.9507665692396682e-05,
      "loss": 0.581,
      "step": 4049
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1143147945404053,
      "learning_rate": 1.950735002959871e-05,
      "loss": 0.486,
      "step": 4050
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0818934440612793,
      "learning_rate": 1.950703426819432e-05,
      "loss": 0.5748,
      "step": 4051
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8598849773406982,
      "learning_rate": 1.9506718408186776e-05,
      "loss": 0.4595,
      "step": 4052
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.938377022743225,
      "learning_rate": 1.950640244957936e-05,
      "loss": 0.601,
      "step": 4053
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8814303874969482,
      "learning_rate": 1.9506086392375353e-05,
      "loss": 0.4704,
      "step": 4054
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8886222839355469,
      "learning_rate": 1.950577023657803e-05,
      "loss": 0.5336,
      "step": 4055
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8016843795776367,
      "learning_rate": 1.9505453982190667e-05,
      "loss": 0.5432,
      "step": 4056
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8586235046386719,
      "learning_rate": 1.9505137629216543e-05,
      "loss": 0.598,
      "step": 4057
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.827128529548645,
      "learning_rate": 1.9504821177658944e-05,
      "loss": 0.4786,
      "step": 4058
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9032541513442993,
      "learning_rate": 1.9504504627521154e-05,
      "loss": 0.5967,
      "step": 4059
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7723280191421509,
      "learning_rate": 1.9504187978806452e-05,
      "loss": 0.4673,
      "step": 4060
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8622806072235107,
      "learning_rate": 1.950387123151812e-05,
      "loss": 0.5058,
      "step": 4061
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8339834213256836,
      "learning_rate": 1.9503554385659446e-05,
      "loss": 0.5166,
      "step": 4062
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.028552532196045,
      "learning_rate": 1.9503237441233717e-05,
      "loss": 0.6436,
      "step": 4063
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8993059396743774,
      "learning_rate": 1.950292039824422e-05,
      "loss": 0.5012,
      "step": 4064
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.866043210029602,
      "learning_rate": 1.9502603256694243e-05,
      "loss": 0.6096,
      "step": 4065
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8871856927871704,
      "learning_rate": 1.950228601658707e-05,
      "loss": 0.5106,
      "step": 4066
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8838229179382324,
      "learning_rate": 1.9501968677926e-05,
      "loss": 0.5778,
      "step": 4067
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.05790638923645,
      "learning_rate": 1.950165124071432e-05,
      "loss": 0.5705,
      "step": 4068
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9026132822036743,
      "learning_rate": 1.9501333704955323e-05,
      "loss": 0.5679,
      "step": 4069
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9025472402572632,
      "learning_rate": 1.9501016070652305e-05,
      "loss": 0.5401,
      "step": 4070
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9566283226013184,
      "learning_rate": 1.9500698337808553e-05,
      "loss": 0.6176,
      "step": 4071
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8014273643493652,
      "learning_rate": 1.9500380506427364e-05,
      "loss": 0.465,
      "step": 4072
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.295957326889038,
      "learning_rate": 1.9500062576512043e-05,
      "loss": 0.925,
      "step": 4073
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9687217473983765,
      "learning_rate": 1.949974454806588e-05,
      "loss": 0.8661,
      "step": 4074
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1427388191223145,
      "learning_rate": 1.9499426421092177e-05,
      "loss": 0.6051,
      "step": 4075
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9324911832809448,
      "learning_rate": 1.949910819559423e-05,
      "loss": 0.5639,
      "step": 4076
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.014993906021118,
      "learning_rate": 1.949878987157534e-05,
      "loss": 0.6025,
      "step": 4077
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.96678626537323,
      "learning_rate": 1.949847144903881e-05,
      "loss": 0.5122,
      "step": 4078
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9509460926055908,
      "learning_rate": 1.9498152927987943e-05,
      "loss": 0.5736,
      "step": 4079
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.851100206375122,
      "learning_rate": 1.9497834308426043e-05,
      "loss": 0.5078,
      "step": 4080
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.901650071144104,
      "learning_rate": 1.949751559035641e-05,
      "loss": 0.5438,
      "step": 4081
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8507333993911743,
      "learning_rate": 1.9497196773782357e-05,
      "loss": 0.4868,
      "step": 4082
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9362343549728394,
      "learning_rate": 1.9496877858707182e-05,
      "loss": 0.5649,
      "step": 4083
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8396084308624268,
      "learning_rate": 1.9496558845134205e-05,
      "loss": 0.4691,
      "step": 4084
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2510647773742676,
      "learning_rate": 1.9496239733066722e-05,
      "loss": 0.6678,
      "step": 4085
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9482635259628296,
      "learning_rate": 1.9495920522508047e-05,
      "loss": 0.4784,
      "step": 4086
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.894971251487732,
      "learning_rate": 1.949560121346149e-05,
      "loss": 0.5837,
      "step": 4087
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8481355905532837,
      "learning_rate": 1.949528180593037e-05,
      "loss": 0.5316,
      "step": 4088
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.034115791320801,
      "learning_rate": 1.949496229991799e-05,
      "loss": 0.6887,
      "step": 4089
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.817619800567627,
      "learning_rate": 1.949464269542767e-05,
      "loss": 0.4694,
      "step": 4090
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7936437129974365,
      "learning_rate": 1.9494322992462718e-05,
      "loss": 0.5361,
      "step": 4091
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8205598592758179,
      "learning_rate": 1.9494003191026457e-05,
      "loss": 0.5807,
      "step": 4092
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8563435077667236,
      "learning_rate": 1.94936832911222e-05,
      "loss": 0.6002,
      "step": 4093
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.918900728225708,
      "learning_rate": 1.949336329275327e-05,
      "loss": 0.5267,
      "step": 4094
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9441711902618408,
      "learning_rate": 1.949304319592298e-05,
      "loss": 0.5107,
      "step": 4095
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.920702576637268,
      "learning_rate": 1.949272300063465e-05,
      "loss": 0.6184,
      "step": 4096
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.148287534713745,
      "learning_rate": 1.9492402706891606e-05,
      "loss": 0.5361,
      "step": 4097
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9358028173446655,
      "learning_rate": 1.949208231469716e-05,
      "loss": 0.556,
      "step": 4098
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8124750852584839,
      "learning_rate": 1.9491761824054653e-05,
      "loss": 0.5166,
      "step": 4099
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7759937047958374,
      "learning_rate": 1.9491441234967387e-05,
      "loss": 0.5315,
      "step": 4100
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8906594514846802,
      "learning_rate": 1.9491120547438703e-05,
      "loss": 0.5351,
      "step": 4101
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9073997735977173,
      "learning_rate": 1.9490799761471924e-05,
      "loss": 0.5176,
      "step": 4102
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9055598974227905,
      "learning_rate": 1.949047887707037e-05,
      "loss": 0.5946,
      "step": 4103
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.020437717437744,
      "learning_rate": 1.949015789423738e-05,
      "loss": 0.5607,
      "step": 4104
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.151731014251709,
      "learning_rate": 1.948983681297627e-05,
      "loss": 0.7098,
      "step": 4105
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.200974225997925,
      "learning_rate": 1.9489515633290378e-05,
      "loss": 0.4786,
      "step": 4106
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9068459272384644,
      "learning_rate": 1.948919435518304e-05,
      "loss": 0.529,
      "step": 4107
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8037669658660889,
      "learning_rate": 1.948887297865758e-05,
      "loss": 0.4311,
      "step": 4108
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.761993646621704,
      "learning_rate": 1.9488551503717332e-05,
      "loss": 0.5264,
      "step": 4109
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9669831991195679,
      "learning_rate": 1.948822993036563e-05,
      "loss": 0.56,
      "step": 4110
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9077658653259277,
      "learning_rate": 1.9487908258605815e-05,
      "loss": 0.6218,
      "step": 4111
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8187130689620972,
      "learning_rate": 1.948758648844122e-05,
      "loss": 0.4776,
      "step": 4112
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0057504177093506,
      "learning_rate": 1.948726461987518e-05,
      "loss": 0.599,
      "step": 4113
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.852292776107788,
      "learning_rate": 1.9486942652911037e-05,
      "loss": 0.4682,
      "step": 4114
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8028780221939087,
      "learning_rate": 1.9486620587552124e-05,
      "loss": 0.5642,
      "step": 4115
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8500951528549194,
      "learning_rate": 1.948629842380179e-05,
      "loss": 0.5776,
      "step": 4116
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8565675020217896,
      "learning_rate": 1.948597616166337e-05,
      "loss": 0.5291,
      "step": 4117
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.976216435432434,
      "learning_rate": 1.948565380114021e-05,
      "loss": 0.5696,
      "step": 4118
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8614822626113892,
      "learning_rate": 1.9485331342235652e-05,
      "loss": 0.5814,
      "step": 4119
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8925813436508179,
      "learning_rate": 1.9485008784953038e-05,
      "loss": 0.4582,
      "step": 4120
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.878674864768982,
      "learning_rate": 1.9484686129295715e-05,
      "loss": 0.5803,
      "step": 4121
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8461941480636597,
      "learning_rate": 1.9484363375267032e-05,
      "loss": 0.4876,
      "step": 4122
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.038681983947754,
      "learning_rate": 1.9484040522870333e-05,
      "loss": 0.6788,
      "step": 4123
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8219434022903442,
      "learning_rate": 1.948371757210897e-05,
      "loss": 0.4666,
      "step": 4124
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9942206144332886,
      "learning_rate": 1.9483394522986288e-05,
      "loss": 0.6246,
      "step": 4125
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8965269327163696,
      "learning_rate": 1.9483071375505644e-05,
      "loss": 0.4858,
      "step": 4126
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9685148000717163,
      "learning_rate": 1.948274812967038e-05,
      "loss": 0.5463,
      "step": 4127
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.924296498298645,
      "learning_rate": 1.948242478548386e-05,
      "loss": 0.5132,
      "step": 4128
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.895729422569275,
      "learning_rate": 1.9482101342949428e-05,
      "loss": 0.5653,
      "step": 4129
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.709655523300171,
      "learning_rate": 1.9481777802070444e-05,
      "loss": 0.4582,
      "step": 4130
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9256339073181152,
      "learning_rate": 1.948145416285026e-05,
      "loss": 0.6205,
      "step": 4131
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7708100080490112,
      "learning_rate": 1.9481130425292235e-05,
      "loss": 0.4766,
      "step": 4132
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.813458800315857,
      "learning_rate": 1.948080658939973e-05,
      "loss": 0.5799,
      "step": 4133
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9165270328521729,
      "learning_rate": 1.9480482655176097e-05,
      "loss": 0.5025,
      "step": 4134
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8756130933761597,
      "learning_rate": 1.9480158622624702e-05,
      "loss": 0.5577,
      "step": 4135
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8394776582717896,
      "learning_rate": 1.94798344917489e-05,
      "loss": 0.5964,
      "step": 4136
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8717719316482544,
      "learning_rate": 1.9479510262552054e-05,
      "loss": 0.5181,
      "step": 4137
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7946710586547852,
      "learning_rate": 1.947918593503753e-05,
      "loss": 0.5247,
      "step": 4138
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.881024718284607,
      "learning_rate": 1.9478861509208687e-05,
      "loss": 0.5412,
      "step": 4139
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9587814807891846,
      "learning_rate": 1.9478536985068897e-05,
      "loss": 0.5558,
      "step": 4140
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9768829345703125,
      "learning_rate": 1.947821236262152e-05,
      "loss": 0.6102,
      "step": 4141
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8678836822509766,
      "learning_rate": 1.9477887641869922e-05,
      "loss": 0.5099,
      "step": 4142
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7978211641311646,
      "learning_rate": 1.947756282281748e-05,
      "loss": 0.489,
      "step": 4143
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2019453048706055,
      "learning_rate": 1.9477237905467546e-05,
      "loss": 0.6405,
      "step": 4144
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.932136058807373,
      "learning_rate": 1.947691288982351e-05,
      "loss": 0.5732,
      "step": 4145
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8182506561279297,
      "learning_rate": 1.9476587775888726e-05,
      "loss": 0.5134,
      "step": 4146
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.069359302520752,
      "learning_rate": 1.9476262563666573e-05,
      "loss": 0.595,
      "step": 4147
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8338556289672852,
      "learning_rate": 1.9475937253160427e-05,
      "loss": 0.4876,
      "step": 4148
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.001422166824341,
      "learning_rate": 1.9475611844373657e-05,
      "loss": 0.6279,
      "step": 4149
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8275189399719238,
      "learning_rate": 1.9475286337309646e-05,
      "loss": 0.5289,
      "step": 4150
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8870095014572144,
      "learning_rate": 1.9474960731971758e-05,
      "loss": 0.6301,
      "step": 4151
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7987151145935059,
      "learning_rate": 1.9474635028363377e-05,
      "loss": 0.4992,
      "step": 4152
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.962246298789978,
      "learning_rate": 1.947430922648788e-05,
      "loss": 0.5986,
      "step": 4153
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8644682168960571,
      "learning_rate": 1.9473983326348646e-05,
      "loss": 0.542,
      "step": 4154
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9026036262512207,
      "learning_rate": 1.9473657327949055e-05,
      "loss": 0.6457,
      "step": 4155
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8631030321121216,
      "learning_rate": 1.9473331231292487e-05,
      "loss": 0.485,
      "step": 4156
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9378821849822998,
      "learning_rate": 1.947300503638233e-05,
      "loss": 0.6095,
      "step": 4157
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.848816156387329,
      "learning_rate": 1.947267874322196e-05,
      "loss": 0.4896,
      "step": 4158
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9207271337509155,
      "learning_rate": 1.9472352351814764e-05,
      "loss": 0.6289,
      "step": 4159
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.033891201019287,
      "learning_rate": 1.9472025862164125e-05,
      "loss": 0.4893,
      "step": 4160
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8452597856521606,
      "learning_rate": 1.947169927427343e-05,
      "loss": 0.6037,
      "step": 4161
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8857364654541016,
      "learning_rate": 1.9471372588146073e-05,
      "loss": 0.5263,
      "step": 4162
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9392671585083008,
      "learning_rate": 1.9471045803785434e-05,
      "loss": 0.6345,
      "step": 4163
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8422536849975586,
      "learning_rate": 1.9470718921194906e-05,
      "loss": 0.5078,
      "step": 4164
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9053016901016235,
      "learning_rate": 1.9470391940377877e-05,
      "loss": 0.5519,
      "step": 4165
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.860049843788147,
      "learning_rate": 1.947006486133774e-05,
      "loss": 0.4695,
      "step": 4166
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9025861024856567,
      "learning_rate": 1.9469737684077888e-05,
      "loss": 0.5646,
      "step": 4167
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9654209613800049,
      "learning_rate": 1.9469410408601712e-05,
      "loss": 0.5299,
      "step": 4168
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9095649719238281,
      "learning_rate": 1.9469083034912607e-05,
      "loss": 0.5926,
      "step": 4169
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7369778156280518,
      "learning_rate": 1.9468755563013973e-05,
      "loss": 0.465,
      "step": 4170
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9830313920974731,
      "learning_rate": 1.94684279929092e-05,
      "loss": 0.5973,
      "step": 4171
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7676888704299927,
      "learning_rate": 1.9468100324601688e-05,
      "loss": 0.4673,
      "step": 4172
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.772761344909668,
      "learning_rate": 1.9467772558094834e-05,
      "loss": 0.6035,
      "step": 4173
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7659674882888794,
      "learning_rate": 1.9467444693392043e-05,
      "loss": 0.4782,
      "step": 4174
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9227187633514404,
      "learning_rate": 1.9467116730496707e-05,
      "loss": 0.5702,
      "step": 4175
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9980231523513794,
      "learning_rate": 1.9466788669412234e-05,
      "loss": 0.5905,
      "step": 4176
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9551963806152344,
      "learning_rate": 1.9466460510142023e-05,
      "loss": 0.577,
      "step": 4177
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8849321603775024,
      "learning_rate": 1.946613225268948e-05,
      "loss": 0.4984,
      "step": 4178
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8937132358551025,
      "learning_rate": 1.9465803897058006e-05,
      "loss": 0.646,
      "step": 4179
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8274552822113037,
      "learning_rate": 1.9465475443251013e-05,
      "loss": 0.5062,
      "step": 4180
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8266408443450928,
      "learning_rate": 1.94651468912719e-05,
      "loss": 0.5433,
      "step": 4181
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7731802463531494,
      "learning_rate": 1.9464818241124083e-05,
      "loss": 0.5356,
      "step": 4182
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9632266759872437,
      "learning_rate": 1.9464489492810958e-05,
      "loss": 0.6194,
      "step": 4183
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9006335735321045,
      "learning_rate": 1.9464160646335947e-05,
      "loss": 0.484,
      "step": 4184
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8981555700302124,
      "learning_rate": 1.946383170170246e-05,
      "loss": 0.5268,
      "step": 4185
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.875251293182373,
      "learning_rate": 1.94635026589139e-05,
      "loss": 0.5656,
      "step": 4186
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9266186952590942,
      "learning_rate": 1.9463173517973685e-05,
      "loss": 0.5753,
      "step": 4187
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9799953699111938,
      "learning_rate": 1.946284427888523e-05,
      "loss": 0.5007,
      "step": 4188
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8532557487487793,
      "learning_rate": 1.9462514941651945e-05,
      "loss": 0.5587,
      "step": 4189
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8917443752288818,
      "learning_rate": 1.9462185506277248e-05,
      "loss": 0.5597,
      "step": 4190
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.884861707687378,
      "learning_rate": 1.946185597276456e-05,
      "loss": 0.5883,
      "step": 4191
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8585824966430664,
      "learning_rate": 1.9461526341117296e-05,
      "loss": 0.4767,
      "step": 4192
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8682749271392822,
      "learning_rate": 1.9461196611338872e-05,
      "loss": 0.6171,
      "step": 4193
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8801004886627197,
      "learning_rate": 1.946086678343271e-05,
      "loss": 0.4558,
      "step": 4194
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9221258163452148,
      "learning_rate": 1.946053685740223e-05,
      "loss": 0.5973,
      "step": 4195
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9277284145355225,
      "learning_rate": 1.9460206833250857e-05,
      "loss": 0.4769,
      "step": 4196
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8627201318740845,
      "learning_rate": 1.9459876710982008e-05,
      "loss": 0.5332,
      "step": 4197
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9207994937896729,
      "learning_rate": 1.9459546490599113e-05,
      "loss": 0.5565,
      "step": 4198
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.96214759349823,
      "learning_rate": 1.9459216172105594e-05,
      "loss": 0.603,
      "step": 4199
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9512925148010254,
      "learning_rate": 1.945888575550488e-05,
      "loss": 0.4973,
      "step": 4200
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.830199122428894,
      "learning_rate": 1.9458555240800388e-05,
      "loss": 0.5406,
      "step": 4201
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.210723400115967,
      "learning_rate": 1.945822462799556e-05,
      "loss": 0.5242,
      "step": 4202
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.967226505279541,
      "learning_rate": 1.9457893917093817e-05,
      "loss": 0.5713,
      "step": 4203
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.775134563446045,
      "learning_rate": 1.9457563108098586e-05,
      "loss": 0.4656,
      "step": 4204
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.297710657119751,
      "learning_rate": 1.9457232201013304e-05,
      "loss": 0.8554,
      "step": 4205
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8520628213882446,
      "learning_rate": 1.9456901195841403e-05,
      "loss": 0.9494,
      "step": 4206
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.236083745956421,
      "learning_rate": 1.9456570092586316e-05,
      "loss": 0.6167,
      "step": 4207
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.144380569458008,
      "learning_rate": 1.945623889125147e-05,
      "loss": 0.5,
      "step": 4208
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.927977442741394,
      "learning_rate": 1.9455907591840308e-05,
      "loss": 0.5532,
      "step": 4209
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8689697980880737,
      "learning_rate": 1.9455576194356265e-05,
      "loss": 0.4936,
      "step": 4210
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.038900375366211,
      "learning_rate": 1.9455244698802774e-05,
      "loss": 0.628,
      "step": 4211
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.958058476448059,
      "learning_rate": 1.945491310518328e-05,
      "loss": 0.4653,
      "step": 4212
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.988099217414856,
      "learning_rate": 1.9454581413501213e-05,
      "loss": 0.5916,
      "step": 4213
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.01764178276062,
      "learning_rate": 1.9454249623760017e-05,
      "loss": 0.5034,
      "step": 4214
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8853133916854858,
      "learning_rate": 1.945391773596314e-05,
      "loss": 0.6261,
      "step": 4215
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8643983602523804,
      "learning_rate": 1.9453585750114016e-05,
      "loss": 0.4753,
      "step": 4216
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8672409057617188,
      "learning_rate": 1.945325366621609e-05,
      "loss": 0.5432,
      "step": 4217
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.940973162651062,
      "learning_rate": 1.9452921484272807e-05,
      "loss": 0.5321,
      "step": 4218
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8769797086715698,
      "learning_rate": 1.9452589204287614e-05,
      "loss": 0.6065,
      "step": 4219
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9527819156646729,
      "learning_rate": 1.9452256826263953e-05,
      "loss": 0.5235,
      "step": 4220
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9478284120559692,
      "learning_rate": 1.9451924350205272e-05,
      "loss": 0.6286,
      "step": 4221
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8947043418884277,
      "learning_rate": 1.945159177611503e-05,
      "loss": 0.4903,
      "step": 4222
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.968814492225647,
      "learning_rate": 1.9451259103996658e-05,
      "loss": 0.5562,
      "step": 4223
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7988390922546387,
      "learning_rate": 1.945092633385362e-05,
      "loss": 0.4853,
      "step": 4224
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8645451068878174,
      "learning_rate": 1.9450593465689364e-05,
      "loss": 0.5432,
      "step": 4225
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9449251890182495,
      "learning_rate": 1.945026049950734e-05,
      "loss": 0.5633,
      "step": 4226
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9267204999923706,
      "learning_rate": 1.9449927435311e-05,
      "loss": 0.604,
      "step": 4227
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7966227531433105,
      "learning_rate": 1.9449594273103803e-05,
      "loss": 0.4823,
      "step": 4228
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0835912227630615,
      "learning_rate": 1.9449261012889206e-05,
      "loss": 0.6022,
      "step": 4229
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8530774116516113,
      "learning_rate": 1.944892765467066e-05,
      "loss": 0.4744,
      "step": 4230
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.857297658920288,
      "learning_rate": 1.9448594198451626e-05,
      "loss": 0.6155,
      "step": 4231
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8797109127044678,
      "learning_rate": 1.9448260644235555e-05,
      "loss": 0.458,
      "step": 4232
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9199578762054443,
      "learning_rate": 1.944792699202592e-05,
      "loss": 0.6288,
      "step": 4233
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9492034912109375,
      "learning_rate": 1.944759324182617e-05,
      "loss": 0.5096,
      "step": 4234
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0530316829681396,
      "learning_rate": 1.9447259393639772e-05,
      "loss": 0.6159,
      "step": 4235
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8079079389572144,
      "learning_rate": 1.9446925447470186e-05,
      "loss": 0.4731,
      "step": 4236
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8760358095169067,
      "learning_rate": 1.944659140332088e-05,
      "loss": 0.5884,
      "step": 4237
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8278944492340088,
      "learning_rate": 1.944625726119531e-05,
      "loss": 0.4857,
      "step": 4238
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.08479380607605,
      "learning_rate": 1.9445923021096954e-05,
      "loss": 0.5283,
      "step": 4239
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9635239839553833,
      "learning_rate": 1.9445588683029265e-05,
      "loss": 0.555,
      "step": 4240
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9957451820373535,
      "learning_rate": 1.944525424699572e-05,
      "loss": 0.5833,
      "step": 4241
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8250764608383179,
      "learning_rate": 1.9444919712999785e-05,
      "loss": 0.4724,
      "step": 4242
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8434127569198608,
      "learning_rate": 1.9444585081044928e-05,
      "loss": 0.5383,
      "step": 4243
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.861030101776123,
      "learning_rate": 1.9444250351134624e-05,
      "loss": 0.4811,
      "step": 4244
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.947570562362671,
      "learning_rate": 1.944391552327234e-05,
      "loss": 0.6091,
      "step": 4245
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8645155429840088,
      "learning_rate": 1.9443580597461547e-05,
      "loss": 0.4967,
      "step": 4246
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9628928899765015,
      "learning_rate": 1.9443245573705725e-05,
      "loss": 0.6125,
      "step": 4247
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8483400344848633,
      "learning_rate": 1.9442910452008345e-05,
      "loss": 0.4906,
      "step": 4248
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8524246215820312,
      "learning_rate": 1.9442575232372885e-05,
      "loss": 0.4805,
      "step": 4249
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8224258422851562,
      "learning_rate": 1.9442239914802822e-05,
      "loss": 0.5478,
      "step": 4250
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.945082664489746,
      "learning_rate": 1.9441904499301628e-05,
      "loss": 0.6384,
      "step": 4251
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.912850260734558,
      "learning_rate": 1.944156898587279e-05,
      "loss": 0.4684,
      "step": 4252
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8519024848937988,
      "learning_rate": 1.9441233374519778e-05,
      "loss": 0.5709,
      "step": 4253
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8817497491836548,
      "learning_rate": 1.9440897665246084e-05,
      "loss": 0.5143,
      "step": 4254
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8980154991149902,
      "learning_rate": 1.9440561858055185e-05,
      "loss": 0.5612,
      "step": 4255
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8121483325958252,
      "learning_rate": 1.944022595295056e-05,
      "loss": 0.4604,
      "step": 4256
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.929325819015503,
      "learning_rate": 1.9439889949935697e-05,
      "loss": 0.6213,
      "step": 4257
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8803974390029907,
      "learning_rate": 1.943955384901408e-05,
      "loss": 0.537,
      "step": 4258
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8071414232254028,
      "learning_rate": 1.9439217650189194e-05,
      "loss": 0.5391,
      "step": 4259
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8786622285842896,
      "learning_rate": 1.943888135346453e-05,
      "loss": 0.489,
      "step": 4260
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.929362177848816,
      "learning_rate": 1.9438544958843574e-05,
      "loss": 0.6416,
      "step": 4261
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7936351299285889,
      "learning_rate": 1.943820846632981e-05,
      "loss": 0.4417,
      "step": 4262
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9807180166244507,
      "learning_rate": 1.943787187592673e-05,
      "loss": 0.5897,
      "step": 4263
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.731611728668213,
      "learning_rate": 1.9437535187637832e-05,
      "loss": 0.4533,
      "step": 4264
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.019266366958618,
      "learning_rate": 1.94371984014666e-05,
      "loss": 0.6047,
      "step": 4265
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9840772151947021,
      "learning_rate": 1.9436861517416527e-05,
      "loss": 0.4882,
      "step": 4266
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8355355262756348,
      "learning_rate": 1.9436524535491114e-05,
      "loss": 0.5536,
      "step": 4267
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9082201719284058,
      "learning_rate": 1.9436187455693852e-05,
      "loss": 0.5128,
      "step": 4268
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.94728422164917,
      "learning_rate": 1.9435850278028236e-05,
      "loss": 0.5772,
      "step": 4269
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9054776430130005,
      "learning_rate": 1.943551300249776e-05,
      "loss": 0.5912,
      "step": 4270
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.015310049057007,
      "learning_rate": 1.943517562910593e-05,
      "loss": 0.5684,
      "step": 4271
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.896301507949829,
      "learning_rate": 1.943483815785624e-05,
      "loss": 0.5573,
      "step": 4272
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9678261280059814,
      "learning_rate": 1.9434500588752193e-05,
      "loss": 0.5967,
      "step": 4273
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8733471632003784,
      "learning_rate": 1.9434162921797287e-05,
      "loss": 0.5342,
      "step": 4274
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9821223020553589,
      "learning_rate": 1.9433825156995024e-05,
      "loss": 0.5809,
      "step": 4275
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9624183177947998,
      "learning_rate": 1.943348729434891e-05,
      "loss": 0.4824,
      "step": 4276
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.832872748374939,
      "learning_rate": 1.943314933386245e-05,
      "loss": 0.5355,
      "step": 4277
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9159669876098633,
      "learning_rate": 1.9432811275539145e-05,
      "loss": 0.4646,
      "step": 4278
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9075878858566284,
      "learning_rate": 1.94324731193825e-05,
      "loss": 0.5624,
      "step": 4279
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7838329076766968,
      "learning_rate": 1.943213486539603e-05,
      "loss": 0.4456,
      "step": 4280
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9368036985397339,
      "learning_rate": 1.9431796513583237e-05,
      "loss": 0.563,
      "step": 4281
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7100626230239868,
      "learning_rate": 1.9431458063947634e-05,
      "loss": 0.4601,
      "step": 4282
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8323287963867188,
      "learning_rate": 1.9431119516492725e-05,
      "loss": 0.5781,
      "step": 4283
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7886602878570557,
      "learning_rate": 1.943078087122203e-05,
      "loss": 0.5046,
      "step": 4284
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9378212690353394,
      "learning_rate": 1.9430442128139054e-05,
      "loss": 0.719,
      "step": 4285
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0686399936676025,
      "learning_rate": 1.9430103287247314e-05,
      "loss": 0.4875,
      "step": 4286
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.6937570571899414,
      "learning_rate": 1.9429764348550324e-05,
      "loss": 0.5899,
      "step": 4287
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8438947200775146,
      "learning_rate": 1.94294253120516e-05,
      "loss": 0.4675,
      "step": 4288
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7898424863815308,
      "learning_rate": 1.9429086177754653e-05,
      "loss": 0.5272,
      "step": 4289
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.768404483795166,
      "learning_rate": 1.9428746945663007e-05,
      "loss": 0.4799,
      "step": 4290
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7803388833999634,
      "learning_rate": 1.9428407615780177e-05,
      "loss": 0.5345,
      "step": 4291
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7998847961425781,
      "learning_rate": 1.9428068188109687e-05,
      "loss": 0.4917,
      "step": 4292
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9622702598571777,
      "learning_rate": 1.9427728662655047e-05,
      "loss": 0.5434,
      "step": 4293
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8579293489456177,
      "learning_rate": 1.942738903941979e-05,
      "loss": 0.5374,
      "step": 4294
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8602650165557861,
      "learning_rate": 1.942704931840743e-05,
      "loss": 0.5709,
      "step": 4295
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.953959584236145,
      "learning_rate": 1.9426709499621496e-05,
      "loss": 0.515,
      "step": 4296
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0580015182495117,
      "learning_rate": 1.942636958306551e-05,
      "loss": 0.587,
      "step": 4297
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.887880563735962,
      "learning_rate": 1.9426029568743e-05,
      "loss": 0.509,
      "step": 4298
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7626652717590332,
      "learning_rate": 1.9425689456657487e-05,
      "loss": 0.4903,
      "step": 4299
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.061350107192993,
      "learning_rate": 1.9425349246812505e-05,
      "loss": 0.5472,
      "step": 4300
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1300885677337646,
      "learning_rate": 1.9425008939211575e-05,
      "loss": 0.6717,
      "step": 4301
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9093105792999268,
      "learning_rate": 1.9424668533858236e-05,
      "loss": 0.5076,
      "step": 4302
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.874529242515564,
      "learning_rate": 1.942432803075601e-05,
      "loss": 0.6062,
      "step": 4303
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.97523033618927,
      "learning_rate": 1.9423987429908435e-05,
      "loss": 0.4675,
      "step": 4304
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0593059062957764,
      "learning_rate": 1.942364673131904e-05,
      "loss": 0.6116,
      "step": 4305
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.933225393295288,
      "learning_rate": 1.9423305934991357e-05,
      "loss": 0.4627,
      "step": 4306
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8972043991088867,
      "learning_rate": 1.942296504092892e-05,
      "loss": 0.5704,
      "step": 4307
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.002673387527466,
      "learning_rate": 1.942262404913528e-05,
      "loss": 0.5002,
      "step": 4308
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.052422285079956,
      "learning_rate": 1.9422282959613946e-05,
      "loss": 0.5663,
      "step": 4309
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0808582305908203,
      "learning_rate": 1.942194177236848e-05,
      "loss": 0.6203,
      "step": 4310
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9671398401260376,
      "learning_rate": 1.9421600487402406e-05,
      "loss": 0.5742,
      "step": 4311
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9807102680206299,
      "learning_rate": 1.9421259104719274e-05,
      "loss": 0.5299,
      "step": 4312
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.033586025238037,
      "learning_rate": 1.9420917624322617e-05,
      "loss": 0.854,
      "step": 4313
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8861885070800781,
      "learning_rate": 1.942057604621598e-05,
      "loss": 0.9123,
      "step": 4314
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.341092824935913,
      "learning_rate": 1.9420234370402907e-05,
      "loss": 0.6535,
      "step": 4315
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2528293132781982,
      "learning_rate": 1.941989259688694e-05,
      "loss": 0.4839,
      "step": 4316
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0098648071289062,
      "learning_rate": 1.9419550725671618e-05,
      "loss": 0.5637,
      "step": 4317
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0915961265563965,
      "learning_rate": 1.9419208756760496e-05,
      "loss": 0.5158,
      "step": 4318
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0430736541748047,
      "learning_rate": 1.9418866690157118e-05,
      "loss": 0.5677,
      "step": 4319
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.02004337310791,
      "learning_rate": 1.941852452586503e-05,
      "loss": 0.5266,
      "step": 4320
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.88844895362854,
      "learning_rate": 1.941818226388778e-05,
      "loss": 0.5294,
      "step": 4321
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0869030952453613,
      "learning_rate": 1.941783990422892e-05,
      "loss": 0.5565,
      "step": 4322
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9875097274780273,
      "learning_rate": 1.9417497446892e-05,
      "loss": 0.5818,
      "step": 4323
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.877027988433838,
      "learning_rate": 1.9417154891880576e-05,
      "loss": 0.4546,
      "step": 4324
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.963531494140625,
      "learning_rate": 1.9416812239198194e-05,
      "loss": 0.5649,
      "step": 4325
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9653483629226685,
      "learning_rate": 1.941646948884841e-05,
      "loss": 0.4729,
      "step": 4326
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8662569522857666,
      "learning_rate": 1.9416126640834784e-05,
      "loss": 0.5021,
      "step": 4327
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9710320234298706,
      "learning_rate": 1.9415783695160863e-05,
      "loss": 0.635,
      "step": 4328
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8645943403244019,
      "learning_rate": 1.941544065183021e-05,
      "loss": 0.5674,
      "step": 4329
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9184266328811646,
      "learning_rate": 1.9415097510846382e-05,
      "loss": 0.5034,
      "step": 4330
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9067904949188232,
      "learning_rate": 1.9414754272212937e-05,
      "loss": 0.6273,
      "step": 4331
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.803424596786499,
      "learning_rate": 1.941441093593343e-05,
      "loss": 0.4738,
      "step": 4332
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.831461787223816,
      "learning_rate": 1.9414067502011434e-05,
      "loss": 0.5541,
      "step": 4333
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0257537364959717,
      "learning_rate": 1.9413723970450503e-05,
      "loss": 0.5208,
      "step": 4334
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.830207109451294,
      "learning_rate": 1.94133803412542e-05,
      "loss": 0.543,
      "step": 4335
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9989376068115234,
      "learning_rate": 1.9413036614426093e-05,
      "loss": 0.5356,
      "step": 4336
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8054890632629395,
      "learning_rate": 1.941269278996974e-05,
      "loss": 0.5478,
      "step": 4337
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7729129791259766,
      "learning_rate": 1.9412348867888713e-05,
      "loss": 0.5087,
      "step": 4338
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7823089361190796,
      "learning_rate": 1.9412004848186576e-05,
      "loss": 0.5323,
      "step": 4339
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7476353645324707,
      "learning_rate": 1.94116607308669e-05,
      "loss": 0.4804,
      "step": 4340
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8018062114715576,
      "learning_rate": 1.9411316515933252e-05,
      "loss": 0.5347,
      "step": 4341
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8830134868621826,
      "learning_rate": 1.94109722033892e-05,
      "loss": 0.5565,
      "step": 4342
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.889418125152588,
      "learning_rate": 1.941062779323832e-05,
      "loss": 0.6652,
      "step": 4343
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8048683404922485,
      "learning_rate": 1.941028328548418e-05,
      "loss": 0.483,
      "step": 4344
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0149497985839844,
      "learning_rate": 1.9409938680130356e-05,
      "loss": 0.6259,
      "step": 4345
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.977652668952942,
      "learning_rate": 1.9409593977180416e-05,
      "loss": 0.5135,
      "step": 4346
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.10931658744812,
      "learning_rate": 1.9409249176637945e-05,
      "loss": 0.6668,
      "step": 4347
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.794225811958313,
      "learning_rate": 1.940890427850651e-05,
      "loss": 0.4477,
      "step": 4348
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7846050262451172,
      "learning_rate": 1.9408559282789695e-05,
      "loss": 0.5447,
      "step": 4349
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.974311351776123,
      "learning_rate": 1.9408214189491075e-05,
      "loss": 0.5839,
      "step": 4350
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9826303720474243,
      "learning_rate": 1.940786899861423e-05,
      "loss": 0.6001,
      "step": 4351
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8489961624145508,
      "learning_rate": 1.9407523710162736e-05,
      "loss": 0.5258,
      "step": 4352
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8873295783996582,
      "learning_rate": 1.9407178324140182e-05,
      "loss": 0.5577,
      "step": 4353
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8872675895690918,
      "learning_rate": 1.9406832840550143e-05,
      "loss": 0.4858,
      "step": 4354
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.4350385665893555,
      "learning_rate": 1.9406487259396206e-05,
      "loss": 0.8946,
      "step": 4355
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.062735080718994,
      "learning_rate": 1.9406141580681956e-05,
      "loss": 0.9408,
      "step": 4356
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.347614049911499,
      "learning_rate": 1.940579580441097e-05,
      "loss": 0.593,
      "step": 4357
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2016286849975586,
      "learning_rate": 1.940544993058685e-05,
      "loss": 0.457,
      "step": 4358
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0563085079193115,
      "learning_rate": 1.940510395921317e-05,
      "loss": 0.6002,
      "step": 4359
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.894607663154602,
      "learning_rate": 1.9404757890293522e-05,
      "loss": 0.4799,
      "step": 4360
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7923119068145752,
      "learning_rate": 1.9404411723831497e-05,
      "loss": 0.5941,
      "step": 4361
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8940662145614624,
      "learning_rate": 1.940406545983068e-05,
      "loss": 0.4945,
      "step": 4362
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9534319639205933,
      "learning_rate": 1.9403719098294674e-05,
      "loss": 0.5498,
      "step": 4363
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9769768714904785,
      "learning_rate": 1.9403372639227056e-05,
      "loss": 0.5261,
      "step": 4364
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8529739379882812,
      "learning_rate": 1.940302608263143e-05,
      "loss": 0.602,
      "step": 4365
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8700971603393555,
      "learning_rate": 1.940267942851139e-05,
      "loss": 0.4863,
      "step": 4366
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8352152109146118,
      "learning_rate": 1.9402332676870523e-05,
      "loss": 0.4885,
      "step": 4367
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.823143720626831,
      "learning_rate": 1.9401985827712435e-05,
      "loss": 0.5167,
      "step": 4368
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.95002019405365,
      "learning_rate": 1.940163888104072e-05,
      "loss": 0.5895,
      "step": 4369
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9826369285583496,
      "learning_rate": 1.9401291836858972e-05,
      "loss": 0.5317,
      "step": 4370
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9742990732192993,
      "learning_rate": 1.94009446951708e-05,
      "loss": 0.6638,
      "step": 4371
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8668168783187866,
      "learning_rate": 1.9400597455979794e-05,
      "loss": 0.4886,
      "step": 4372
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9209529161453247,
      "learning_rate": 1.9400250119289562e-05,
      "loss": 0.6149,
      "step": 4373
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8899966478347778,
      "learning_rate": 1.9399902685103703e-05,
      "loss": 0.4646,
      "step": 4374
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.962948203086853,
      "learning_rate": 1.9399555153425822e-05,
      "loss": 0.5941,
      "step": 4375
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8832498788833618,
      "learning_rate": 1.9399207524259524e-05,
      "loss": 0.4736,
      "step": 4376
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8622945547103882,
      "learning_rate": 1.9398859797608415e-05,
      "loss": 0.5406,
      "step": 4377
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8688255548477173,
      "learning_rate": 1.93985119734761e-05,
      "loss": 0.48,
      "step": 4378
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8794957399368286,
      "learning_rate": 1.9398164051866187e-05,
      "loss": 0.5446,
      "step": 4379
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0303776264190674,
      "learning_rate": 1.9397816032782282e-05,
      "loss": 0.5479,
      "step": 4380
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9165209531784058,
      "learning_rate": 1.9397467916228e-05,
      "loss": 0.6248,
      "step": 4381
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8399317264556885,
      "learning_rate": 1.9397119702206945e-05,
      "loss": 0.4971,
      "step": 4382
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.97161865234375,
      "learning_rate": 1.9396771390722734e-05,
      "loss": 0.6319,
      "step": 4383
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8109220266342163,
      "learning_rate": 1.939642298177898e-05,
      "loss": 0.4617,
      "step": 4384
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9700733423233032,
      "learning_rate": 1.939607447537929e-05,
      "loss": 0.5105,
      "step": 4385
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9642874002456665,
      "learning_rate": 1.9395725871527283e-05,
      "loss": 0.5472,
      "step": 4386
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.038161277770996,
      "learning_rate": 1.9395377170226576e-05,
      "loss": 0.6686,
      "step": 4387
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.897392749786377,
      "learning_rate": 1.9395028371480783e-05,
      "loss": 0.4746,
      "step": 4388
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1072170734405518,
      "learning_rate": 1.939467947529352e-05,
      "loss": 0.5787,
      "step": 4389
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9418033361434937,
      "learning_rate": 1.939433048166841e-05,
      "loss": 0.515,
      "step": 4390
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.710397720336914,
      "learning_rate": 1.939398139060907e-05,
      "loss": 0.9431,
      "step": 4391
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3134243488311768,
      "learning_rate": 1.9393632202119125e-05,
      "loss": 0.9484,
      "step": 4392
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1456191539764404,
      "learning_rate": 1.939328291620219e-05,
      "loss": 0.6176,
      "step": 4393
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.203986644744873,
      "learning_rate": 1.939293353286189e-05,
      "loss": 0.4759,
      "step": 4394
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9167375564575195,
      "learning_rate": 1.9392584052101848e-05,
      "loss": 0.5409,
      "step": 4395
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8922147750854492,
      "learning_rate": 1.939223447392569e-05,
      "loss": 0.482,
      "step": 4396
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.024137496948242,
      "learning_rate": 1.9391884798337046e-05,
      "loss": 0.6268,
      "step": 4397
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0080161094665527,
      "learning_rate": 1.9391535025339537e-05,
      "loss": 0.4678,
      "step": 4398
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.041029691696167,
      "learning_rate": 1.939118515493679e-05,
      "loss": 0.5749,
      "step": 4399
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.886048674583435,
      "learning_rate": 1.9390835187132434e-05,
      "loss": 0.524,
      "step": 4400
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9825748205184937,
      "learning_rate": 1.9390485121930104e-05,
      "loss": 0.595,
      "step": 4401
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0015814304351807,
      "learning_rate": 1.9390134959333428e-05,
      "loss": 0.4754,
      "step": 4402
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.053370952606201,
      "learning_rate": 1.9389784699346032e-05,
      "loss": 0.5453,
      "step": 4403
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0457310676574707,
      "learning_rate": 1.938943434197156e-05,
      "loss": 0.5768,
      "step": 4404
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.87328040599823,
      "learning_rate": 1.9389083887213633e-05,
      "loss": 0.5188,
      "step": 4405
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7545788288116455,
      "learning_rate": 1.9388733335075893e-05,
      "loss": 0.4813,
      "step": 4406
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8922009468078613,
      "learning_rate": 1.938838268556198e-05,
      "loss": 0.5366,
      "step": 4407
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9607462882995605,
      "learning_rate": 1.938803193867552e-05,
      "loss": 0.4963,
      "step": 4408
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8723138570785522,
      "learning_rate": 1.9387681094420157e-05,
      "loss": 0.5924,
      "step": 4409
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.926626205444336,
      "learning_rate": 1.9387330152799535e-05,
      "loss": 0.4945,
      "step": 4410
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.810889720916748,
      "learning_rate": 1.9386979113817283e-05,
      "loss": 0.5677,
      "step": 4411
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8108572959899902,
      "learning_rate": 1.9386627977477047e-05,
      "loss": 0.4776,
      "step": 4412
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2287418842315674,
      "learning_rate": 1.9386276743782472e-05,
      "loss": 0.6329,
      "step": 4413
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8775845766067505,
      "learning_rate": 1.9385925412737195e-05,
      "loss": 0.5034,
      "step": 4414
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8929712772369385,
      "learning_rate": 1.9385573984344864e-05,
      "loss": 0.6639,
      "step": 4415
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.917648434638977,
      "learning_rate": 1.9385222458609125e-05,
      "loss": 0.4945,
      "step": 4416
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7740106582641602,
      "learning_rate": 1.9384870835533617e-05,
      "loss": 0.4812,
      "step": 4417
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.867918848991394,
      "learning_rate": 1.9384519115121995e-05,
      "loss": 0.5674,
      "step": 4418
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8221943378448486,
      "learning_rate": 1.9384167297377902e-05,
      "loss": 0.5364,
      "step": 4419
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.907130479812622,
      "learning_rate": 1.9383815382304987e-05,
      "loss": 0.5466,
      "step": 4420
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8689466714859009,
      "learning_rate": 1.9383463369906898e-05,
      "loss": 0.5243,
      "step": 4421
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8761202096939087,
      "learning_rate": 1.9383111260187296e-05,
      "loss": 0.5482,
      "step": 4422
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.016998052597046,
      "learning_rate": 1.938275905314982e-05,
      "loss": 0.5894,
      "step": 4423
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8521323204040527,
      "learning_rate": 1.938240674879813e-05,
      "loss": 0.4478,
      "step": 4424
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9931639432907104,
      "learning_rate": 1.938205434713588e-05,
      "loss": 0.6216,
      "step": 4425
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8814417123794556,
      "learning_rate": 1.9381701848166725e-05,
      "loss": 0.473,
      "step": 4426
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9503602981567383,
      "learning_rate": 1.9381349251894318e-05,
      "loss": 0.5806,
      "step": 4427
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9551588296890259,
      "learning_rate": 1.9380996558322318e-05,
      "loss": 0.5103,
      "step": 4428
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7690095901489258,
      "learning_rate": 1.9380643767454385e-05,
      "loss": 0.5499,
      "step": 4429
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8769822120666504,
      "learning_rate": 1.938029087929417e-05,
      "loss": 0.4792,
      "step": 4430
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8316422700881958,
      "learning_rate": 1.9379937893845343e-05,
      "loss": 0.5449,
      "step": 4431
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9249267578125,
      "learning_rate": 1.937958481111156e-05,
      "loss": 0.5065,
      "step": 4432
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9028936624526978,
      "learning_rate": 1.9379231631096484e-05,
      "loss": 0.6302,
      "step": 4433
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.824889063835144,
      "learning_rate": 1.937887835380378e-05,
      "loss": 0.4567,
      "step": 4434
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8649362325668335,
      "learning_rate": 1.9378524979237104e-05,
      "loss": 0.5963,
      "step": 4435
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8486727476119995,
      "learning_rate": 1.9378171507400133e-05,
      "loss": 0.4835,
      "step": 4436
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.978868007659912,
      "learning_rate": 1.937781793829652e-05,
      "loss": 0.6828,
      "step": 4437
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8606724739074707,
      "learning_rate": 1.9377464271929947e-05,
      "loss": 0.4516,
      "step": 4438
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9344416856765747,
      "learning_rate": 1.937711050830407e-05,
      "loss": 0.5353,
      "step": 4439
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.84767746925354,
      "learning_rate": 1.9376756647422562e-05,
      "loss": 0.5198,
      "step": 4440
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9713845252990723,
      "learning_rate": 1.9376402689289095e-05,
      "loss": 0.6031,
      "step": 4441
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8756667375564575,
      "learning_rate": 1.937604863390734e-05,
      "loss": 0.4526,
      "step": 4442
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8266788721084595,
      "learning_rate": 1.9375694481280965e-05,
      "loss": 0.5708,
      "step": 4443
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8678463697433472,
      "learning_rate": 1.9375340231413643e-05,
      "loss": 0.4633,
      "step": 4444
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.058945894241333,
      "learning_rate": 1.9374985884309057e-05,
      "loss": 0.6986,
      "step": 4445
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.842505693435669,
      "learning_rate": 1.937463143997087e-05,
      "loss": 0.4604,
      "step": 4446
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8789151906967163,
      "learning_rate": 1.9374276898402768e-05,
      "loss": 0.5322,
      "step": 4447
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8725107908248901,
      "learning_rate": 1.9373922259608423e-05,
      "loss": 0.559,
      "step": 4448
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8350260257720947,
      "learning_rate": 1.9373567523591516e-05,
      "loss": 0.503,
      "step": 4449
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9802570343017578,
      "learning_rate": 1.9373212690355724e-05,
      "loss": 0.5854,
      "step": 4450
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8185820579528809,
      "learning_rate": 1.9372857759904727e-05,
      "loss": 0.5326,
      "step": 4451
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9088959693908691,
      "learning_rate": 1.937250273224221e-05,
      "loss": 0.4896,
      "step": 4452
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0278780460357666,
      "learning_rate": 1.937214760737185e-05,
      "loss": 0.6729,
      "step": 4453
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8482189178466797,
      "learning_rate": 1.937179238529733e-05,
      "loss": 0.483,
      "step": 4454
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8937350511550903,
      "learning_rate": 1.937143706602234e-05,
      "loss": 0.5771,
      "step": 4455
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7755130529403687,
      "learning_rate": 1.937108164955056e-05,
      "loss": 0.501,
      "step": 4456
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.882376790046692,
      "learning_rate": 1.937072613588568e-05,
      "loss": 0.5854,
      "step": 4457
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.852743148803711,
      "learning_rate": 1.9370370525031385e-05,
      "loss": 0.5227,
      "step": 4458
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8679656982421875,
      "learning_rate": 1.9370014816991363e-05,
      "loss": 0.524,
      "step": 4459
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8694430589675903,
      "learning_rate": 1.9369659011769302e-05,
      "loss": 0.5438,
      "step": 4460
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.493434190750122,
      "learning_rate": 1.93693031093689e-05,
      "loss": 0.8986,
      "step": 4461
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.111854314804077,
      "learning_rate": 1.9368947109793836e-05,
      "loss": 0.9785,
      "step": 4462
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1390461921691895,
      "learning_rate": 1.9368591013047812e-05,
      "loss": 0.566,
      "step": 4463
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0216071605682373,
      "learning_rate": 1.936823481913452e-05,
      "loss": 0.4953,
      "step": 4464
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9995959997177124,
      "learning_rate": 1.936787852805765e-05,
      "loss": 0.6277,
      "step": 4465
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8041902780532837,
      "learning_rate": 1.9367522139820898e-05,
      "loss": 0.4745,
      "step": 4466
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.174489736557007,
      "learning_rate": 1.9367165654427963e-05,
      "loss": 0.5659,
      "step": 4467
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9662396907806396,
      "learning_rate": 1.9366809071882543e-05,
      "loss": 0.48,
      "step": 4468
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9005885124206543,
      "learning_rate": 1.9366452392188334e-05,
      "loss": 0.6087,
      "step": 4469
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8882091045379639,
      "learning_rate": 1.9366095615349037e-05,
      "loss": 0.4915,
      "step": 4470
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.941193699836731,
      "learning_rate": 1.936573874136835e-05,
      "loss": 0.5727,
      "step": 4471
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.038630485534668,
      "learning_rate": 1.9365381770249977e-05,
      "loss": 0.4779,
      "step": 4472
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1195452213287354,
      "learning_rate": 1.9365024701997617e-05,
      "loss": 0.6224,
      "step": 4473
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.802616834640503,
      "learning_rate": 1.9364667536614977e-05,
      "loss": 0.4733,
      "step": 4474
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.830702543258667,
      "learning_rate": 1.9364310274105758e-05,
      "loss": 0.5383,
      "step": 4475
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7630046606063843,
      "learning_rate": 1.9363952914473668e-05,
      "loss": 0.4444,
      "step": 4476
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.019425630569458,
      "learning_rate": 1.9363595457722414e-05,
      "loss": 0.5436,
      "step": 4477
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9985448122024536,
      "learning_rate": 1.9363237903855698e-05,
      "loss": 0.5377,
      "step": 4478
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8391602039337158,
      "learning_rate": 1.9362880252877234e-05,
      "loss": 0.5667,
      "step": 4479
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7744532823562622,
      "learning_rate": 1.936252250479073e-05,
      "loss": 0.4825,
      "step": 4480
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8651542663574219,
      "learning_rate": 1.9362164659599898e-05,
      "loss": 0.5357,
      "step": 4481
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9729284048080444,
      "learning_rate": 1.9361806717308447e-05,
      "loss": 0.5365,
      "step": 4482
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8976367712020874,
      "learning_rate": 1.936144867792009e-05,
      "loss": 0.5891,
      "step": 4483
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.850023865699768,
      "learning_rate": 1.936109054143854e-05,
      "loss": 0.4694,
      "step": 4484
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8021482229232788,
      "learning_rate": 1.936073230786751e-05,
      "loss": 0.58,
      "step": 4485
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.98749577999115,
      "learning_rate": 1.936037397721072e-05,
      "loss": 0.536,
      "step": 4486
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8669942617416382,
      "learning_rate": 1.9360015549471884e-05,
      "loss": 0.5431,
      "step": 4487
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8254553079605103,
      "learning_rate": 1.935965702465472e-05,
      "loss": 0.5648,
      "step": 4488
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8168960809707642,
      "learning_rate": 1.9359298402762944e-05,
      "loss": 0.609,
      "step": 4489
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.9455881118774414,
      "learning_rate": 1.935893968380028e-05,
      "loss": 0.5667,
      "step": 4490
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.123680353164673,
      "learning_rate": 1.935858086777044e-05,
      "loss": 0.5396,
      "step": 4491
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.97905433177948,
      "learning_rate": 1.9358221954677158e-05,
      "loss": 0.4843,
      "step": 4492
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9278279542922974,
      "learning_rate": 1.9357862944524145e-05,
      "loss": 0.5523,
      "step": 4493
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9642677307128906,
      "learning_rate": 1.9357503837315132e-05,
      "loss": 0.5021,
      "step": 4494
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0679502487182617,
      "learning_rate": 1.935714463305384e-05,
      "loss": 0.5943,
      "step": 4495
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9174333810806274,
      "learning_rate": 1.9356785331743994e-05,
      "loss": 0.4694,
      "step": 4496
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9681141376495361,
      "learning_rate": 1.9356425933389323e-05,
      "loss": 0.5299,
      "step": 4497
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9079639911651611,
      "learning_rate": 1.9356066437993554e-05,
      "loss": 0.5107,
      "step": 4498
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.872630000114441,
      "learning_rate": 1.9355706845560412e-05,
      "loss": 0.522,
      "step": 4499
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9376813173294067,
      "learning_rate": 1.935534715609363e-05,
      "loss": 0.5322,
      "step": 4500
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0518345832824707,
      "learning_rate": 1.9354987369596943e-05,
      "loss": 0.6279,
      "step": 4501
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8884094953536987,
      "learning_rate": 1.9354627486074075e-05,
      "loss": 0.5438,
      "step": 4502
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1320226192474365,
      "learning_rate": 1.935426750552876e-05,
      "loss": 0.6663,
      "step": 4503
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8698397874832153,
      "learning_rate": 1.9353907427964732e-05,
      "loss": 0.4708,
      "step": 4504
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.758699655532837,
      "learning_rate": 1.9353547253385727e-05,
      "loss": 0.4784,
      "step": 4505
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9407552480697632,
      "learning_rate": 1.935318698179548e-05,
      "loss": 0.526,
      "step": 4506
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.120392084121704,
      "learning_rate": 1.9352826613197728e-05,
      "loss": 0.611,
      "step": 4507
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8363959789276123,
      "learning_rate": 1.9352466147596208e-05,
      "loss": 0.4935,
      "step": 4508
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9079670906066895,
      "learning_rate": 1.935210558499466e-05,
      "loss": 0.5727,
      "step": 4509
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0254709720611572,
      "learning_rate": 1.935174492539682e-05,
      "loss": 0.626,
      "step": 4510
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.80394446849823,
      "learning_rate": 1.935138416880643e-05,
      "loss": 0.492,
      "step": 4511
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9667572975158691,
      "learning_rate": 1.9351023315227236e-05,
      "loss": 0.4592,
      "step": 4512
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8072941303253174,
      "learning_rate": 1.9350662364662975e-05,
      "loss": 0.5113,
      "step": 4513
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.851575493812561,
      "learning_rate": 1.9350301317117395e-05,
      "loss": 0.4837,
      "step": 4514
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9220350980758667,
      "learning_rate": 1.9349940172594234e-05,
      "loss": 0.5494,
      "step": 4515
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8265832662582397,
      "learning_rate": 1.934957893109725e-05,
      "loss": 0.4651,
      "step": 4516
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9136078357696533,
      "learning_rate": 1.9349217592630173e-05,
      "loss": 0.5327,
      "step": 4517
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0233116149902344,
      "learning_rate": 1.9348856157196762e-05,
      "loss": 0.5921,
      "step": 4518
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8991178274154663,
      "learning_rate": 1.9348494624800765e-05,
      "loss": 0.5823,
      "step": 4519
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9403327703475952,
      "learning_rate": 1.934813299544593e-05,
      "loss": 0.5645,
      "step": 4520
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9290900230407715,
      "learning_rate": 1.9347771269136007e-05,
      "loss": 0.5797,
      "step": 4521
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.877866268157959,
      "learning_rate": 1.9347409445874747e-05,
      "loss": 0.5007,
      "step": 4522
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.731155276298523,
      "learning_rate": 1.9347047525665904e-05,
      "loss": 0.5191,
      "step": 4523
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9239394664764404,
      "learning_rate": 1.9346685508513233e-05,
      "loss": 0.5759,
      "step": 4524
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.919970154762268,
      "learning_rate": 1.9346323394420486e-05,
      "loss": 0.577,
      "step": 4525
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9508686065673828,
      "learning_rate": 1.9345961183391416e-05,
      "loss": 0.5594,
      "step": 4526
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9247490167617798,
      "learning_rate": 1.934559887542979e-05,
      "loss": 0.541,
      "step": 4527
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8090535402297974,
      "learning_rate": 1.9345236470539356e-05,
      "loss": 0.525,
      "step": 4528
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.898991346359253,
      "learning_rate": 1.9344873968723878e-05,
      "loss": 0.5424,
      "step": 4529
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.813488245010376,
      "learning_rate": 1.9344511369987113e-05,
      "loss": 0.4446,
      "step": 4530
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0932323932647705,
      "learning_rate": 1.934414867433282e-05,
      "loss": 0.6574,
      "step": 4531
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8721518516540527,
      "learning_rate": 1.9343785881764766e-05,
      "loss": 0.5191,
      "step": 4532
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8933804035186768,
      "learning_rate": 1.934342299228671e-05,
      "loss": 0.6095,
      "step": 4533
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.78635835647583,
      "learning_rate": 1.9343060005902415e-05,
      "loss": 0.507,
      "step": 4534
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8563112020492554,
      "learning_rate": 1.934269692261565e-05,
      "loss": 0.5857,
      "step": 4535
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9937512874603271,
      "learning_rate": 1.934233374243018e-05,
      "loss": 0.4673,
      "step": 4536
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.907296895980835,
      "learning_rate": 1.9341970465349764e-05,
      "loss": 0.5067,
      "step": 4537
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8888717889785767,
      "learning_rate": 1.934160709137818e-05,
      "loss": 0.5548,
      "step": 4538
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8126420974731445,
      "learning_rate": 1.934124362051919e-05,
      "loss": 0.5393,
      "step": 4539
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.85120689868927,
      "learning_rate": 1.9340880052776568e-05,
      "loss": 0.4878,
      "step": 4540
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8768292665481567,
      "learning_rate": 1.9340516388154084e-05,
      "loss": 0.506,
      "step": 4541
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1431021690368652,
      "learning_rate": 1.9340152626655507e-05,
      "loss": 0.4639,
      "step": 4542
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9034181833267212,
      "learning_rate": 1.933978876828461e-05,
      "loss": 0.5946,
      "step": 4543
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.030870199203491,
      "learning_rate": 1.933942481304517e-05,
      "loss": 0.5132,
      "step": 4544
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0766971111297607,
      "learning_rate": 1.9339060760940962e-05,
      "loss": 0.554,
      "step": 4545
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8503271341323853,
      "learning_rate": 1.933869661197576e-05,
      "loss": 0.5038,
      "step": 4546
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8874647617340088,
      "learning_rate": 1.9338332366153337e-05,
      "loss": 0.5798,
      "step": 4547
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0607028007507324,
      "learning_rate": 1.933796802347748e-05,
      "loss": 0.5917,
      "step": 4548
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1686925888061523,
      "learning_rate": 1.933760358395196e-05,
      "loss": 0.595,
      "step": 4549
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.875584602355957,
      "learning_rate": 1.9337239047580556e-05,
      "loss": 0.4897,
      "step": 4550
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2721667289733887,
      "learning_rate": 1.9336874414367054e-05,
      "loss": 0.6425,
      "step": 4551
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8063163757324219,
      "learning_rate": 1.9336509684315236e-05,
      "loss": 0.4894,
      "step": 4552
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.038710594177246,
      "learning_rate": 1.9336144857428884e-05,
      "loss": 0.6239,
      "step": 4553
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8307075500488281,
      "learning_rate": 1.9335779933711775e-05,
      "loss": 0.4395,
      "step": 4554
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3655357360839844,
      "learning_rate": 1.9335414913167704e-05,
      "loss": 0.9145,
      "step": 4555
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2162604331970215,
      "learning_rate": 1.933504979580045e-05,
      "loss": 0.9233,
      "step": 4556
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0110013484954834,
      "learning_rate": 1.9334684581613804e-05,
      "loss": 0.6027,
      "step": 4557
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0617146492004395,
      "learning_rate": 1.9334319270611547e-05,
      "loss": 0.5095,
      "step": 4558
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9531652927398682,
      "learning_rate": 1.933395386279748e-05,
      "loss": 0.6063,
      "step": 4559
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8201476335525513,
      "learning_rate": 1.933358835817538e-05,
      "loss": 0.5132,
      "step": 4560
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.017296075820923,
      "learning_rate": 1.9333222756749047e-05,
      "loss": 0.6808,
      "step": 4561
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.876480221748352,
      "learning_rate": 1.9332857058522267e-05,
      "loss": 0.4555,
      "step": 4562
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.181636333465576,
      "learning_rate": 1.9332491263498838e-05,
      "loss": 0.8373,
      "step": 4563
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.915508508682251,
      "learning_rate": 1.933212537168255e-05,
      "loss": 0.863,
      "step": 4564
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.4149792194366455,
      "learning_rate": 1.93317593830772e-05,
      "loss": 0.5953,
      "step": 4565
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1579344272613525,
      "learning_rate": 1.9331393297686582e-05,
      "loss": 0.4985,
      "step": 4566
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0417351722717285,
      "learning_rate": 1.9331027115514493e-05,
      "loss": 0.5634,
      "step": 4567
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1979355812072754,
      "learning_rate": 1.9330660836564733e-05,
      "loss": 0.578,
      "step": 4568
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9629384279251099,
      "learning_rate": 1.9330294460841096e-05,
      "loss": 0.5778,
      "step": 4569
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0188937187194824,
      "learning_rate": 1.932992798834739e-05,
      "loss": 0.4966,
      "step": 4570
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.065433979034424,
      "learning_rate": 1.932956141908741e-05,
      "loss": 0.5782,
      "step": 4571
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0083489418029785,
      "learning_rate": 1.932919475306496e-05,
      "loss": 0.4692,
      "step": 4572
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.034608840942383,
      "learning_rate": 1.932882799028384e-05,
      "loss": 0.6141,
      "step": 4573
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8058863878250122,
      "learning_rate": 1.932846113074786e-05,
      "loss": 0.4841,
      "step": 4574
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.816680669784546,
      "learning_rate": 1.9328094174460817e-05,
      "loss": 0.5489,
      "step": 4575
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.030606269836426,
      "learning_rate": 1.9327727121426525e-05,
      "loss": 0.5461,
      "step": 4576
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9421281814575195,
      "learning_rate": 1.9327359971648785e-05,
      "loss": 0.5994,
      "step": 4577
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.859100580215454,
      "learning_rate": 1.9326992725131406e-05,
      "loss": 0.4549,
      "step": 4578
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0133392810821533,
      "learning_rate": 1.9326625381878198e-05,
      "loss": 0.6135,
      "step": 4579
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8317134380340576,
      "learning_rate": 1.9326257941892974e-05,
      "loss": 0.4695,
      "step": 4580
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9776009321212769,
      "learning_rate": 1.932589040517954e-05,
      "loss": 0.6198,
      "step": 4581
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.708101511001587,
      "learning_rate": 1.9325522771741707e-05,
      "loss": 0.4725,
      "step": 4582
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8430116176605225,
      "learning_rate": 1.9325155041583293e-05,
      "loss": 0.5535,
      "step": 4583
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8844610452651978,
      "learning_rate": 1.932478721470811e-05,
      "loss": 0.492,
      "step": 4584
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.172644853591919,
      "learning_rate": 1.932441929111997e-05,
      "loss": 0.6067,
      "step": 4585
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8321110010147095,
      "learning_rate": 1.932405127082269e-05,
      "loss": 0.4732,
      "step": 4586
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8891031742095947,
      "learning_rate": 1.9323683153820095e-05,
      "loss": 0.5791,
      "step": 4587
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8595985174179077,
      "learning_rate": 1.932331494011599e-05,
      "loss": 0.4978,
      "step": 4588
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.904899001121521,
      "learning_rate": 1.9322946629714204e-05,
      "loss": 0.5572,
      "step": 4589
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9007240533828735,
      "learning_rate": 1.932257822261855e-05,
      "loss": 0.5416,
      "step": 4590
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9343807697296143,
      "learning_rate": 1.9322209718832855e-05,
      "loss": 0.5922,
      "step": 4591
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8081586360931396,
      "learning_rate": 1.932184111836094e-05,
      "loss": 0.4726,
      "step": 4592
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9196666479110718,
      "learning_rate": 1.9321472421206623e-05,
      "loss": 0.5934,
      "step": 4593
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7002348899841309,
      "learning_rate": 1.9321103627373735e-05,
      "loss": 0.4749,
      "step": 4594
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7748934030532837,
      "learning_rate": 1.9320734736866093e-05,
      "loss": 0.5465,
      "step": 4595
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7220194339752197,
      "learning_rate": 1.932036574968753e-05,
      "loss": 0.4628,
      "step": 4596
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8698031902313232,
      "learning_rate": 1.9319996665841867e-05,
      "loss": 0.553,
      "step": 4597
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9885450601577759,
      "learning_rate": 1.9319627485332936e-05,
      "loss": 0.4729,
      "step": 4598
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7829700708389282,
      "learning_rate": 1.931925820816457e-05,
      "loss": 0.5244,
      "step": 4599
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8220995664596558,
      "learning_rate": 1.931888883434059e-05,
      "loss": 0.5716,
      "step": 4600
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8668774366378784,
      "learning_rate": 1.9318519363864826e-05,
      "loss": 0.5319,
      "step": 4601
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8851258754730225,
      "learning_rate": 1.9318149796741123e-05,
      "loss": 0.5547,
      "step": 4602
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0450353622436523,
      "learning_rate": 1.9317780132973302e-05,
      "loss": 0.6349,
      "step": 4603
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8583786487579346,
      "learning_rate": 1.9317410372565205e-05,
      "loss": 0.4898,
      "step": 4604
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.87354576587677,
      "learning_rate": 1.931704051552066e-05,
      "loss": 0.5621,
      "step": 4605
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.927802562713623,
      "learning_rate": 1.9316670561843505e-05,
      "loss": 0.5774,
      "step": 4606
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8435206413269043,
      "learning_rate": 1.9316300511537583e-05,
      "loss": 0.572,
      "step": 4607
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8294543027877808,
      "learning_rate": 1.9315930364606723e-05,
      "loss": 0.5329,
      "step": 4608
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0299386978149414,
      "learning_rate": 1.931556012105477e-05,
      "loss": 0.6937,
      "step": 4609
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7807265520095825,
      "learning_rate": 1.931518978088556e-05,
      "loss": 0.4716,
      "step": 4610
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8263986110687256,
      "learning_rate": 1.931481934410294e-05,
      "loss": 0.5288,
      "step": 4611
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9502034187316895,
      "learning_rate": 1.9314448810710745e-05,
      "loss": 0.4875,
      "step": 4612
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1268701553344727,
      "learning_rate": 1.9314078180712824e-05,
      "loss": 0.5565,
      "step": 4613
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.760243535041809,
      "learning_rate": 1.9313707454113015e-05,
      "loss": 0.4616,
      "step": 4614
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9649662971496582,
      "learning_rate": 1.931333663091517e-05,
      "loss": 0.5941,
      "step": 4615
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8496503829956055,
      "learning_rate": 1.931296571112313e-05,
      "loss": 0.4858,
      "step": 4616
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8561451435089111,
      "learning_rate": 1.9312594694740742e-05,
      "loss": 0.5089,
      "step": 4617
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8752975463867188,
      "learning_rate": 1.9312223581771855e-05,
      "loss": 0.5145,
      "step": 4618
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0069401264190674,
      "learning_rate": 1.931185237222032e-05,
      "loss": 0.5961,
      "step": 4619
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8403754234313965,
      "learning_rate": 1.9311481066089984e-05,
      "loss": 0.4611,
      "step": 4620
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8938851356506348,
      "learning_rate": 1.93111096633847e-05,
      "loss": 0.5584,
      "step": 4621
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.902166485786438,
      "learning_rate": 1.9310738164108322e-05,
      "loss": 0.5198,
      "step": 4622
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8801405429840088,
      "learning_rate": 1.93103665682647e-05,
      "loss": 0.5443,
      "step": 4623
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8625580072402954,
      "learning_rate": 1.930999487585769e-05,
      "loss": 0.4785,
      "step": 4624
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8672984838485718,
      "learning_rate": 1.9309623086891142e-05,
      "loss": 0.547,
      "step": 4625
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9052278995513916,
      "learning_rate": 1.9309251201368914e-05,
      "loss": 0.5015,
      "step": 4626
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8693492412567139,
      "learning_rate": 1.9308879219294868e-05,
      "loss": 0.5626,
      "step": 4627
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8480459451675415,
      "learning_rate": 1.930850714067286e-05,
      "loss": 0.4614,
      "step": 4628
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8928990364074707,
      "learning_rate": 1.9308134965506746e-05,
      "loss": 0.5267,
      "step": 4629
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8623515367507935,
      "learning_rate": 1.930776269380039e-05,
      "loss": 0.4859,
      "step": 4630
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8526859283447266,
      "learning_rate": 1.930739032555765e-05,
      "loss": 0.5896,
      "step": 4631
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8917596340179443,
      "learning_rate": 1.9307017860782386e-05,
      "loss": 0.5211,
      "step": 4632
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.023104429244995,
      "learning_rate": 1.930664529947847e-05,
      "loss": 0.6601,
      "step": 4633
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7615798711776733,
      "learning_rate": 1.9306272641649753e-05,
      "loss": 0.4502,
      "step": 4634
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8237828016281128,
      "learning_rate": 1.9305899887300112e-05,
      "loss": 0.5347,
      "step": 4635
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.794581651687622,
      "learning_rate": 1.930552703643341e-05,
      "loss": 0.461,
      "step": 4636
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9699233770370483,
      "learning_rate": 1.9305154089053507e-05,
      "loss": 0.5535,
      "step": 4637
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9184871912002563,
      "learning_rate": 1.930478104516428e-05,
      "loss": 0.5572,
      "step": 4638
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.078559637069702,
      "learning_rate": 1.9304407904769593e-05,
      "loss": 0.6962,
      "step": 4639
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.971011996269226,
      "learning_rate": 1.9304034667873315e-05,
      "loss": 0.521,
      "step": 4640
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8564032316207886,
      "learning_rate": 1.9303661334479328e-05,
      "loss": 0.5659,
      "step": 4641
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8616373538970947,
      "learning_rate": 1.930328790459149e-05,
      "loss": 0.4894,
      "step": 4642
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.905205249786377,
      "learning_rate": 1.9302914378213678e-05,
      "loss": 0.5792,
      "step": 4643
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.806523084640503,
      "learning_rate": 1.930254075534977e-05,
      "loss": 0.4694,
      "step": 4644
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9563984870910645,
      "learning_rate": 1.930216703600364e-05,
      "loss": 0.6176,
      "step": 4645
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8110198974609375,
      "learning_rate": 1.930179322017916e-05,
      "loss": 0.4767,
      "step": 4646
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9271583557128906,
      "learning_rate": 1.9301419307880213e-05,
      "loss": 0.6521,
      "step": 4647
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9405591487884521,
      "learning_rate": 1.930104529911067e-05,
      "loss": 0.5468,
      "step": 4648
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8783973455429077,
      "learning_rate": 1.9300671193874417e-05,
      "loss": 0.5895,
      "step": 4649
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9629460573196411,
      "learning_rate": 1.930029699217533e-05,
      "loss": 0.5111,
      "step": 4650
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8992891311645508,
      "learning_rate": 1.9299922694017294e-05,
      "loss": 0.5789,
      "step": 4651
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7610410451889038,
      "learning_rate": 1.9299548299404187e-05,
      "loss": 0.4751,
      "step": 4652
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8723567724227905,
      "learning_rate": 1.9299173808339892e-05,
      "loss": 0.5865,
      "step": 4653
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9000850915908813,
      "learning_rate": 1.9298799220828296e-05,
      "loss": 0.4727,
      "step": 4654
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.4373528957366943,
      "learning_rate": 1.929842453687328e-05,
      "loss": 0.884,
      "step": 4655
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.168877363204956,
      "learning_rate": 1.9298049756478736e-05,
      "loss": 0.9908,
      "step": 4656
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.92729651927948,
      "learning_rate": 1.9297674879648546e-05,
      "loss": 0.618,
      "step": 4657
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.656289577484131,
      "learning_rate": 1.9297299906386602e-05,
      "loss": 0.4615,
      "step": 4658
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8937686681747437,
      "learning_rate": 1.9296924836696788e-05,
      "loss": 0.5535,
      "step": 4659
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.824832558631897,
      "learning_rate": 1.9296549670583e-05,
      "loss": 0.5062,
      "step": 4660
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8019837141036987,
      "learning_rate": 1.9296174408049122e-05,
      "loss": 0.5977,
      "step": 4661
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7839800119400024,
      "learning_rate": 1.9295799049099054e-05,
      "loss": 0.5049,
      "step": 4662
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9789966344833374,
      "learning_rate": 1.929542359373668e-05,
      "loss": 0.5915,
      "step": 4663
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.816117763519287,
      "learning_rate": 1.9295048041965904e-05,
      "loss": 0.5001,
      "step": 4664
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.920786738395691,
      "learning_rate": 1.9294672393790615e-05,
      "loss": 0.5626,
      "step": 4665
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8322887420654297,
      "learning_rate": 1.9294296649214706e-05,
      "loss": 0.5719,
      "step": 4666
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9013913869857788,
      "learning_rate": 1.9293920808242084e-05,
      "loss": 0.6009,
      "step": 4667
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8244249820709229,
      "learning_rate": 1.9293544870876638e-05,
      "loss": 0.4889,
      "step": 4668
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.92914879322052,
      "learning_rate": 1.9293168837122275e-05,
      "loss": 0.5714,
      "step": 4669
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8810241222381592,
      "learning_rate": 1.9292792706982884e-05,
      "loss": 0.4609,
      "step": 4670
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8035424947738647,
      "learning_rate": 1.929241648046238e-05,
      "loss": 0.559,
      "step": 4671
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9965609312057495,
      "learning_rate": 1.9292040157564652e-05,
      "loss": 0.564,
      "step": 4672
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.990127682685852,
      "learning_rate": 1.9291663738293614e-05,
      "loss": 0.5965,
      "step": 4673
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.827016830444336,
      "learning_rate": 1.929128722265316e-05,
      "loss": 0.4904,
      "step": 4674
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.89156174659729,
      "learning_rate": 1.9290910610647203e-05,
      "loss": 0.5575,
      "step": 4675
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6603538990020752,
      "learning_rate": 1.9290533902279646e-05,
      "loss": 0.4534,
      "step": 4676
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9592057466506958,
      "learning_rate": 1.9290157097554393e-05,
      "loss": 0.5967,
      "step": 4677
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7933621406555176,
      "learning_rate": 1.928978019647536e-05,
      "loss": 0.4428,
      "step": 4678
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8013548851013184,
      "learning_rate": 1.9289403199046444e-05,
      "loss": 0.5851,
      "step": 4679
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7542682886123657,
      "learning_rate": 1.9289026105271568e-05,
      "loss": 0.4659,
      "step": 4680
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9225705862045288,
      "learning_rate": 1.928864891515464e-05,
      "loss": 0.5597,
      "step": 4681
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9403799772262573,
      "learning_rate": 1.9288271628699563e-05,
      "loss": 0.541,
      "step": 4682
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9979370832443237,
      "learning_rate": 1.9287894245910262e-05,
      "loss": 0.5351,
      "step": 4683
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.945855975151062,
      "learning_rate": 1.928751676679064e-05,
      "loss": 0.5683,
      "step": 4684
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.054295539855957,
      "learning_rate": 1.928713919134462e-05,
      "loss": 0.6391,
      "step": 4685
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8184486627578735,
      "learning_rate": 1.928676151957612e-05,
      "loss": 0.488,
      "step": 4686
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9118095636367798,
      "learning_rate": 1.928638375148905e-05,
      "loss": 0.5901,
      "step": 4687
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9054895639419556,
      "learning_rate": 1.9286005887087328e-05,
      "loss": 0.5178,
      "step": 4688
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.985416054725647,
      "learning_rate": 1.9285627926374877e-05,
      "loss": 0.562,
      "step": 4689
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.844887614250183,
      "learning_rate": 1.9285249869355617e-05,
      "loss": 0.4766,
      "step": 4690
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.925352931022644,
      "learning_rate": 1.9284871716033467e-05,
      "loss": 0.5516,
      "step": 4691
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8414980173110962,
      "learning_rate": 1.9284493466412352e-05,
      "loss": 0.5238,
      "step": 4692
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8379744291305542,
      "learning_rate": 1.9284115120496188e-05,
      "loss": 0.5718,
      "step": 4693
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9194450378417969,
      "learning_rate": 1.9283736678288912e-05,
      "loss": 0.5275,
      "step": 4694
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8121908903121948,
      "learning_rate": 1.9283358139794434e-05,
      "loss": 0.5625,
      "step": 4695
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.836934208869934,
      "learning_rate": 1.9282979505016694e-05,
      "loss": 0.4956,
      "step": 4696
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7840031385421753,
      "learning_rate": 1.9282600773959605e-05,
      "loss": 0.6114,
      "step": 4697
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0583837032318115,
      "learning_rate": 1.9282221946627106e-05,
      "loss": 0.4633,
      "step": 4698
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.985701084136963,
      "learning_rate": 1.9281843023023125e-05,
      "loss": 0.6201,
      "step": 4699
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.926385521888733,
      "learning_rate": 1.9281464003151586e-05,
      "loss": 0.5082,
      "step": 4700
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9038339853286743,
      "learning_rate": 1.9281084887016423e-05,
      "loss": 0.6067,
      "step": 4701
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7210694551467896,
      "learning_rate": 1.928070567462157e-05,
      "loss": 0.4747,
      "step": 4702
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8737499713897705,
      "learning_rate": 1.928032636597096e-05,
      "loss": 0.5377,
      "step": 4703
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8373305797576904,
      "learning_rate": 1.9279946961068523e-05,
      "loss": 0.519,
      "step": 4704
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.918034553527832,
      "learning_rate": 1.9279567459918196e-05,
      "loss": 0.5712,
      "step": 4705
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7718451023101807,
      "learning_rate": 1.9279187862523917e-05,
      "loss": 0.5003,
      "step": 4706
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8473355770111084,
      "learning_rate": 1.9278808168889623e-05,
      "loss": 0.503,
      "step": 4707
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9073752164840698,
      "learning_rate": 1.927842837901925e-05,
      "loss": 0.4831,
      "step": 4708
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8217271566390991,
      "learning_rate": 1.9278048492916738e-05,
      "loss": 0.5479,
      "step": 4709
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7700773477554321,
      "learning_rate": 1.9277668510586025e-05,
      "loss": 0.4493,
      "step": 4710
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9460097551345825,
      "learning_rate": 1.9277288432031055e-05,
      "loss": 0.6269,
      "step": 4711
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.865323543548584,
      "learning_rate": 1.9276908257255768e-05,
      "loss": 0.4543,
      "step": 4712
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9330816268920898,
      "learning_rate": 1.9276527986264108e-05,
      "loss": 0.6004,
      "step": 4713
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.797551155090332,
      "learning_rate": 1.927614761906002e-05,
      "loss": 0.4742,
      "step": 4714
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8654136657714844,
      "learning_rate": 1.9275767155647446e-05,
      "loss": 0.5583,
      "step": 4715
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.861737608909607,
      "learning_rate": 1.9275386596030336e-05,
      "loss": 0.5051,
      "step": 4716
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9301601648330688,
      "learning_rate": 1.9275005940212638e-05,
      "loss": 0.5877,
      "step": 4717
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7541017532348633,
      "learning_rate": 1.927462518819829e-05,
      "loss": 0.4552,
      "step": 4718
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9373066425323486,
      "learning_rate": 1.927424433999125e-05,
      "loss": 0.5456,
      "step": 4719
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.804097056388855,
      "learning_rate": 1.9273863395595466e-05,
      "loss": 0.4555,
      "step": 4720
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0146329402923584,
      "learning_rate": 1.927348235501489e-05,
      "loss": 0.5607,
      "step": 4721
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.949149489402771,
      "learning_rate": 1.9273101218253475e-05,
      "loss": 0.463,
      "step": 4722
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9444818496704102,
      "learning_rate": 1.927271998531517e-05,
      "loss": 0.5178,
      "step": 4723
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9074634313583374,
      "learning_rate": 1.927233865620393e-05,
      "loss": 0.5455,
      "step": 4724
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.858330249786377,
      "learning_rate": 1.9271957230923713e-05,
      "loss": 0.5204,
      "step": 4725
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7368611097335815,
      "learning_rate": 1.9271575709478468e-05,
      "loss": 0.4734,
      "step": 4726
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.850813865661621,
      "learning_rate": 1.927119409187216e-05,
      "loss": 0.5617,
      "step": 4727
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9736946821212769,
      "learning_rate": 1.9270812378108746e-05,
      "loss": 0.5448,
      "step": 4728
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9477205276489258,
      "learning_rate": 1.927043056819218e-05,
      "loss": 0.531,
      "step": 4729
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.899040699005127,
      "learning_rate": 1.927004866212643e-05,
      "loss": 0.5525,
      "step": 4730
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0324816703796387,
      "learning_rate": 1.9269666659915444e-05,
      "loss": 0.6242,
      "step": 4731
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8290693759918213,
      "learning_rate": 1.9269284561563198e-05,
      "loss": 0.4544,
      "step": 4732
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9040850400924683,
      "learning_rate": 1.9268902367073644e-05,
      "loss": 0.549,
      "step": 4733
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8047548532485962,
      "learning_rate": 1.9268520076450753e-05,
      "loss": 0.4798,
      "step": 4734
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8552417755126953,
      "learning_rate": 1.926813768969849e-05,
      "loss": 0.5509,
      "step": 4735
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8657938241958618,
      "learning_rate": 1.9267755206820816e-05,
      "loss": 0.5203,
      "step": 4736
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8225785493850708,
      "learning_rate": 1.9267372627821703e-05,
      "loss": 0.4822,
      "step": 4737
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8353338241577148,
      "learning_rate": 1.9266989952705112e-05,
      "loss": 0.4496,
      "step": 4738
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9435405731201172,
      "learning_rate": 1.9266607181475015e-05,
      "loss": 0.5431,
      "step": 4739
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.806702971458435,
      "learning_rate": 1.926622431413539e-05,
      "loss": 0.5051,
      "step": 4740
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.833217740058899,
      "learning_rate": 1.92658413506902e-05,
      "loss": 0.5719,
      "step": 4741
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7827088832855225,
      "learning_rate": 1.9265458291143417e-05,
      "loss": 0.5012,
      "step": 4742
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9555323123931885,
      "learning_rate": 1.9265075135499015e-05,
      "loss": 0.6055,
      "step": 4743
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8420523405075073,
      "learning_rate": 1.926469188376097e-05,
      "loss": 0.4936,
      "step": 4744
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.023651361465454,
      "learning_rate": 1.9264308535933253e-05,
      "loss": 0.5623,
      "step": 4745
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8635152578353882,
      "learning_rate": 1.9263925092019845e-05,
      "loss": 0.4951,
      "step": 4746
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9987989664077759,
      "learning_rate": 1.9263541552024716e-05,
      "loss": 0.5848,
      "step": 4747
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7942012548446655,
      "learning_rate": 1.9263157915951852e-05,
      "loss": 0.4848,
      "step": 4748
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9116120338439941,
      "learning_rate": 1.9262774183805225e-05,
      "loss": 0.5476,
      "step": 4749
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7617231607437134,
      "learning_rate": 1.926239035558882e-05,
      "loss": 0.4906,
      "step": 4750
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8978636264801025,
      "learning_rate": 1.926200643130661e-05,
      "loss": 0.6033,
      "step": 4751
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8587123155593872,
      "learning_rate": 1.9261622410962587e-05,
      "loss": 0.5636,
      "step": 4752
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8807734251022339,
      "learning_rate": 1.9261238294560734e-05,
      "loss": 0.5955,
      "step": 4753
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7448034286499023,
      "learning_rate": 1.9260854082105026e-05,
      "loss": 0.4941,
      "step": 4754
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9670714139938354,
      "learning_rate": 1.926046977359945e-05,
      "loss": 0.5908,
      "step": 4755
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8110285997390747,
      "learning_rate": 1.9260085369047997e-05,
      "loss": 0.5123,
      "step": 4756
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9454014301300049,
      "learning_rate": 1.9259700868454652e-05,
      "loss": 0.5504,
      "step": 4757
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9888840913772583,
      "learning_rate": 1.9259316271823397e-05,
      "loss": 0.5915,
      "step": 4758
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.735120415687561,
      "learning_rate": 1.925893157915823e-05,
      "loss": 0.4836,
      "step": 4759
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8984267711639404,
      "learning_rate": 1.9258546790463138e-05,
      "loss": 0.5675,
      "step": 4760
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8963137865066528,
      "learning_rate": 1.925816190574211e-05,
      "loss": 0.612,
      "step": 4761
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.775295376777649,
      "learning_rate": 1.9257776924999137e-05,
      "loss": 0.4706,
      "step": 4762
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8886595964431763,
      "learning_rate": 1.9257391848238212e-05,
      "loss": 0.5655,
      "step": 4763
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7668002843856812,
      "learning_rate": 1.9257006675463332e-05,
      "loss": 0.4735,
      "step": 4764
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8490411043167114,
      "learning_rate": 1.925662140667849e-05,
      "loss": 0.5668,
      "step": 4765
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7723432779312134,
      "learning_rate": 1.9256236041887682e-05,
      "loss": 0.4865,
      "step": 4766
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8772257566452026,
      "learning_rate": 1.9255850581094905e-05,
      "loss": 0.5676,
      "step": 4767
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8165205717086792,
      "learning_rate": 1.9255465024304157e-05,
      "loss": 0.4908,
      "step": 4768
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.90349543094635,
      "learning_rate": 1.9255079371519432e-05,
      "loss": 0.5974,
      "step": 4769
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7948150634765625,
      "learning_rate": 1.925469362274474e-05,
      "loss": 0.4578,
      "step": 4770
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8630644083023071,
      "learning_rate": 1.9254307777984077e-05,
      "loss": 0.5616,
      "step": 4771
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8029156923294067,
      "learning_rate": 1.925392183724144e-05,
      "loss": 0.4749,
      "step": 4772
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8436511754989624,
      "learning_rate": 1.9253535800520837e-05,
      "loss": 0.5229,
      "step": 4773
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7752323150634766,
      "learning_rate": 1.925314966782627e-05,
      "loss": 0.4736,
      "step": 4774
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.000542640686035,
      "learning_rate": 1.9252763439161743e-05,
      "loss": 0.6238,
      "step": 4775
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9284703731536865,
      "learning_rate": 1.9252377114531268e-05,
      "loss": 0.4708,
      "step": 4776
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7803035974502563,
      "learning_rate": 1.925199069393884e-05,
      "loss": 0.5038,
      "step": 4777
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.048765182495117,
      "learning_rate": 1.925160417738848e-05,
      "loss": 0.5939,
      "step": 4778
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8956586122512817,
      "learning_rate": 1.9251217564884188e-05,
      "loss": 0.5788,
      "step": 4779
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.880500078201294,
      "learning_rate": 1.9250830856429976e-05,
      "loss": 0.4852,
      "step": 4780
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9528628587722778,
      "learning_rate": 1.9250444052029858e-05,
      "loss": 0.5114,
      "step": 4781
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8452433347702026,
      "learning_rate": 1.925005715168784e-05,
      "loss": 0.5349,
      "step": 4782
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8709219694137573,
      "learning_rate": 1.924967015540794e-05,
      "loss": 0.5664,
      "step": 4783
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.918903112411499,
      "learning_rate": 1.924928306319417e-05,
      "loss": 0.5469,
      "step": 4784
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.929957628250122,
      "learning_rate": 1.9248895875050538e-05,
      "loss": 0.5313,
      "step": 4785
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8943134546279907,
      "learning_rate": 1.9248508590981073e-05,
      "loss": 0.4601,
      "step": 4786
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8972837924957275,
      "learning_rate": 1.9248121210989782e-05,
      "loss": 0.5222,
      "step": 4787
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8411813974380493,
      "learning_rate": 1.9247733735080686e-05,
      "loss": 0.4974,
      "step": 4788
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8764194250106812,
      "learning_rate": 1.92473461632578e-05,
      "loss": 0.5637,
      "step": 4789
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8253036737442017,
      "learning_rate": 1.9246958495525156e-05,
      "loss": 0.5291,
      "step": 4790
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9182254076004028,
      "learning_rate": 1.924657073188676e-05,
      "loss": 0.5849,
      "step": 4791
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8069084882736206,
      "learning_rate": 1.9246182872346638e-05,
      "loss": 0.4563,
      "step": 4792
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8990373611450195,
      "learning_rate": 1.9245794916908817e-05,
      "loss": 0.5168,
      "step": 4793
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8152371644973755,
      "learning_rate": 1.9245406865577318e-05,
      "loss": 0.507,
      "step": 4794
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.5071609020233154,
      "learning_rate": 1.924501871835616e-05,
      "loss": 0.5791,
      "step": 4795
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8117873668670654,
      "learning_rate": 1.924463047524938e-05,
      "loss": 0.5381,
      "step": 4796
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8348780870437622,
      "learning_rate": 1.9244242136260997e-05,
      "loss": 0.5272,
      "step": 4797
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8423866033554077,
      "learning_rate": 1.9243853701395043e-05,
      "loss": 0.5037,
      "step": 4798
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8420767784118652,
      "learning_rate": 1.9243465170655544e-05,
      "loss": 0.5392,
      "step": 4799
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8404828310012817,
      "learning_rate": 1.9243076544046528e-05,
      "loss": 0.5262,
      "step": 4800
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8180510997772217,
      "learning_rate": 1.9242687821572027e-05,
      "loss": 0.5514,
      "step": 4801
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9773069620132446,
      "learning_rate": 1.9242299003236077e-05,
      "loss": 0.4639,
      "step": 4802
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.892061471939087,
      "learning_rate": 1.924191008904271e-05,
      "loss": 0.5474,
      "step": 4803
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.824189305305481,
      "learning_rate": 1.924152107899595e-05,
      "loss": 0.5104,
      "step": 4804
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0237700939178467,
      "learning_rate": 1.924113197309984e-05,
      "loss": 0.6178,
      "step": 4805
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.908378005027771,
      "learning_rate": 1.9240742771358416e-05,
      "loss": 0.4615,
      "step": 4806
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8851457834243774,
      "learning_rate": 1.924035347377571e-05,
      "loss": 0.6432,
      "step": 4807
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9913930892944336,
      "learning_rate": 1.9239964080355763e-05,
      "loss": 0.4658,
      "step": 4808
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9112498760223389,
      "learning_rate": 1.9239574591102617e-05,
      "loss": 0.5853,
      "step": 4809
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.786943793296814,
      "learning_rate": 1.9239185006020303e-05,
      "loss": 0.4917,
      "step": 4810
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0063819885253906,
      "learning_rate": 1.9238795325112867e-05,
      "loss": 0.5422,
      "step": 4811
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8271187543869019,
      "learning_rate": 1.9238405548384352e-05,
      "loss": 0.4815,
      "step": 4812
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.82125723361969,
      "learning_rate": 1.92380156758388e-05,
      "loss": 0.5744,
      "step": 4813
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8425661325454712,
      "learning_rate": 1.923762570748025e-05,
      "loss": 0.4886,
      "step": 4814
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0101397037506104,
      "learning_rate": 1.923723564331275e-05,
      "loss": 0.5255,
      "step": 4815
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9391164779663086,
      "learning_rate": 1.9236845483340346e-05,
      "loss": 0.4903,
      "step": 4816
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9195966720581055,
      "learning_rate": 1.9236455227567085e-05,
      "loss": 0.5693,
      "step": 4817
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7774921655654907,
      "learning_rate": 1.923606487599701e-05,
      "loss": 0.4742,
      "step": 4818
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.85361647605896,
      "learning_rate": 1.9235674428634174e-05,
      "loss": 0.5183,
      "step": 4819
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8709250688552856,
      "learning_rate": 1.9235283885482628e-05,
      "loss": 0.5333,
      "step": 4820
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9185595512390137,
      "learning_rate": 1.923489324654642e-05,
      "loss": 0.5232,
      "step": 4821
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.854774832725525,
      "learning_rate": 1.9234502511829598e-05,
      "loss": 0.5024,
      "step": 4822
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0844342708587646,
      "learning_rate": 1.923411168133622e-05,
      "loss": 0.6358,
      "step": 4823
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9149342775344849,
      "learning_rate": 1.923372075507034e-05,
      "loss": 0.4704,
      "step": 4824
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8419533967971802,
      "learning_rate": 1.9233329733036008e-05,
      "loss": 0.5527,
      "step": 4825
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9232183694839478,
      "learning_rate": 1.9232938615237283e-05,
      "loss": 0.5356,
      "step": 4826
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9265234470367432,
      "learning_rate": 1.9232547401678217e-05,
      "loss": 0.6004,
      "step": 4827
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.860949158668518,
      "learning_rate": 1.9232156092362876e-05,
      "loss": 0.4916,
      "step": 4828
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8457632064819336,
      "learning_rate": 1.9231764687295313e-05,
      "loss": 0.5487,
      "step": 4829
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9218480587005615,
      "learning_rate": 1.9231373186479585e-05,
      "loss": 0.5247,
      "step": 4830
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8582388162612915,
      "learning_rate": 1.9230981589919756e-05,
      "loss": 0.5353,
      "step": 4831
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7892273664474487,
      "learning_rate": 1.9230589897619888e-05,
      "loss": 0.4579,
      "step": 4832
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8175959587097168,
      "learning_rate": 1.9230198109584043e-05,
      "loss": 0.5062,
      "step": 4833
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7395566701889038,
      "learning_rate": 1.922980622581628e-05,
      "loss": 0.5277,
      "step": 4834
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9275590181350708,
      "learning_rate": 1.922941424632067e-05,
      "loss": 0.5828,
      "step": 4835
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9228463172912598,
      "learning_rate": 1.9229022171101274e-05,
      "loss": 0.5201,
      "step": 4836
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8300306797027588,
      "learning_rate": 1.9228630000162162e-05,
      "loss": 0.5545,
      "step": 4837
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8293507099151611,
      "learning_rate": 1.92282377335074e-05,
      "loss": 0.5006,
      "step": 4838
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8211050033569336,
      "learning_rate": 1.9227845371141054e-05,
      "loss": 0.5105,
      "step": 4839
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7337981462478638,
      "learning_rate": 1.9227452913067196e-05,
      "loss": 0.5066,
      "step": 4840
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0791802406311035,
      "learning_rate": 1.9227060359289894e-05,
      "loss": 0.5244,
      "step": 4841
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.913557767868042,
      "learning_rate": 1.9226667709813223e-05,
      "loss": 0.4723,
      "step": 4842
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8472745418548584,
      "learning_rate": 1.9226274964641258e-05,
      "loss": 0.5648,
      "step": 4843
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7479610443115234,
      "learning_rate": 1.9225882123778062e-05,
      "loss": 0.4585,
      "step": 4844
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7337415218353271,
      "learning_rate": 1.9225489187227716e-05,
      "loss": 0.5299,
      "step": 4845
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.754441738128662,
      "learning_rate": 1.9225096154994296e-05,
      "loss": 0.4766,
      "step": 4846
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9239248037338257,
      "learning_rate": 1.922470302708188e-05,
      "loss": 0.5741,
      "step": 4847
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.842492699623108,
      "learning_rate": 1.922430980349454e-05,
      "loss": 0.473,
      "step": 4848
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8904180526733398,
      "learning_rate": 1.9223916484236357e-05,
      "loss": 0.5664,
      "step": 4849
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8897197246551514,
      "learning_rate": 1.9223523069311408e-05,
      "loss": 0.5065,
      "step": 4850
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8453471660614014,
      "learning_rate": 1.922312955872378e-05,
      "loss": 0.5908,
      "step": 4851
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8447775840759277,
      "learning_rate": 1.922273595247755e-05,
      "loss": 0.5018,
      "step": 4852
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8006646633148193,
      "learning_rate": 1.9222342250576797e-05,
      "loss": 0.5202,
      "step": 4853
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9127144813537598,
      "learning_rate": 1.922194845302561e-05,
      "loss": 0.5724,
      "step": 4854
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.015915870666504,
      "learning_rate": 1.9221554559828066e-05,
      "loss": 0.6467,
      "step": 4855
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6981751918792725,
      "learning_rate": 1.922116057098826e-05,
      "loss": 0.4554,
      "step": 4856
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8892862796783447,
      "learning_rate": 1.922076648651027e-05,
      "loss": 0.5232,
      "step": 4857
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8383493423461914,
      "learning_rate": 1.922037230639819e-05,
      "loss": 0.5204,
      "step": 4858
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.963261365890503,
      "learning_rate": 1.9219978030656103e-05,
      "loss": 0.6547,
      "step": 4859
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8732820749282837,
      "learning_rate": 1.9219583659288103e-05,
      "loss": 0.4516,
      "step": 4860
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9281063079833984,
      "learning_rate": 1.921918919229828e-05,
      "loss": 0.5423,
      "step": 4861
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7794612646102905,
      "learning_rate": 1.9218794629690713e-05,
      "loss": 0.5254,
      "step": 4862
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.19511342048645,
      "learning_rate": 1.9218399971469513e-05,
      "loss": 0.6672,
      "step": 4863
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.833837628364563,
      "learning_rate": 1.921800521763876e-05,
      "loss": 0.4828,
      "step": 4864
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9814289808273315,
      "learning_rate": 1.9217610368202553e-05,
      "loss": 0.6405,
      "step": 4865
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7471671104431152,
      "learning_rate": 1.9217215423164986e-05,
      "loss": 0.4952,
      "step": 4866
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.782528281211853,
      "learning_rate": 1.9216820382530162e-05,
      "loss": 0.5717,
      "step": 4867
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8133972883224487,
      "learning_rate": 1.9216425246302164e-05,
      "loss": 0.4906,
      "step": 4868
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1142640113830566,
      "learning_rate": 1.9216030014485106e-05,
      "loss": 0.6779,
      "step": 4869
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7188435792922974,
      "learning_rate": 1.9215634687083073e-05,
      "loss": 0.4273,
      "step": 4870
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9057703018188477,
      "learning_rate": 1.9215239264100176e-05,
      "loss": 0.5809,
      "step": 4871
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7516309022903442,
      "learning_rate": 1.921484374554051e-05,
      "loss": 0.425,
      "step": 4872
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.869159460067749,
      "learning_rate": 1.921444813140818e-05,
      "loss": 0.5594,
      "step": 4873
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8073371648788452,
      "learning_rate": 1.9214052421707287e-05,
      "loss": 0.4934,
      "step": 4874
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0944314002990723,
      "learning_rate": 1.9213656616441937e-05,
      "loss": 0.6126,
      "step": 4875
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7237849235534668,
      "learning_rate": 1.9213260715616235e-05,
      "loss": 0.4476,
      "step": 4876
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.871024250984192,
      "learning_rate": 1.921286471923429e-05,
      "loss": 0.5502,
      "step": 4877
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7668277025222778,
      "learning_rate": 1.92124686273002e-05,
      "loss": 0.4507,
      "step": 4878
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9218271970748901,
      "learning_rate": 1.921207243981808e-05,
      "loss": 0.5543,
      "step": 4879
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9214142560958862,
      "learning_rate": 1.9211676156792038e-05,
      "loss": 0.4717,
      "step": 4880
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8358973264694214,
      "learning_rate": 1.9211279778226184e-05,
      "loss": 0.5514,
      "step": 4881
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8970569372177124,
      "learning_rate": 1.921088330412463e-05,
      "loss": 0.4828,
      "step": 4882
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8561465740203857,
      "learning_rate": 1.9210486734491488e-05,
      "loss": 0.5441,
      "step": 4883
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8857834339141846,
      "learning_rate": 1.921009006933087e-05,
      "loss": 0.5463,
      "step": 4884
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9970999956130981,
      "learning_rate": 1.920969330864689e-05,
      "loss": 0.6492,
      "step": 4885
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.718985676765442,
      "learning_rate": 1.9209296452443663e-05,
      "loss": 0.4579,
      "step": 4886
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.83659827709198,
      "learning_rate": 1.9208899500725304e-05,
      "loss": 0.6062,
      "step": 4887
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7681471109390259,
      "learning_rate": 1.9208502453495934e-05,
      "loss": 0.5221,
      "step": 4888
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8842253684997559,
      "learning_rate": 1.9208105310759665e-05,
      "loss": 0.5714,
      "step": 4889
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.824210524559021,
      "learning_rate": 1.9207708072520624e-05,
      "loss": 0.501,
      "step": 4890
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8779094219207764,
      "learning_rate": 1.920731073878292e-05,
      "loss": 0.5631,
      "step": 4891
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8325783014297485,
      "learning_rate": 1.9206913309550687e-05,
      "loss": 0.4918,
      "step": 4892
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9265424013137817,
      "learning_rate": 1.9206515784828038e-05,
      "loss": 0.592,
      "step": 4893
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.861007571220398,
      "learning_rate": 1.92061181646191e-05,
      "loss": 0.5133,
      "step": 4894
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.03877329826355,
      "learning_rate": 1.9205720448927993e-05,
      "loss": 0.6061,
      "step": 4895
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.844435453414917,
      "learning_rate": 1.9205322637758845e-05,
      "loss": 0.5194,
      "step": 4896
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8958325386047363,
      "learning_rate": 1.920492473111578e-05,
      "loss": 0.5409,
      "step": 4897
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9055585861206055,
      "learning_rate": 1.9204526729002924e-05,
      "loss": 0.546,
      "step": 4898
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.972044825553894,
      "learning_rate": 1.9204128631424415e-05,
      "loss": 0.5986,
      "step": 4899
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.761129379272461,
      "learning_rate": 1.9203730438384368e-05,
      "loss": 0.4601,
      "step": 4900
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9156755208969116,
      "learning_rate": 1.920333214988692e-05,
      "loss": 0.5493,
      "step": 4901
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7791491746902466,
      "learning_rate": 1.92029337659362e-05,
      "loss": 0.515,
      "step": 4902
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.902208924293518,
      "learning_rate": 1.9202535286536342e-05,
      "loss": 0.5828,
      "step": 4903
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.795875906944275,
      "learning_rate": 1.9202136711691474e-05,
      "loss": 0.4651,
      "step": 4904
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9523429870605469,
      "learning_rate": 1.920173804140574e-05,
      "loss": 0.5743,
      "step": 4905
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8827966451644897,
      "learning_rate": 1.9201339275683263e-05,
      "loss": 0.555,
      "step": 4906
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8725911378860474,
      "learning_rate": 1.9200940414528184e-05,
      "loss": 0.5419,
      "step": 4907
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7629190683364868,
      "learning_rate": 1.9200541457944642e-05,
      "loss": 0.449,
      "step": 4908
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.886344313621521,
      "learning_rate": 1.920014240593677e-05,
      "loss": 0.5313,
      "step": 4909
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9555606842041016,
      "learning_rate": 1.9199743258508712e-05,
      "loss": 0.4936,
      "step": 4910
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7046841382980347,
      "learning_rate": 1.9199344015664605e-05,
      "loss": 0.5345,
      "step": 4911
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8944287300109863,
      "learning_rate": 1.919894467740859e-05,
      "loss": 0.5024,
      "step": 4912
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9768534898757935,
      "learning_rate": 1.919854524374481e-05,
      "loss": 0.5613,
      "step": 4913
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8983497619628906,
      "learning_rate": 1.91981457146774e-05,
      "loss": 0.6049,
      "step": 4914
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9063310623168945,
      "learning_rate": 1.919774609021052e-05,
      "loss": 0.5773,
      "step": 4915
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7423862218856812,
      "learning_rate": 1.91973463703483e-05,
      "loss": 0.4737,
      "step": 4916
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7210795879364014,
      "learning_rate": 1.919694655509489e-05,
      "loss": 0.5503,
      "step": 4917
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8556941747665405,
      "learning_rate": 1.9196546644454438e-05,
      "loss": 0.4887,
      "step": 4918
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7546738386154175,
      "learning_rate": 1.9196146638431094e-05,
      "loss": 0.5241,
      "step": 4919
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8501265048980713,
      "learning_rate": 1.9195746537029e-05,
      "loss": 0.5959,
      "step": 4920
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8473583459854126,
      "learning_rate": 1.9195346340252313e-05,
      "loss": 0.5881,
      "step": 4921
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8136842250823975,
      "learning_rate": 1.9194946048105177e-05,
      "loss": 0.541,
      "step": 4922
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8720300197601318,
      "learning_rate": 1.9194545660591753e-05,
      "loss": 0.5621,
      "step": 4923
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7734110355377197,
      "learning_rate": 1.9194145177716182e-05,
      "loss": 0.5207,
      "step": 4924
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8289088010787964,
      "learning_rate": 1.9193744599482624e-05,
      "loss": 0.6088,
      "step": 4925
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.711186408996582,
      "learning_rate": 1.9193343925895234e-05,
      "loss": 0.4441,
      "step": 4926
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7827064990997314,
      "learning_rate": 1.9192943156958168e-05,
      "loss": 0.5094,
      "step": 4927
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7748174667358398,
      "learning_rate": 1.919254229267558e-05,
      "loss": 0.5224,
      "step": 4928
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8873215913772583,
      "learning_rate": 1.9192141333051628e-05,
      "loss": 0.5544,
      "step": 4929
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8967487812042236,
      "learning_rate": 1.9191740278090472e-05,
      "loss": 0.5081,
      "step": 4930
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8731462955474854,
      "learning_rate": 1.9191339127796273e-05,
      "loss": 0.574,
      "step": 4931
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7259504795074463,
      "learning_rate": 1.9190937882173187e-05,
      "loss": 0.4956,
      "step": 4932
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8418852090835571,
      "learning_rate": 1.9190536541225377e-05,
      "loss": 0.5909,
      "step": 4933
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.713331699371338,
      "learning_rate": 1.9190135104957008e-05,
      "loss": 0.4437,
      "step": 4934
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.841333031654358,
      "learning_rate": 1.9189733573372242e-05,
      "loss": 0.5444,
      "step": 4935
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.76797354221344,
      "learning_rate": 1.9189331946475246e-05,
      "loss": 0.465,
      "step": 4936
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.894454002380371,
      "learning_rate": 1.918893022427018e-05,
      "loss": 0.578,
      "step": 4937
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.009838104248047,
      "learning_rate": 1.9188528406761214e-05,
      "loss": 0.4631,
      "step": 4938
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9908374547958374,
      "learning_rate": 1.918812649395252e-05,
      "loss": 0.5746,
      "step": 4939
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8148987293243408,
      "learning_rate": 1.9187724485848254e-05,
      "loss": 0.4868,
      "step": 4940
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7836540937423706,
      "learning_rate": 1.91873223824526e-05,
      "loss": 0.5528,
      "step": 4941
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.836881399154663,
      "learning_rate": 1.918692018376972e-05,
      "loss": 0.5643,
      "step": 4942
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9137656688690186,
      "learning_rate": 1.9186517889803788e-05,
      "loss": 0.536,
      "step": 4943
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.934380054473877,
      "learning_rate": 1.9186115500558972e-05,
      "loss": 0.5663,
      "step": 4944
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.867645502090454,
      "learning_rate": 1.918571301603945e-05,
      "loss": 0.6205,
      "step": 4945
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8663609027862549,
      "learning_rate": 1.9185310436249396e-05,
      "loss": 0.4898,
      "step": 4946
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.870807409286499,
      "learning_rate": 1.9184907761192984e-05,
      "loss": 0.584,
      "step": 4947
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7233831882476807,
      "learning_rate": 1.918450499087439e-05,
      "loss": 0.4673,
      "step": 4948
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.209056854248047,
      "learning_rate": 1.9184102125297793e-05,
      "loss": 0.5689,
      "step": 4949
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8362202644348145,
      "learning_rate": 1.9183699164467376e-05,
      "loss": 0.4619,
      "step": 4950
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8288469314575195,
      "learning_rate": 1.9183296108387307e-05,
      "loss": 0.5061,
      "step": 4951
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.815363883972168,
      "learning_rate": 1.9182892957061775e-05,
      "loss": 0.4793,
      "step": 4952
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.959865689277649,
      "learning_rate": 1.9182489710494958e-05,
      "loss": 0.5493,
      "step": 4953
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.837193250656128,
      "learning_rate": 1.918208636869104e-05,
      "loss": 0.4992,
      "step": 4954
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.915542483329773,
      "learning_rate": 1.9181682931654202e-05,
      "loss": 0.5598,
      "step": 4955
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.798477292060852,
      "learning_rate": 1.918127939938863e-05,
      "loss": 0.4834,
      "step": 4956
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8299322128295898,
      "learning_rate": 1.9180875771898512e-05,
      "loss": 0.5596,
      "step": 4957
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9690463542938232,
      "learning_rate": 1.918047204918803e-05,
      "loss": 0.5291,
      "step": 4958
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8483468294143677,
      "learning_rate": 1.918006823126137e-05,
      "loss": 0.5187,
      "step": 4959
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8827253580093384,
      "learning_rate": 1.9179664318122726e-05,
      "loss": 0.4969,
      "step": 4960
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.087249755859375,
      "learning_rate": 1.917926030977628e-05,
      "loss": 0.5379,
      "step": 4961
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9007235765457153,
      "learning_rate": 1.9178856206226227e-05,
      "loss": 0.4625,
      "step": 4962
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8473247289657593,
      "learning_rate": 1.917845200747676e-05,
      "loss": 0.5113,
      "step": 4963
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8612703084945679,
      "learning_rate": 1.917804771353207e-05,
      "loss": 0.5059,
      "step": 4964
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8210859298706055,
      "learning_rate": 1.9177643324396348e-05,
      "loss": 0.5946,
      "step": 4965
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9016973972320557,
      "learning_rate": 1.9177238840073786e-05,
      "loss": 0.5639,
      "step": 4966
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9044116735458374,
      "learning_rate": 1.9176834260568584e-05,
      "loss": 0.4959,
      "step": 4967
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8919938802719116,
      "learning_rate": 1.9176429585884935e-05,
      "loss": 0.5397,
      "step": 4968
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.06219482421875,
      "learning_rate": 1.9176024816027038e-05,
      "loss": 0.8043,
      "step": 4969
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8080519437789917,
      "learning_rate": 1.9175619950999088e-05,
      "loss": 0.9288,
      "step": 4970
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9835273027420044,
      "learning_rate": 1.917521499080529e-05,
      "loss": 0.5381,
      "step": 4971
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.05853271484375,
      "learning_rate": 1.917480993544984e-05,
      "loss": 0.5003,
      "step": 4972
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8113566637039185,
      "learning_rate": 1.917440478493694e-05,
      "loss": 0.5039,
      "step": 4973
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7724827527999878,
      "learning_rate": 1.917399953927079e-05,
      "loss": 0.4839,
      "step": 4974
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8909276723861694,
      "learning_rate": 1.9173594198455597e-05,
      "loss": 0.5089,
      "step": 4975
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8832892179489136,
      "learning_rate": 1.917318876249556e-05,
      "loss": 0.4868,
      "step": 4976
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9914780855178833,
      "learning_rate": 1.917278323139489e-05,
      "loss": 0.6025,
      "step": 4977
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9631558656692505,
      "learning_rate": 1.9172377605157787e-05,
      "loss": 0.5533,
      "step": 4978
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8704941272735596,
      "learning_rate": 1.9171971883788464e-05,
      "loss": 0.5211,
      "step": 4979
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.021596908569336,
      "learning_rate": 1.9171566067291125e-05,
      "loss": 0.4628,
      "step": 4980
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7777020931243896,
      "learning_rate": 1.9171160155669976e-05,
      "loss": 0.5049,
      "step": 4981
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.875245213508606,
      "learning_rate": 1.9170754148929233e-05,
      "loss": 0.4847,
      "step": 4982
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.000251054763794,
      "learning_rate": 1.9170348047073107e-05,
      "loss": 0.5133,
      "step": 4983
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.021825075149536,
      "learning_rate": 1.9169941850105806e-05,
      "loss": 0.5483,
      "step": 4984
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8228665590286255,
      "learning_rate": 1.9169535558031546e-05,
      "loss": 0.5409,
      "step": 4985
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0049047470092773,
      "learning_rate": 1.9169129170854537e-05,
      "loss": 0.6018,
      "step": 4986
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.916919469833374,
      "learning_rate": 1.9168722688578997e-05,
      "loss": 0.5512,
      "step": 4987
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8723721504211426,
      "learning_rate": 1.9168316111209142e-05,
      "loss": 0.4814,
      "step": 4988
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8461518287658691,
      "learning_rate": 1.9167909438749187e-05,
      "loss": 0.5418,
      "step": 4989
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8352580070495605,
      "learning_rate": 1.9167502671203352e-05,
      "loss": 0.5208,
      "step": 4990
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8232181072235107,
      "learning_rate": 1.9167095808575854e-05,
      "loss": 0.5237,
      "step": 4991
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8102171421051025,
      "learning_rate": 1.9166688850870912e-05,
      "loss": 0.4696,
      "step": 4992
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.287506103515625,
      "learning_rate": 1.916628179809275e-05,
      "loss": 0.6343,
      "step": 4993
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7560216188430786,
      "learning_rate": 1.916587465024559e-05,
      "loss": 0.4492,
      "step": 4994
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9020755290985107,
      "learning_rate": 1.916546740733365e-05,
      "loss": 0.5491,
      "step": 4995
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9025837182998657,
      "learning_rate": 1.9165060069361157e-05,
      "loss": 0.5892,
      "step": 4996
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.714767575263977,
      "learning_rate": 1.9164652636332338e-05,
      "loss": 0.5077,
      "step": 4997
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.84165358543396,
      "learning_rate": 1.9164245108251412e-05,
      "loss": 0.4737,
      "step": 4998
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9819658994674683,
      "learning_rate": 1.9163837485122615e-05,
      "loss": 0.5753,
      "step": 4999
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.869085669517517,
      "learning_rate": 1.9163429766950166e-05,
      "loss": 0.4572,
      "step": 5000
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.912899136543274,
      "learning_rate": 1.9163021953738296e-05,
      "loss": 0.5957,
      "step": 5001
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.848681092262268,
      "learning_rate": 1.916261404549124e-05,
      "loss": 0.4482,
      "step": 5002
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8822295665740967,
      "learning_rate": 1.916220604221322e-05,
      "loss": 0.6187,
      "step": 5003
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.811444640159607,
      "learning_rate": 1.9161797943908477e-05,
      "loss": 0.4372,
      "step": 5004
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9997395277023315,
      "learning_rate": 1.9161389750581237e-05,
      "loss": 0.5949,
      "step": 5005
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.834157109260559,
      "learning_rate": 1.9160981462235732e-05,
      "loss": 0.4988,
      "step": 5006
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3664939403533936,
      "learning_rate": 1.9160573078876206e-05,
      "loss": 0.9593,
      "step": 5007
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.171679735183716,
      "learning_rate": 1.9160164600506886e-05,
      "loss": 0.959,
      "step": 5008
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8885527849197388,
      "learning_rate": 1.915975602713201e-05,
      "loss": 0.5421,
      "step": 5009
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8797684907913208,
      "learning_rate": 1.9159347358755817e-05,
      "loss": 0.4717,
      "step": 5010
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8193329572677612,
      "learning_rate": 1.9158938595382546e-05,
      "loss": 0.5339,
      "step": 5011
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8337633609771729,
      "learning_rate": 1.9158529737016434e-05,
      "loss": 0.5628,
      "step": 5012
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7859679460525513,
      "learning_rate": 1.915812078366173e-05,
      "loss": 0.4963,
      "step": 5013
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.782004714012146,
      "learning_rate": 1.9157711735322662e-05,
      "loss": 0.4736,
      "step": 5014
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.794215440750122,
      "learning_rate": 1.9157302592003483e-05,
      "loss": 0.568,
      "step": 5015
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7538584470748901,
      "learning_rate": 1.9156893353708433e-05,
      "loss": 0.4617,
      "step": 5016
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7472474575042725,
      "learning_rate": 1.9156484020441753e-05,
      "loss": 0.5169,
      "step": 5017
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7329312562942505,
      "learning_rate": 1.9156074592207693e-05,
      "loss": 0.4725,
      "step": 5018
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.89667546749115,
      "learning_rate": 1.91556650690105e-05,
      "loss": 0.6001,
      "step": 5019
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8505678176879883,
      "learning_rate": 1.9155255450854418e-05,
      "loss": 0.4549,
      "step": 5020
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.920030117034912,
      "learning_rate": 1.9154845737743696e-05,
      "loss": 0.6052,
      "step": 5021
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.874977946281433,
      "learning_rate": 1.9154435929682587e-05,
      "loss": 0.505,
      "step": 5022
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.804725170135498,
      "learning_rate": 1.9154026026675335e-05,
      "loss": 0.5104,
      "step": 5023
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8105870485305786,
      "learning_rate": 1.9153616028726197e-05,
      "loss": 0.531,
      "step": 5024
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8678632974624634,
      "learning_rate": 1.9153205935839427e-05,
      "loss": 0.552,
      "step": 5025
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8656107187271118,
      "learning_rate": 1.9152795748019272e-05,
      "loss": 0.4707,
      "step": 5026
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7739049196243286,
      "learning_rate": 1.915238546526999e-05,
      "loss": 0.5619,
      "step": 5027
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7198684215545654,
      "learning_rate": 1.915197508759583e-05,
      "loss": 0.4899,
      "step": 5028
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9530571699142456,
      "learning_rate": 1.9151564615001058e-05,
      "loss": 0.5686,
      "step": 5029
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8757463693618774,
      "learning_rate": 1.9151154047489927e-05,
      "loss": 0.5202,
      "step": 5030
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8824608325958252,
      "learning_rate": 1.9150743385066698e-05,
      "loss": 0.5213,
      "step": 5031
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8156874179840088,
      "learning_rate": 1.9150332627735622e-05,
      "loss": 0.4478,
      "step": 5032
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9199634790420532,
      "learning_rate": 1.9149921775500966e-05,
      "loss": 0.5443,
      "step": 5033
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7710365056991577,
      "learning_rate": 1.914951082836699e-05,
      "loss": 0.4688,
      "step": 5034
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7538219690322876,
      "learning_rate": 1.9149099786337958e-05,
      "loss": 0.5209,
      "step": 5035
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.801772952079773,
      "learning_rate": 1.9148688649418127e-05,
      "loss": 0.4745,
      "step": 5036
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.845483422279358,
      "learning_rate": 1.9148277417611768e-05,
      "loss": 0.5101,
      "step": 5037
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8072370290756226,
      "learning_rate": 1.9147866090923143e-05,
      "loss": 0.5159,
      "step": 5038
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.881081461906433,
      "learning_rate": 1.9147454669356514e-05,
      "loss": 0.5828,
      "step": 5039
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.443042278289795,
      "learning_rate": 1.914704315291616e-05,
      "loss": 0.5014,
      "step": 5040
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9786882400512695,
      "learning_rate": 1.9146631541606336e-05,
      "loss": 0.569,
      "step": 5041
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7706791162490845,
      "learning_rate": 1.914621983543132e-05,
      "loss": 0.5216,
      "step": 5042
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9148600101470947,
      "learning_rate": 1.914580803439538e-05,
      "loss": 0.5573,
      "step": 5043
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.833533525466919,
      "learning_rate": 1.9145396138502783e-05,
      "loss": 0.5394,
      "step": 5044
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8780457973480225,
      "learning_rate": 1.9144984147757803e-05,
      "loss": 0.592,
      "step": 5045
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8080989122390747,
      "learning_rate": 1.9144572062164717e-05,
      "loss": 0.4509,
      "step": 5046
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7241623401641846,
      "learning_rate": 1.9144159881727793e-05,
      "loss": 0.5188,
      "step": 5047
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8386812210083008,
      "learning_rate": 1.914374760645131e-05,
      "loss": 0.5859,
      "step": 5048
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8238452672958374,
      "learning_rate": 1.9143335236339542e-05,
      "loss": 0.518,
      "step": 5049
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9078819751739502,
      "learning_rate": 1.9142922771396767e-05,
      "loss": 0.491,
      "step": 5050
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9799532890319824,
      "learning_rate": 1.9142510211627264e-05,
      "loss": 0.5872,
      "step": 5051
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.828300952911377,
      "learning_rate": 1.914209755703531e-05,
      "loss": 0.4893,
      "step": 5052
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9602257013320923,
      "learning_rate": 1.9141684807625183e-05,
      "loss": 0.5634,
      "step": 5053
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8224215507507324,
      "learning_rate": 1.914127196340117e-05,
      "loss": 0.4476,
      "step": 5054
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8724242448806763,
      "learning_rate": 1.9140859024367545e-05,
      "loss": 0.598,
      "step": 5055
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8386056423187256,
      "learning_rate": 1.9140445990528596e-05,
      "loss": 0.5132,
      "step": 5056
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9001104831695557,
      "learning_rate": 1.9140032861888608e-05,
      "loss": 0.5981,
      "step": 5057
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7479822635650635,
      "learning_rate": 1.913961963845186e-05,
      "loss": 0.5027,
      "step": 5058
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.732102870941162,
      "learning_rate": 1.9139206320222647e-05,
      "loss": 0.5204,
      "step": 5059
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.839618444442749,
      "learning_rate": 1.9138792907205245e-05,
      "loss": 0.4903,
      "step": 5060
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8129483461380005,
      "learning_rate": 1.9138379399403948e-05,
      "loss": 0.5716,
      "step": 5061
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7758256196975708,
      "learning_rate": 1.9137965796823046e-05,
      "loss": 0.4881,
      "step": 5062
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7373701333999634,
      "learning_rate": 1.913755209946682e-05,
      "loss": 0.505,
      "step": 5063
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7817161083221436,
      "learning_rate": 1.9137138307339573e-05,
      "loss": 0.4907,
      "step": 5064
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7676554918289185,
      "learning_rate": 1.913672442044559e-05,
      "loss": 0.6125,
      "step": 5065
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.737910270690918,
      "learning_rate": 1.9136310438789165e-05,
      "loss": 0.476,
      "step": 5066
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8214221000671387,
      "learning_rate": 1.913589636237459e-05,
      "loss": 0.5173,
      "step": 5067
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7381618022918701,
      "learning_rate": 1.9135482191206162e-05,
      "loss": 0.5002,
      "step": 5068
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7493013143539429,
      "learning_rate": 1.9135067925288172e-05,
      "loss": 0.5412,
      "step": 5069
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.874690294265747,
      "learning_rate": 1.9134653564624922e-05,
      "loss": 0.5066,
      "step": 5070
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8406561613082886,
      "learning_rate": 1.9134239109220708e-05,
      "loss": 0.5572,
      "step": 5071
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.759098768234253,
      "learning_rate": 1.9133824559079826e-05,
      "loss": 0.4645,
      "step": 5072
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.803952693939209,
      "learning_rate": 1.913340991420658e-05,
      "loss": 0.488,
      "step": 5073
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9843124151229858,
      "learning_rate": 1.9132995174605265e-05,
      "loss": 0.5686,
      "step": 5074
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9063212871551514,
      "learning_rate": 1.9132580340280188e-05,
      "loss": 0.565,
      "step": 5075
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8112804889678955,
      "learning_rate": 1.9132165411235652e-05,
      "loss": 0.5531,
      "step": 5076
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8376106023788452,
      "learning_rate": 1.9131750387475952e-05,
      "loss": 0.5801,
      "step": 5077
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0143659114837646,
      "learning_rate": 1.91313352690054e-05,
      "loss": 0.4526,
      "step": 5078
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7960325479507446,
      "learning_rate": 1.91309200558283e-05,
      "loss": 0.4953,
      "step": 5079
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7684894800186157,
      "learning_rate": 1.9130504747948958e-05,
      "loss": 0.4784,
      "step": 5080
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9038015604019165,
      "learning_rate": 1.9130089345371682e-05,
      "loss": 0.5298,
      "step": 5081
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.873427152633667,
      "learning_rate": 1.9129673848100778e-05,
      "loss": 0.4963,
      "step": 5082
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9026607275009155,
      "learning_rate": 1.9129258256140556e-05,
      "loss": 0.5242,
      "step": 5083
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7820987701416016,
      "learning_rate": 1.9128842569495328e-05,
      "loss": 0.4659,
      "step": 5084
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7918646335601807,
      "learning_rate": 1.9128426788169408e-05,
      "loss": 0.5322,
      "step": 5085
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7731168270111084,
      "learning_rate": 1.9128010912167102e-05,
      "loss": 0.4753,
      "step": 5086
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9033371210098267,
      "learning_rate": 1.912759494149273e-05,
      "loss": 0.5922,
      "step": 5087
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.758414387702942,
      "learning_rate": 1.91271788761506e-05,
      "loss": 0.4588,
      "step": 5088
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.954348087310791,
      "learning_rate": 1.912676271614503e-05,
      "loss": 0.6061,
      "step": 5089
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.806420087814331,
      "learning_rate": 1.9126346461480337e-05,
      "loss": 0.4674,
      "step": 5090
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8227123022079468,
      "learning_rate": 1.9125930112160838e-05,
      "loss": 0.5086,
      "step": 5091
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.775588035583496,
      "learning_rate": 1.912551366819085e-05,
      "loss": 0.4627,
      "step": 5092
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9799398183822632,
      "learning_rate": 1.912509712957469e-05,
      "loss": 0.5614,
      "step": 5093
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7481279373168945,
      "learning_rate": 1.9124680496316684e-05,
      "loss": 0.4728,
      "step": 5094
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8724822998046875,
      "learning_rate": 1.912426376842115e-05,
      "loss": 0.538,
      "step": 5095
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6887223720550537,
      "learning_rate": 1.9123846945892413e-05,
      "loss": 0.4842,
      "step": 5096
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7893767356872559,
      "learning_rate": 1.912343002873479e-05,
      "loss": 0.5684,
      "step": 5097
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8083959817886353,
      "learning_rate": 1.9123013016952607e-05,
      "loss": 0.4881,
      "step": 5098
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8117293119430542,
      "learning_rate": 1.912259591055019e-05,
      "loss": 0.5473,
      "step": 5099
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.760789394378662,
      "learning_rate": 1.912217870953187e-05,
      "loss": 0.4611,
      "step": 5100
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.784502387046814,
      "learning_rate": 1.912176141390197e-05,
      "loss": 0.5267,
      "step": 5101
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.803989291191101,
      "learning_rate": 1.9121344023664813e-05,
      "loss": 0.4614,
      "step": 5102
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9028985500335693,
      "learning_rate": 1.9120926538824735e-05,
      "loss": 0.5642,
      "step": 5103
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.946922779083252,
      "learning_rate": 1.9120508959386063e-05,
      "loss": 0.5042,
      "step": 5104
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9356697797775269,
      "learning_rate": 1.912009128535313e-05,
      "loss": 0.572,
      "step": 5105
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8021684885025024,
      "learning_rate": 1.911967351673027e-05,
      "loss": 0.4901,
      "step": 5106
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8174066543579102,
      "learning_rate": 1.9119255653521808e-05,
      "loss": 0.542,
      "step": 5107
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9027677774429321,
      "learning_rate": 1.9118837695732082e-05,
      "loss": 0.5046,
      "step": 5108
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8513597249984741,
      "learning_rate": 1.9118419643365425e-05,
      "loss": 0.5714,
      "step": 5109
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0207061767578125,
      "learning_rate": 1.911800149642618e-05,
      "loss": 0.5309,
      "step": 5110
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8117691278457642,
      "learning_rate": 1.911758325491868e-05,
      "loss": 0.5011,
      "step": 5111
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9047366380691528,
      "learning_rate": 1.911716491884726e-05,
      "loss": 0.5661,
      "step": 5112
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7841380834579468,
      "learning_rate": 1.911674648821626e-05,
      "loss": 0.5511,
      "step": 5113
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8622281551361084,
      "learning_rate": 1.911632796303002e-05,
      "loss": 0.5089,
      "step": 5114
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1306941509246826,
      "learning_rate": 1.911590934329288e-05,
      "loss": 0.485,
      "step": 5115
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7904233932495117,
      "learning_rate": 1.9115490629009187e-05,
      "loss": 0.5203,
      "step": 5116
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8725889921188354,
      "learning_rate": 1.911507182018328e-05,
      "loss": 0.5762,
      "step": 5117
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8796195983886719,
      "learning_rate": 1.9114652916819498e-05,
      "loss": 0.4836,
      "step": 5118
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8128254413604736,
      "learning_rate": 1.9114233918922193e-05,
      "loss": 0.5197,
      "step": 5119
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.758009910583496,
      "learning_rate": 1.9113814826495707e-05,
      "loss": 0.4759,
      "step": 5120
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9752451181411743,
      "learning_rate": 1.911339563954439e-05,
      "loss": 0.6063,
      "step": 5121
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8538037538528442,
      "learning_rate": 1.9112976358072583e-05,
      "loss": 0.482,
      "step": 5122
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7594925165176392,
      "learning_rate": 1.9112556982084643e-05,
      "loss": 0.5442,
      "step": 5123
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9177632331848145,
      "learning_rate": 1.9112137511584912e-05,
      "loss": 0.4669,
      "step": 5124
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.931626319885254,
      "learning_rate": 1.9111717946577745e-05,
      "loss": 0.5428,
      "step": 5125
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.840744137763977,
      "learning_rate": 1.911129828706749e-05,
      "loss": 0.4902,
      "step": 5126
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7958788871765137,
      "learning_rate": 1.9110878533058508e-05,
      "loss": 0.5286,
      "step": 5127
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8237117528915405,
      "learning_rate": 1.9110458684555143e-05,
      "loss": 0.5023,
      "step": 5128
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9241136312484741,
      "learning_rate": 1.9110038741561755e-05,
      "loss": 0.596,
      "step": 5129
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7586796283721924,
      "learning_rate": 1.9109618704082697e-05,
      "loss": 0.4685,
      "step": 5130
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7133469581604004,
      "learning_rate": 1.9109198572122324e-05,
      "loss": 0.5066,
      "step": 5131
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9150477647781372,
      "learning_rate": 1.9108778345684993e-05,
      "loss": 0.534,
      "step": 5132
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.862290859222412,
      "learning_rate": 1.910835802477507e-05,
      "loss": 0.5117,
      "step": 5133
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7011001110076904,
      "learning_rate": 1.910793760939691e-05,
      "loss": 0.4484,
      "step": 5134
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8756812810897827,
      "learning_rate": 1.910751709955487e-05,
      "loss": 0.5733,
      "step": 5135
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8053654432296753,
      "learning_rate": 1.910709649525331e-05,
      "loss": 0.4822,
      "step": 5136
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8704901933670044,
      "learning_rate": 1.9106675796496603e-05,
      "loss": 0.5822,
      "step": 5137
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9697997570037842,
      "learning_rate": 1.9106255003289104e-05,
      "loss": 0.5609,
      "step": 5138
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.804623007774353,
      "learning_rate": 1.910583411563518e-05,
      "loss": 0.5276,
      "step": 5139
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8268423080444336,
      "learning_rate": 1.9105413133539193e-05,
      "loss": 0.4841,
      "step": 5140
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6714519262313843,
      "learning_rate": 1.9104992057005513e-05,
      "loss": 0.4934,
      "step": 5141
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8400394916534424,
      "learning_rate": 1.9104570886038504e-05,
      "loss": 0.5599,
      "step": 5142
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.866988182067871,
      "learning_rate": 1.9104149620642534e-05,
      "loss": 0.5139,
      "step": 5143
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.951237440109253,
      "learning_rate": 1.9103728260821978e-05,
      "loss": 0.5744,
      "step": 5144
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8114243745803833,
      "learning_rate": 1.9103306806581202e-05,
      "loss": 0.5438,
      "step": 5145
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.917512059211731,
      "learning_rate": 1.9102885257924577e-05,
      "loss": 0.5011,
      "step": 5146
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.877364993095398,
      "learning_rate": 1.9102463614856474e-05,
      "loss": 0.5984,
      "step": 5147
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8692190647125244,
      "learning_rate": 1.910204187738127e-05,
      "loss": 0.528,
      "step": 5148
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.413282632827759,
      "learning_rate": 1.9101620045503337e-05,
      "loss": 0.8828,
      "step": 5149
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9135991334915161,
      "learning_rate": 1.9101198119227047e-05,
      "loss": 0.9321,
      "step": 5150
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.103381633758545,
      "learning_rate": 1.9100776098556782e-05,
      "loss": 0.5672,
      "step": 5151
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2344307899475098,
      "learning_rate": 1.9100353983496913e-05,
      "loss": 0.4678,
      "step": 5152
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0587966442108154,
      "learning_rate": 1.9099931774051824e-05,
      "loss": 0.6326,
      "step": 5153
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8040560483932495,
      "learning_rate": 1.909950947022589e-05,
      "loss": 0.4554,
      "step": 5154
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.995689034461975,
      "learning_rate": 1.9099087072023492e-05,
      "loss": 0.6395,
      "step": 5155
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8928505182266235,
      "learning_rate": 1.9098664579449007e-05,
      "loss": 0.4632,
      "step": 5156
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.880656361579895,
      "learning_rate": 1.9098241992506826e-05,
      "loss": 0.5502,
      "step": 5157
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8158420324325562,
      "learning_rate": 1.9097819311201325e-05,
      "loss": 0.5095,
      "step": 5158
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8827528953552246,
      "learning_rate": 1.9097396535536893e-05,
      "loss": 0.5342,
      "step": 5159
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8494338989257812,
      "learning_rate": 1.909697366551791e-05,
      "loss": 0.4478,
      "step": 5160
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.868194580078125,
      "learning_rate": 1.9096550701148757e-05,
      "loss": 0.5362,
      "step": 5161
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8044819831848145,
      "learning_rate": 1.9096127642433832e-05,
      "loss": 0.5406,
      "step": 5162
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9628249406814575,
      "learning_rate": 1.909570448937752e-05,
      "loss": 0.584,
      "step": 5163
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7477635145187378,
      "learning_rate": 1.9095281241984206e-05,
      "loss": 0.4509,
      "step": 5164
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0313260555267334,
      "learning_rate": 1.9094857900258277e-05,
      "loss": 0.5886,
      "step": 5165
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8412532806396484,
      "learning_rate": 1.9094434464204133e-05,
      "loss": 0.4625,
      "step": 5166
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.898390769958496,
      "learning_rate": 1.9094010933826158e-05,
      "loss": 0.5778,
      "step": 5167
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8536628484725952,
      "learning_rate": 1.909358730912875e-05,
      "loss": 0.5351,
      "step": 5168
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7939444780349731,
      "learning_rate": 1.90931635901163e-05,
      "loss": 0.5588,
      "step": 5169
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9627268314361572,
      "learning_rate": 1.9092739776793202e-05,
      "loss": 0.5504,
      "step": 5170
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.889646053314209,
      "learning_rate": 1.9092315869163853e-05,
      "loss": 0.5423,
      "step": 5171
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.923561930656433,
      "learning_rate": 1.909189186723265e-05,
      "loss": 0.5288,
      "step": 5172
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7624635696411133,
      "learning_rate": 1.9091467771003986e-05,
      "loss": 0.5289,
      "step": 5173
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.908248782157898,
      "learning_rate": 1.909104358048227e-05,
      "loss": 0.5748,
      "step": 5174
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8086910247802734,
      "learning_rate": 1.9090619295671885e-05,
      "loss": 0.501,
      "step": 5175
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8411558866500854,
      "learning_rate": 1.9090194916577248e-05,
      "loss": 0.529,
      "step": 5176
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8154104948043823,
      "learning_rate": 1.9089770443202752e-05,
      "loss": 0.5892,
      "step": 5177
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7912453413009644,
      "learning_rate": 1.9089345875552804e-05,
      "loss": 0.4937,
      "step": 5178
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.817152976989746,
      "learning_rate": 1.9088921213631803e-05,
      "loss": 0.9545,
      "step": 5179
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.269690990447998,
      "learning_rate": 1.9088496457444156e-05,
      "loss": 0.9546,
      "step": 5180
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9317328929901123,
      "learning_rate": 1.9088071606994264e-05,
      "loss": 0.6037,
      "step": 5181
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.836784839630127,
      "learning_rate": 1.908764666228654e-05,
      "loss": 0.4861,
      "step": 5182
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9176111221313477,
      "learning_rate": 1.9087221623325386e-05,
      "loss": 0.5336,
      "step": 5183
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8290084600448608,
      "learning_rate": 1.9086796490115212e-05,
      "loss": 0.5353,
      "step": 5184
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8689417839050293,
      "learning_rate": 1.908637126266043e-05,
      "loss": 0.572,
      "step": 5185
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8760790824890137,
      "learning_rate": 1.9085945940965448e-05,
      "loss": 0.4573,
      "step": 5186
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9098261594772339,
      "learning_rate": 1.9085520525034677e-05,
      "loss": 0.5469,
      "step": 5187
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8377126455307007,
      "learning_rate": 1.908509501487253e-05,
      "loss": 0.4682,
      "step": 5188
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9500447511672974,
      "learning_rate": 1.9084669410483417e-05,
      "loss": 0.5241,
      "step": 5189
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.818149209022522,
      "learning_rate": 1.9084243711871758e-05,
      "loss": 0.5036,
      "step": 5190
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7401328086853027,
      "learning_rate": 1.9083817919041968e-05,
      "loss": 0.549,
      "step": 5191
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7149584293365479,
      "learning_rate": 1.9083392031998458e-05,
      "loss": 0.4592,
      "step": 5192
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8926634788513184,
      "learning_rate": 1.9082966050745646e-05,
      "loss": 0.5616,
      "step": 5193
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7966334819793701,
      "learning_rate": 1.908253997528795e-05,
      "loss": 0.4708,
      "step": 5194
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7776235342025757,
      "learning_rate": 1.9082113805629798e-05,
      "loss": 0.5033,
      "step": 5195
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6572191715240479,
      "learning_rate": 1.9081687541775597e-05,
      "loss": 0.454,
      "step": 5196
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9713619947433472,
      "learning_rate": 1.9081261183729774e-05,
      "loss": 0.6486,
      "step": 5197
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8650548458099365,
      "learning_rate": 1.908083473149675e-05,
      "loss": 0.4447,
      "step": 5198
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9117459058761597,
      "learning_rate": 1.9080408185080954e-05,
      "loss": 0.5392,
      "step": 5199
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8813961744308472,
      "learning_rate": 1.90799815444868e-05,
      "loss": 0.5093,
      "step": 5200
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7327760457992554,
      "learning_rate": 1.907955480971872e-05,
      "loss": 0.4986,
      "step": 5201
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.892933964729309,
      "learning_rate": 1.9079127980781136e-05,
      "loss": 0.5221,
      "step": 5202
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0207502841949463,
      "learning_rate": 1.907870105767848e-05,
      "loss": 0.5573,
      "step": 5203
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.818088412284851,
      "learning_rate": 1.9078274040415173e-05,
      "loss": 0.4678,
      "step": 5204
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8032166957855225,
      "learning_rate": 1.9077846928995648e-05,
      "loss": 0.5404,
      "step": 5205
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9513617753982544,
      "learning_rate": 1.9077419723424335e-05,
      "loss": 0.5098,
      "step": 5206
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8490439653396606,
      "learning_rate": 1.9076992423705662e-05,
      "loss": 0.5883,
      "step": 5207
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8398176431655884,
      "learning_rate": 1.9076565029844065e-05,
      "loss": 0.4533,
      "step": 5208
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8608258962631226,
      "learning_rate": 1.9076137541843974e-05,
      "loss": 0.5571,
      "step": 5209
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.87156081199646,
      "learning_rate": 1.907570995970982e-05,
      "loss": 0.5166,
      "step": 5210
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.533160924911499,
      "learning_rate": 1.9075282283446045e-05,
      "loss": 0.6576,
      "step": 5211
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8233773708343506,
      "learning_rate": 1.907485451305708e-05,
      "loss": 0.4396,
      "step": 5212
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8905426263809204,
      "learning_rate": 1.907442664854736e-05,
      "loss": 0.5598,
      "step": 5213
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.810970425605774,
      "learning_rate": 1.9073998689921328e-05,
      "loss": 0.4787,
      "step": 5214
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8214316368103027,
      "learning_rate": 1.9073570637183416e-05,
      "loss": 0.559,
      "step": 5215
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.157392978668213,
      "learning_rate": 1.907314249033807e-05,
      "loss": 0.5454,
      "step": 5216
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0041723251342773,
      "learning_rate": 1.9072714249389728e-05,
      "loss": 0.545,
      "step": 5217
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.81858491897583,
      "learning_rate": 1.907228591434283e-05,
      "loss": 0.4862,
      "step": 5218
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7151615619659424,
      "learning_rate": 1.907185748520182e-05,
      "loss": 0.4932,
      "step": 5219
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7850178480148315,
      "learning_rate": 1.907142896197114e-05,
      "loss": 0.4921,
      "step": 5220
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7366126775741577,
      "learning_rate": 1.9071000344655237e-05,
      "loss": 0.5126,
      "step": 5221
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8218666315078735,
      "learning_rate": 1.9070571633258556e-05,
      "loss": 0.5127,
      "step": 5222
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8226916790008545,
      "learning_rate": 1.907014282778554e-05,
      "loss": 0.5084,
      "step": 5223
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9569251537322998,
      "learning_rate": 1.9069713928240642e-05,
      "loss": 0.5492,
      "step": 5224
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9330030679702759,
      "learning_rate": 1.9069284934628304e-05,
      "loss": 0.6058,
      "step": 5225
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9002097845077515,
      "learning_rate": 1.906885584695298e-05,
      "loss": 0.5325,
      "step": 5226
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1514246463775635,
      "learning_rate": 1.906842666521912e-05,
      "loss": 0.6737,
      "step": 5227
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8586450815200806,
      "learning_rate": 1.9067997389431173e-05,
      "loss": 0.4525,
      "step": 5228
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9164119958877563,
      "learning_rate": 1.9067568019593596e-05,
      "loss": 0.5386,
      "step": 5229
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7002309560775757,
      "learning_rate": 1.9067138555710835e-05,
      "loss": 0.4495,
      "step": 5230
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.999521255493164,
      "learning_rate": 1.906670899778735e-05,
      "loss": 0.6344,
      "step": 5231
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8816089630126953,
      "learning_rate": 1.9066279345827594e-05,
      "loss": 0.4558,
      "step": 5232
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0036542415618896,
      "learning_rate": 1.9065849599836025e-05,
      "loss": 0.6817,
      "step": 5233
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7461968660354614,
      "learning_rate": 1.9065419759817096e-05,
      "loss": 0.4435,
      "step": 5234
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9214122295379639,
      "learning_rate": 1.9064989825775267e-05,
      "loss": 0.6294,
      "step": 5235
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8022733926773071,
      "learning_rate": 1.9064559797715e-05,
      "loss": 0.4673,
      "step": 5236
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8399462699890137,
      "learning_rate": 1.9064129675640752e-05,
      "loss": 0.5209,
      "step": 5237
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8060688972473145,
      "learning_rate": 1.9063699459556986e-05,
      "loss": 0.4847,
      "step": 5238
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9029103517532349,
      "learning_rate": 1.9063269149468162e-05,
      "loss": 0.5069,
      "step": 5239
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8430954217910767,
      "learning_rate": 1.9062838745378745e-05,
      "loss": 0.4624,
      "step": 5240
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8789629936218262,
      "learning_rate": 1.9062408247293194e-05,
      "loss": 0.605,
      "step": 5241
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7869616746902466,
      "learning_rate": 1.9061977655215985e-05,
      "loss": 0.4546,
      "step": 5242
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8529601097106934,
      "learning_rate": 1.906154696915157e-05,
      "loss": 0.6016,
      "step": 5243
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.819029688835144,
      "learning_rate": 1.906111618910443e-05,
      "loss": 0.4998,
      "step": 5244
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.732591986656189,
      "learning_rate": 1.9060685315079022e-05,
      "loss": 0.5032,
      "step": 5245
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8037976026535034,
      "learning_rate": 1.906025434707982e-05,
      "loss": 0.5065,
      "step": 5246
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6799325942993164,
      "learning_rate": 1.9059823285111286e-05,
      "loss": 0.4788,
      "step": 5247
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8812172412872314,
      "learning_rate": 1.9059392129177902e-05,
      "loss": 0.5614,
      "step": 5248
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8975234031677246,
      "learning_rate": 1.9058960879284136e-05,
      "loss": 0.5525,
      "step": 5249
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7645169496536255,
      "learning_rate": 1.905852953543446e-05,
      "loss": 0.4853,
      "step": 5250
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.36053466796875,
      "learning_rate": 1.9058098097633345e-05,
      "loss": 0.9839,
      "step": 5251
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2788732051849365,
      "learning_rate": 1.9057666565885266e-05,
      "loss": 0.9298,
      "step": 5252
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.119354009628296,
      "learning_rate": 1.9057234940194704e-05,
      "loss": 0.5609,
      "step": 5253
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0271408557891846,
      "learning_rate": 1.9056803220566134e-05,
      "loss": 0.582,
      "step": 5254
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8365216255187988,
      "learning_rate": 1.9056371407004027e-05,
      "loss": 0.5713,
      "step": 5255
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8622217178344727,
      "learning_rate": 1.905593949951287e-05,
      "loss": 0.5162,
      "step": 5256
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1020469665527344,
      "learning_rate": 1.905550749809714e-05,
      "loss": 0.558,
      "step": 5257
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8352375030517578,
      "learning_rate": 1.9055075402761315e-05,
      "loss": 0.452,
      "step": 5258
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8285679817199707,
      "learning_rate": 1.905464321350988e-05,
      "loss": 0.5317,
      "step": 5259
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7533670663833618,
      "learning_rate": 1.9054210930347313e-05,
      "loss": 0.4593,
      "step": 5260
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9973080158233643,
      "learning_rate": 1.9053778553278103e-05,
      "loss": 0.5583,
      "step": 5261
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8692848682403564,
      "learning_rate": 1.905334608230673e-05,
      "loss": 0.4934,
      "step": 5262
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0926783084869385,
      "learning_rate": 1.905291351743768e-05,
      "loss": 0.5506,
      "step": 5263
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7543964385986328,
      "learning_rate": 1.9052480858675445e-05,
      "loss": 0.4488,
      "step": 5264
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.910869836807251,
      "learning_rate": 1.9052048106024504e-05,
      "loss": 0.5004,
      "step": 5265
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.799957036972046,
      "learning_rate": 1.905161525948935e-05,
      "loss": 0.4965,
      "step": 5266
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8134697675704956,
      "learning_rate": 1.9051182319074474e-05,
      "loss": 0.5332,
      "step": 5267
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0082855224609375,
      "learning_rate": 1.905074928478436e-05,
      "loss": 0.5702,
      "step": 5268
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9833191633224487,
      "learning_rate": 1.9050316156623504e-05,
      "loss": 0.5961,
      "step": 5269
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.82334303855896,
      "learning_rate": 1.9049882934596397e-05,
      "loss": 0.4804,
      "step": 5270
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8745505809783936,
      "learning_rate": 1.9049449618707534e-05,
      "loss": 0.5523,
      "step": 5271
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7967028617858887,
      "learning_rate": 1.9049016208961406e-05,
      "loss": 0.4578,
      "step": 5272
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7866617441177368,
      "learning_rate": 1.9048582705362513e-05,
      "loss": 0.5009,
      "step": 5273
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.818825364112854,
      "learning_rate": 1.9048149107915344e-05,
      "loss": 0.5036,
      "step": 5274
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9396294355392456,
      "learning_rate": 1.9047715416624402e-05,
      "loss": 0.5538,
      "step": 5275
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7516032457351685,
      "learning_rate": 1.904728163149418e-05,
      "loss": 0.4752,
      "step": 5276
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9112873077392578,
      "learning_rate": 1.9046847752529184e-05,
      "loss": 0.5269,
      "step": 5277
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7921806573867798,
      "learning_rate": 1.904641377973391e-05,
      "loss": 0.4408,
      "step": 5278
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8512028455734253,
      "learning_rate": 1.904597971311286e-05,
      "loss": 0.5588,
      "step": 5279
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8738458156585693,
      "learning_rate": 1.9045545552670532e-05,
      "loss": 0.5116,
      "step": 5280
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.753324031829834,
      "learning_rate": 1.904511129841143e-05,
      "loss": 0.5512,
      "step": 5281
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7493422031402588,
      "learning_rate": 1.9044676950340066e-05,
      "loss": 0.4438,
      "step": 5282
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7497177124023438,
      "learning_rate": 1.9044242508460933e-05,
      "loss": 0.467,
      "step": 5283
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8341072797775269,
      "learning_rate": 1.904380797277855e-05,
      "loss": 0.603,
      "step": 5284
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8510023355484009,
      "learning_rate": 1.9043373343297413e-05,
      "loss": 0.5467,
      "step": 5285
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6900169849395752,
      "learning_rate": 1.904293862002203e-05,
      "loss": 0.4663,
      "step": 5286
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0190505981445312,
      "learning_rate": 1.9042503802956918e-05,
      "loss": 0.638,
      "step": 5287
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7908011674880981,
      "learning_rate": 1.904206889210658e-05,
      "loss": 0.4674,
      "step": 5288
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8299779891967773,
      "learning_rate": 1.9041633887475526e-05,
      "loss": 0.5176,
      "step": 5289
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9304004907608032,
      "learning_rate": 1.9041198789068275e-05,
      "loss": 0.5102,
      "step": 5290
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8607463836669922,
      "learning_rate": 1.904076359688933e-05,
      "loss": 0.5525,
      "step": 5291
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7158050537109375,
      "learning_rate": 1.9040328310943214e-05,
      "loss": 0.4741,
      "step": 5292
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.85078763961792,
      "learning_rate": 1.9039892931234434e-05,
      "loss": 0.535,
      "step": 5293
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6845465898513794,
      "learning_rate": 1.903945745776751e-05,
      "loss": 0.4792,
      "step": 5294
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8033417463302612,
      "learning_rate": 1.9039021890546957e-05,
      "loss": 0.5495,
      "step": 5295
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8809726238250732,
      "learning_rate": 1.9038586229577295e-05,
      "loss": 0.5092,
      "step": 5296
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8889029026031494,
      "learning_rate": 1.9038150474863038e-05,
      "loss": 0.5786,
      "step": 5297
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7369914054870605,
      "learning_rate": 1.903771462640871e-05,
      "loss": 0.4414,
      "step": 5298
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8214420080184937,
      "learning_rate": 1.9037278684218827e-05,
      "loss": 0.5204,
      "step": 5299
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.846443772315979,
      "learning_rate": 1.9036842648297912e-05,
      "loss": 0.4706,
      "step": 5300
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8124425411224365,
      "learning_rate": 1.903640651865049e-05,
      "loss": 0.5075,
      "step": 5301
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9001927375793457,
      "learning_rate": 1.9035970295281083e-05,
      "loss": 0.5133,
      "step": 5302
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9436979293823242,
      "learning_rate": 1.9035533978194215e-05,
      "loss": 0.5188,
      "step": 5303
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8430864810943604,
      "learning_rate": 1.903509756739441e-05,
      "loss": 0.4929,
      "step": 5304
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8277939558029175,
      "learning_rate": 1.9034661062886192e-05,
      "loss": 0.6379,
      "step": 5305
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8462251424789429,
      "learning_rate": 1.90342244646741e-05,
      "loss": 0.4572,
      "step": 5306
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9476547241210938,
      "learning_rate": 1.9033787772762647e-05,
      "loss": 0.6051,
      "step": 5307
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.745269775390625,
      "learning_rate": 1.903335098715637e-05,
      "loss": 0.4718,
      "step": 5308
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.720687985420227,
      "learning_rate": 1.90329141078598e-05,
      "loss": 0.5091,
      "step": 5309
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8845903873443604,
      "learning_rate": 1.9032477134877462e-05,
      "loss": 0.519,
      "step": 5310
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9019548892974854,
      "learning_rate": 1.90320400682139e-05,
      "loss": 0.619,
      "step": 5311
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7290022373199463,
      "learning_rate": 1.9031602907873635e-05,
      "loss": 0.4678,
      "step": 5312
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.846156358718872,
      "learning_rate": 1.9031165653861204e-05,
      "loss": 0.5448,
      "step": 5313
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8264758586883545,
      "learning_rate": 1.9030728306181147e-05,
      "loss": 0.4486,
      "step": 5314
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7914295196533203,
      "learning_rate": 1.9030290864837996e-05,
      "loss": 0.5485,
      "step": 5315
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7879023551940918,
      "learning_rate": 1.9029853329836287e-05,
      "loss": 0.4504,
      "step": 5316
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9373886585235596,
      "learning_rate": 1.9029415701180557e-05,
      "loss": 0.6065,
      "step": 5317
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.835737943649292,
      "learning_rate": 1.902897797887535e-05,
      "loss": 0.4698,
      "step": 5318
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0307319164276123,
      "learning_rate": 1.9028540162925207e-05,
      "loss": 0.5534,
      "step": 5319
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.815391182899475,
      "learning_rate": 1.9028102253334663e-05,
      "loss": 0.4878,
      "step": 5320
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.87639582157135,
      "learning_rate": 1.902766425010826e-05,
      "loss": 0.5838,
      "step": 5321
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.762264370918274,
      "learning_rate": 1.9027226153250545e-05,
      "loss": 0.4372,
      "step": 5322
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8729740381240845,
      "learning_rate": 1.9026787962766057e-05,
      "loss": 0.5844,
      "step": 5323
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7926262617111206,
      "learning_rate": 1.9026349678659343e-05,
      "loss": 0.4446,
      "step": 5324
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7599588632583618,
      "learning_rate": 1.902591130093495e-05,
      "loss": 0.5112,
      "step": 5325
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.754658818244934,
      "learning_rate": 1.9025472829597427e-05,
      "loss": 0.4893,
      "step": 5326
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9285120964050293,
      "learning_rate": 1.9025034264651316e-05,
      "loss": 0.5776,
      "step": 5327
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.853811264038086,
      "learning_rate": 1.9024595606101165e-05,
      "loss": 0.5029,
      "step": 5328
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.78536856174469,
      "learning_rate": 1.902415685395153e-05,
      "loss": 0.5219,
      "step": 5329
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8321707248687744,
      "learning_rate": 1.9023718008206958e-05,
      "loss": 0.5189,
      "step": 5330
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7683147192001343,
      "learning_rate": 1.9023279068872e-05,
      "loss": 0.5521,
      "step": 5331
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9509499073028564,
      "learning_rate": 1.9022840035951205e-05,
      "loss": 0.4889,
      "step": 5332
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.847135305404663,
      "learning_rate": 1.9022400909449134e-05,
      "loss": 0.5718,
      "step": 5333
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.819588303565979,
      "learning_rate": 1.902196168937034e-05,
      "loss": 0.5143,
      "step": 5334
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7037826776504517,
      "learning_rate": 1.9021522375719374e-05,
      "loss": 0.5462,
      "step": 5335
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8068126440048218,
      "learning_rate": 1.9021082968500796e-05,
      "loss": 0.5265,
      "step": 5336
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.917428731918335,
      "learning_rate": 1.9020643467719162e-05,
      "loss": 0.578,
      "step": 5337
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7713509798049927,
      "learning_rate": 1.902020387337903e-05,
      "loss": 0.4532,
      "step": 5338
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8076045513153076,
      "learning_rate": 1.901976418548496e-05,
      "loss": 0.5433,
      "step": 5339
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7879873514175415,
      "learning_rate": 1.9019324404041514e-05,
      "loss": 0.5132,
      "step": 5340
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8228180408477783,
      "learning_rate": 1.9018884529053246e-05,
      "loss": 0.5363,
      "step": 5341
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8348029851913452,
      "learning_rate": 1.9018444560524728e-05,
      "loss": 0.4958,
      "step": 5342
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.741034984588623,
      "learning_rate": 1.901800449846052e-05,
      "loss": 0.9365,
      "step": 5343
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2575228214263916,
      "learning_rate": 1.9017564342865182e-05,
      "loss": 1.023,
      "step": 5344
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.890937089920044,
      "learning_rate": 1.901712409374328e-05,
      "loss": 0.5885,
      "step": 5345
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8439764976501465,
      "learning_rate": 1.9016683751099385e-05,
      "loss": 0.4794,
      "step": 5346
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9778884649276733,
      "learning_rate": 1.9016243314938058e-05,
      "loss": 0.5578,
      "step": 5347
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8171372413635254,
      "learning_rate": 1.9015802785263877e-05,
      "loss": 0.494,
      "step": 5348
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0271050930023193,
      "learning_rate": 1.9015362162081398e-05,
      "loss": 0.5979,
      "step": 5349
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9513421058654785,
      "learning_rate": 1.9014921445395196e-05,
      "loss": 0.4852,
      "step": 5350
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0165233612060547,
      "learning_rate": 1.901448063520985e-05,
      "loss": 0.5953,
      "step": 5351
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8278359174728394,
      "learning_rate": 1.901403973152992e-05,
      "loss": 0.4743,
      "step": 5352
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7628505229949951,
      "learning_rate": 1.9013598734359982e-05,
      "loss": 0.5047,
      "step": 5353
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7669590711593628,
      "learning_rate": 1.9013157643704615e-05,
      "loss": 0.5095,
      "step": 5354
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.807263731956482,
      "learning_rate": 1.901271645956839e-05,
      "loss": 0.5271,
      "step": 5355
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7931888103485107,
      "learning_rate": 1.9012275181955883e-05,
      "loss": 0.5239,
      "step": 5356
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8787678480148315,
      "learning_rate": 1.901183381087167e-05,
      "loss": 0.49,
      "step": 5357
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9765321016311646,
      "learning_rate": 1.901139234632033e-05,
      "loss": 0.555,
      "step": 5358
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.461015462875366,
      "learning_rate": 1.9010950788306443e-05,
      "loss": 0.937,
      "step": 5359
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.178568124771118,
      "learning_rate": 1.9010509136834586e-05,
      "loss": 0.8606,
      "step": 5360
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.162726879119873,
      "learning_rate": 1.901006739190934e-05,
      "loss": 0.566,
      "step": 5361
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0709502696990967,
      "learning_rate": 1.900962555353529e-05,
      "loss": 0.5088,
      "step": 5362
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9015120267868042,
      "learning_rate": 1.9009183621717014e-05,
      "loss": 0.566,
      "step": 5363
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8259286880493164,
      "learning_rate": 1.9008741596459096e-05,
      "loss": 0.481,
      "step": 5364
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9394372701644897,
      "learning_rate": 1.9008299477766125e-05,
      "loss": 0.5484,
      "step": 5365
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7337411642074585,
      "learning_rate": 1.900785726564268e-05,
      "loss": 0.4694,
      "step": 5366
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.896527647972107,
      "learning_rate": 1.9007414960093355e-05,
      "loss": 0.6011,
      "step": 5367
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7669543027877808,
      "learning_rate": 1.900697256112273e-05,
      "loss": 0.5026,
      "step": 5368
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7855405807495117,
      "learning_rate": 1.90065300687354e-05,
      "loss": 0.5301,
      "step": 5369
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7315629720687866,
      "learning_rate": 1.9006087482935945e-05,
      "loss": 0.4835,
      "step": 5370
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.846113681793213,
      "learning_rate": 1.9005644803728967e-05,
      "loss": 0.5806,
      "step": 5371
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7526298761367798,
      "learning_rate": 1.900520203111905e-05,
      "loss": 0.5064,
      "step": 5372
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7951472997665405,
      "learning_rate": 1.9004759165110787e-05,
      "loss": 0.5177,
      "step": 5373
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7948733568191528,
      "learning_rate": 1.9004316205708772e-05,
      "loss": 0.4838,
      "step": 5374
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7367563247680664,
      "learning_rate": 1.9003873152917598e-05,
      "loss": 0.5433,
      "step": 5375
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8501121997833252,
      "learning_rate": 1.9003430006741864e-05,
      "loss": 0.4891,
      "step": 5376
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8203904628753662,
      "learning_rate": 1.9002986767186163e-05,
      "loss": 0.5419,
      "step": 5377
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7617275714874268,
      "learning_rate": 1.9002543434255094e-05,
      "loss": 0.4727,
      "step": 5378
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.073113203048706,
      "learning_rate": 1.9002100007953252e-05,
      "loss": 0.6406,
      "step": 5379
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8011564016342163,
      "learning_rate": 1.900165648828524e-05,
      "loss": 0.4484,
      "step": 5380
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8556253910064697,
      "learning_rate": 1.900121287525565e-05,
      "loss": 0.5114,
      "step": 5381
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.816510796546936,
      "learning_rate": 1.9000769168869094e-05,
      "loss": 0.4677,
      "step": 5382
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.769787311553955,
      "learning_rate": 1.900032536913017e-05,
      "loss": 0.5441,
      "step": 5383
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8612618446350098,
      "learning_rate": 1.8999881476043476e-05,
      "loss": 0.4717,
      "step": 5384
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7557566165924072,
      "learning_rate": 1.8999437489613623e-05,
      "loss": 0.5061,
      "step": 5385
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.801806092262268,
      "learning_rate": 1.899899340984521e-05,
      "loss": 0.4474,
      "step": 5386
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9259016513824463,
      "learning_rate": 1.8998549236742844e-05,
      "loss": 0.5457,
      "step": 5387
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7443994283676147,
      "learning_rate": 1.8998104970311137e-05,
      "loss": 0.4463,
      "step": 5388
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9280402660369873,
      "learning_rate": 1.8997660610554687e-05,
      "loss": 0.5557,
      "step": 5389
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7639799118041992,
      "learning_rate": 1.8997216157478115e-05,
      "loss": 0.4414,
      "step": 5390
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8378231525421143,
      "learning_rate": 1.899677161108602e-05,
      "loss": 0.516,
      "step": 5391
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8350316286087036,
      "learning_rate": 1.899632697138302e-05,
      "loss": 0.4706,
      "step": 5392
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8187662363052368,
      "learning_rate": 1.899588223837372e-05,
      "loss": 0.5021,
      "step": 5393
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.940742015838623,
      "learning_rate": 1.8995437412062737e-05,
      "loss": 0.4358,
      "step": 5394
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9649662971496582,
      "learning_rate": 1.8994992492454684e-05,
      "loss": 0.5425,
      "step": 5395
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7678711414337158,
      "learning_rate": 1.8994547479554176e-05,
      "loss": 0.4654,
      "step": 5396
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0034191608428955,
      "learning_rate": 1.8994102373365825e-05,
      "loss": 0.6175,
      "step": 5397
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7555220127105713,
      "learning_rate": 1.8993657173894255e-05,
      "loss": 0.4905,
      "step": 5398
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.796033501625061,
      "learning_rate": 1.8993211881144075e-05,
      "loss": 0.4943,
      "step": 5399
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7526973485946655,
      "learning_rate": 1.8992766495119905e-05,
      "loss": 0.5244,
      "step": 5400
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7706276178359985,
      "learning_rate": 1.899232101582637e-05,
      "loss": 0.5481,
      "step": 5401
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7599252462387085,
      "learning_rate": 1.899187544326808e-05,
      "loss": 0.4583,
      "step": 5402
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.798761248588562,
      "learning_rate": 1.8991429777449674e-05,
      "loss": 0.5431,
      "step": 5403
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8037124872207642,
      "learning_rate": 1.8990984018375754e-05,
      "loss": 0.4958,
      "step": 5404
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7306959629058838,
      "learning_rate": 1.8990538166050954e-05,
      "loss": 0.491,
      "step": 5405
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.777060866355896,
      "learning_rate": 1.8990092220479902e-05,
      "loss": 0.4672,
      "step": 5406
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9298779964447021,
      "learning_rate": 1.8989646181667213e-05,
      "loss": 0.5327,
      "step": 5407
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8773494958877563,
      "learning_rate": 1.898920004961752e-05,
      "loss": 0.467,
      "step": 5408
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7999709844589233,
      "learning_rate": 1.8988753824335444e-05,
      "loss": 0.5189,
      "step": 5409
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9570974111557007,
      "learning_rate": 1.898830750582562e-05,
      "loss": 0.5305,
      "step": 5410
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.88591468334198,
      "learning_rate": 1.8987861094092673e-05,
      "loss": 0.5059,
      "step": 5411
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8380614519119263,
      "learning_rate": 1.898741458914124e-05,
      "loss": 0.5274,
      "step": 5412
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.769387125968933,
      "learning_rate": 1.898696799097594e-05,
      "loss": 0.5419,
      "step": 5413
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7941229343414307,
      "learning_rate": 1.898652129960141e-05,
      "loss": 0.4471,
      "step": 5414
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.994460105895996,
      "learning_rate": 1.8986074515022286e-05,
      "loss": 0.6176,
      "step": 5415
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.793868064880371,
      "learning_rate": 1.8985627637243202e-05,
      "loss": 0.5143,
      "step": 5416
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8839784860610962,
      "learning_rate": 1.898518066626879e-05,
      "loss": 0.5697,
      "step": 5417
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.72696852684021,
      "learning_rate": 1.8984733602103686e-05,
      "loss": 0.4664,
      "step": 5418
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7423962354660034,
      "learning_rate": 1.8984286444752524e-05,
      "loss": 0.4993,
      "step": 5419
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8517507314682007,
      "learning_rate": 1.898383919421995e-05,
      "loss": 0.523,
      "step": 5420
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.84169602394104,
      "learning_rate": 1.8983391850510594e-05,
      "loss": 0.5448,
      "step": 5421
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7875175476074219,
      "learning_rate": 1.8982944413629103e-05,
      "loss": 0.4738,
      "step": 5422
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9431782960891724,
      "learning_rate": 1.898249688358011e-05,
      "loss": 0.652,
      "step": 5423
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.832029938697815,
      "learning_rate": 1.898204926036826e-05,
      "loss": 0.5105,
      "step": 5424
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8432598114013672,
      "learning_rate": 1.89816015439982e-05,
      "loss": 0.5478,
      "step": 5425
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7598662376403809,
      "learning_rate": 1.898115373447457e-05,
      "loss": 0.4773,
      "step": 5426
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.852590560913086,
      "learning_rate": 1.8980705831802007e-05,
      "loss": 0.8585,
      "step": 5427
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.4121623039245605,
      "learning_rate": 1.8980257835985172e-05,
      "loss": 0.9476,
      "step": 5428
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.099799156188965,
      "learning_rate": 1.8979809747028702e-05,
      "loss": 0.5978,
      "step": 5429
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.798836350440979,
      "learning_rate": 1.8979361564937243e-05,
      "loss": 0.4601,
      "step": 5430
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0995676517486572,
      "learning_rate": 1.8978913289715444e-05,
      "loss": 0.6294,
      "step": 5431
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7681593894958496,
      "learning_rate": 1.897846492136796e-05,
      "loss": 0.4955,
      "step": 5432
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9559248685836792,
      "learning_rate": 1.8978016459899436e-05,
      "loss": 0.5489,
      "step": 5433
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9483894109725952,
      "learning_rate": 1.8977567905314522e-05,
      "loss": 0.5889,
      "step": 5434
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7986621856689453,
      "learning_rate": 1.8977119257617878e-05,
      "loss": 0.5102,
      "step": 5435
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9199011325836182,
      "learning_rate": 1.897667051681415e-05,
      "loss": 0.4638,
      "step": 5436
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7326194047927856,
      "learning_rate": 1.8976221682907995e-05,
      "loss": 0.5273,
      "step": 5437
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7910034656524658,
      "learning_rate": 1.8975772755904066e-05,
      "loss": 0.4959,
      "step": 5438
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.792101502418518,
      "learning_rate": 1.8975323735807022e-05,
      "loss": 0.5536,
      "step": 5439
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9094315767288208,
      "learning_rate": 1.8974874622621515e-05,
      "loss": 0.4772,
      "step": 5440
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9206104278564453,
      "learning_rate": 1.897442541635221e-05,
      "loss": 0.6036,
      "step": 5441
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8141720294952393,
      "learning_rate": 1.897397611700376e-05,
      "loss": 0.4474,
      "step": 5442
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.8008315563201904,
      "learning_rate": 1.897352672458083e-05,
      "loss": 0.8432,
      "step": 5443
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.631359815597534,
      "learning_rate": 1.8973077239088077e-05,
      "loss": 0.9492,
      "step": 5444
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.96185302734375,
      "learning_rate": 1.8972627660530166e-05,
      "loss": 0.5101,
      "step": 5445
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.026352643966675,
      "learning_rate": 1.897217798891176e-05,
      "loss": 0.5431,
      "step": 5446
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.964097261428833,
      "learning_rate": 1.897172822423752e-05,
      "loss": 0.6411,
      "step": 5447
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.962024450302124,
      "learning_rate": 1.8971278366512107e-05,
      "loss": 0.5319,
      "step": 5448
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8540163040161133,
      "learning_rate": 1.8970828415740194e-05,
      "loss": 0.5421,
      "step": 5449
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8946012258529663,
      "learning_rate": 1.8970378371926448e-05,
      "loss": 0.5341,
      "step": 5450
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9575749635696411,
      "learning_rate": 1.896992823507553e-05,
      "loss": 0.533,
      "step": 5451
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7591440677642822,
      "learning_rate": 1.8969478005192114e-05,
      "loss": 0.4479,
      "step": 5452
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.831624984741211,
      "learning_rate": 1.896902768228087e-05,
      "loss": 0.5349,
      "step": 5453
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8987380266189575,
      "learning_rate": 1.8968577266346465e-05,
      "loss": 0.4996,
      "step": 5454
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9033598899841309,
      "learning_rate": 1.896812675739357e-05,
      "loss": 0.5359,
      "step": 5455
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7841345071792603,
      "learning_rate": 1.8967676155426865e-05,
      "loss": 0.4632,
      "step": 5456
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7172108888626099,
      "learning_rate": 1.8967225460451015e-05,
      "loss": 0.5012,
      "step": 5457
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7366242408752441,
      "learning_rate": 1.89667746724707e-05,
      "loss": 0.4973,
      "step": 5458
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9135327339172363,
      "learning_rate": 1.896632379149059e-05,
      "loss": 0.5869,
      "step": 5459
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8093104362487793,
      "learning_rate": 1.8965872817515368e-05,
      "loss": 0.4997,
      "step": 5460
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.85829758644104,
      "learning_rate": 1.8965421750549705e-05,
      "loss": 0.5662,
      "step": 5461
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6714056730270386,
      "learning_rate": 1.896497059059828e-05,
      "loss": 0.4766,
      "step": 5462
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8096470832824707,
      "learning_rate": 1.8964519337665778e-05,
      "loss": 0.5491,
      "step": 5463
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.653221845626831,
      "learning_rate": 1.8964067991756875e-05,
      "loss": 0.4393,
      "step": 5464
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8904497623443604,
      "learning_rate": 1.896361655287625e-05,
      "loss": 0.5612,
      "step": 5465
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8149495124816895,
      "learning_rate": 1.896316502102859e-05,
      "loss": 0.4544,
      "step": 5466
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.822593331336975,
      "learning_rate": 1.8962713396218576e-05,
      "loss": 0.5777,
      "step": 5467
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8671138286590576,
      "learning_rate": 1.896226167845089e-05,
      "loss": 0.5297,
      "step": 5468
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7391144037246704,
      "learning_rate": 1.8961809867730224e-05,
      "loss": 0.5617,
      "step": 5469
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.747308611869812,
      "learning_rate": 1.8961357964061252e-05,
      "loss": 0.4484,
      "step": 5470
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.788051962852478,
      "learning_rate": 1.8960905967448673e-05,
      "loss": 0.4828,
      "step": 5471
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8868099451065063,
      "learning_rate": 1.8960453877897168e-05,
      "loss": 0.5351,
      "step": 5472
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7670880556106567,
      "learning_rate": 1.8960001695411428e-05,
      "loss": 0.5567,
      "step": 5473
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.692555546760559,
      "learning_rate": 1.8959549419996145e-05,
      "loss": 0.4368,
      "step": 5474
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8034716844558716,
      "learning_rate": 1.8959097051656004e-05,
      "loss": 0.517,
      "step": 5475
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7937219142913818,
      "learning_rate": 1.8958644590395702e-05,
      "loss": 0.5075,
      "step": 5476
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9523885250091553,
      "learning_rate": 1.8958192036219927e-05,
      "loss": 0.5859,
      "step": 5477
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7516626119613647,
      "learning_rate": 1.895773938913338e-05,
      "loss": 0.488,
      "step": 5478
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7312195301055908,
      "learning_rate": 1.8957286649140752e-05,
      "loss": 0.5711,
      "step": 5479
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8196505308151245,
      "learning_rate": 1.8956833816246733e-05,
      "loss": 0.4769,
      "step": 5480
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.924721598625183,
      "learning_rate": 1.895638089045603e-05,
      "loss": 0.5907,
      "step": 5481
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.675217628479004,
      "learning_rate": 1.895592787177333e-05,
      "loss": 0.4706,
      "step": 5482
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6972992420196533,
      "learning_rate": 1.895547476020334e-05,
      "loss": 0.5207,
      "step": 5483
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.882028341293335,
      "learning_rate": 1.8955021555750756e-05,
      "loss": 0.4983,
      "step": 5484
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.673065185546875,
      "learning_rate": 1.8954568258420274e-05,
      "loss": 0.8708,
      "step": 5485
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.4189436435699463,
      "learning_rate": 1.8954114868216608e-05,
      "loss": 0.9472,
      "step": 5486
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.019256591796875,
      "learning_rate": 1.895366138514445e-05,
      "loss": 0.5879,
      "step": 5487
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0202977657318115,
      "learning_rate": 1.8953207809208504e-05,
      "loss": 0.4959,
      "step": 5488
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8225958347320557,
      "learning_rate": 1.8952754140413475e-05,
      "loss": 0.4917,
      "step": 5489
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7523071765899658,
      "learning_rate": 1.8952300378764072e-05,
      "loss": 0.4946,
      "step": 5490
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8585014343261719,
      "learning_rate": 1.8951846524264998e-05,
      "loss": 0.5373,
      "step": 5491
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9114974737167358,
      "learning_rate": 1.8951392576920958e-05,
      "loss": 0.4755,
      "step": 5492
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9159607887268066,
      "learning_rate": 1.8950938536736666e-05,
      "loss": 0.5289,
      "step": 5493
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.924088716506958,
      "learning_rate": 1.8950484403716828e-05,
      "loss": 0.5673,
      "step": 5494
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7886227369308472,
      "learning_rate": 1.8950030177866153e-05,
      "loss": 0.5093,
      "step": 5495
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7307084798812866,
      "learning_rate": 1.8949575859189355e-05,
      "loss": 0.4988,
      "step": 5496
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9833890199661255,
      "learning_rate": 1.8949121447691145e-05,
      "loss": 0.5997,
      "step": 5497
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8601337671279907,
      "learning_rate": 1.894866694337623e-05,
      "loss": 0.4397,
      "step": 5498
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8646923303604126,
      "learning_rate": 1.8948212346249334e-05,
      "loss": 0.5538,
      "step": 5499
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6676982641220093,
      "learning_rate": 1.8947757656315166e-05,
      "loss": 0.4491,
      "step": 5500
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.6523382663726807,
      "learning_rate": 1.894730287357844e-05,
      "loss": 0.9234,
      "step": 5501
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.317460775375366,
      "learning_rate": 1.894684799804388e-05,
      "loss": 0.9536,
      "step": 5502
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.08953595161438,
      "learning_rate": 1.8946393029716196e-05,
      "loss": 0.5935,
      "step": 5503
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0056283473968506,
      "learning_rate": 1.8945937968600114e-05,
      "loss": 0.4983,
      "step": 5504
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.878981113433838,
      "learning_rate": 1.894548281470035e-05,
      "loss": 0.5257,
      "step": 5505
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8068526983261108,
      "learning_rate": 1.8945027568021622e-05,
      "loss": 0.4774,
      "step": 5506
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9089103937149048,
      "learning_rate": 1.894457222856865e-05,
      "loss": 0.5566,
      "step": 5507
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9080772399902344,
      "learning_rate": 1.894411679634617e-05,
      "loss": 0.4664,
      "step": 5508
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.161170721054077,
      "learning_rate": 1.894366127135889e-05,
      "loss": 0.5985,
      "step": 5509
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8284415006637573,
      "learning_rate": 1.8943205653611546e-05,
      "loss": 0.4958,
      "step": 5510
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9806735515594482,
      "learning_rate": 1.8942749943108857e-05,
      "loss": 0.6308,
      "step": 5511
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8029946088790894,
      "learning_rate": 1.8942294139855545e-05,
      "loss": 0.49,
      "step": 5512
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9463177919387817,
      "learning_rate": 1.894183824385635e-05,
      "loss": 0.5738,
      "step": 5513
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.064117193222046,
      "learning_rate": 1.8941382255115992e-05,
      "loss": 0.4635,
      "step": 5514
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8194103240966797,
      "learning_rate": 1.89409261736392e-05,
      "loss": 0.5451,
      "step": 5515
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.762105941772461,
      "learning_rate": 1.894046999943071e-05,
      "loss": 0.5142,
      "step": 5516
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8315598964691162,
      "learning_rate": 1.8940013732495245e-05,
      "loss": 0.5487,
      "step": 5517
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8168435096740723,
      "learning_rate": 1.8939557372837547e-05,
      "loss": 0.4722,
      "step": 5518
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8474984169006348,
      "learning_rate": 1.8939100920462338e-05,
      "loss": 0.5596,
      "step": 5519
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.74087393283844,
      "learning_rate": 1.8938644375374366e-05,
      "loss": 0.4731,
      "step": 5520
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7138755321502686,
      "learning_rate": 1.893818773757835e-05,
      "loss": 0.4598,
      "step": 5521
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.351989269256592,
      "learning_rate": 1.893773100707904e-05,
      "loss": 0.5281,
      "step": 5522
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7713463306427002,
      "learning_rate": 1.8937274183881164e-05,
      "loss": 0.5089,
      "step": 5523
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8117170333862305,
      "learning_rate": 1.8936817267989466e-05,
      "loss": 0.4907,
      "step": 5524
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8372502326965332,
      "learning_rate": 1.893636025940868e-05,
      "loss": 0.5442,
      "step": 5525
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0367727279663086,
      "learning_rate": 1.893590315814355e-05,
      "loss": 0.5424,
      "step": 5526
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9583845138549805,
      "learning_rate": 1.893544596419881e-05,
      "loss": 0.6041,
      "step": 5527
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.902868390083313,
      "learning_rate": 1.8934988677579213e-05,
      "loss": 0.4976,
      "step": 5528
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8865087032318115,
      "learning_rate": 1.8934531298289494e-05,
      "loss": 0.523,
      "step": 5529
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7102999687194824,
      "learning_rate": 1.8934073826334394e-05,
      "loss": 0.4455,
      "step": 5530
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8886555433273315,
      "learning_rate": 1.893361626171867e-05,
      "loss": 0.5027,
      "step": 5531
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9328639507293701,
      "learning_rate": 1.8933158604447053e-05,
      "loss": 0.5681,
      "step": 5532
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8003439903259277,
      "learning_rate": 1.89327008545243e-05,
      "loss": 0.5231,
      "step": 5533
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8514288663864136,
      "learning_rate": 1.8932243011955154e-05,
      "loss": 0.4728,
      "step": 5534
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9081106185913086,
      "learning_rate": 1.8931785076744366e-05,
      "loss": 0.5572,
      "step": 5535
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7812620401382446,
      "learning_rate": 1.893132704889668e-05,
      "loss": 0.4793,
      "step": 5536
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8458712100982666,
      "learning_rate": 1.8930868928416853e-05,
      "loss": 0.5524,
      "step": 5537
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8932851552963257,
      "learning_rate": 1.893041071530964e-05,
      "loss": 0.5219,
      "step": 5538
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7519354820251465,
      "learning_rate": 1.892995240957978e-05,
      "loss": 0.5418,
      "step": 5539
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8217720985412598,
      "learning_rate": 1.8929494011232036e-05,
      "loss": 0.4841,
      "step": 5540
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8864948749542236,
      "learning_rate": 1.8929035520271157e-05,
      "loss": 0.5637,
      "step": 5541
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.789561152458191,
      "learning_rate": 1.8928576936701906e-05,
      "loss": 0.4899,
      "step": 5542
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.97007155418396,
      "learning_rate": 1.892811826052903e-05,
      "loss": 0.5397,
      "step": 5543
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8492883443832397,
      "learning_rate": 1.8927659491757296e-05,
      "loss": 0.5094,
      "step": 5544
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.801473617553711,
      "learning_rate": 1.8927200630391454e-05,
      "loss": 0.5045,
      "step": 5545
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9198447465896606,
      "learning_rate": 1.8926741676436268e-05,
      "loss": 0.5142,
      "step": 5546
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7163532972335815,
      "learning_rate": 1.8926282629896495e-05,
      "loss": 0.4752,
      "step": 5547
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7349954843521118,
      "learning_rate": 1.8925823490776894e-05,
      "loss": 0.4861,
      "step": 5548
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9389241933822632,
      "learning_rate": 1.8925364259082237e-05,
      "loss": 0.5787,
      "step": 5549
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.722149133682251,
      "learning_rate": 1.8924904934817277e-05,
      "loss": 0.4693,
      "step": 5550
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.836719274520874,
      "learning_rate": 1.892444551798678e-05,
      "loss": 0.5151,
      "step": 5551
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8165981769561768,
      "learning_rate": 1.8923986008595513e-05,
      "loss": 0.5279,
      "step": 5552
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7545201778411865,
      "learning_rate": 1.8923526406648243e-05,
      "loss": 0.5235,
      "step": 5553
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9641602039337158,
      "learning_rate": 1.8923066712149732e-05,
      "loss": 0.5077,
      "step": 5554
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8279021978378296,
      "learning_rate": 1.892260692510475e-05,
      "loss": 0.5683,
      "step": 5555
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6737277507781982,
      "learning_rate": 1.8922147045518068e-05,
      "loss": 0.4419,
      "step": 5556
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7493590116500854,
      "learning_rate": 1.8921687073394453e-05,
      "loss": 0.5498,
      "step": 5557
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.881885051727295,
      "learning_rate": 1.8921227008738678e-05,
      "loss": 0.4917,
      "step": 5558
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8676115274429321,
      "learning_rate": 1.892076685155551e-05,
      "loss": 0.5775,
      "step": 5559
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8126895427703857,
      "learning_rate": 1.892030660184973e-05,
      "loss": 0.455,
      "step": 5560
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.119567632675171,
      "learning_rate": 1.8919846259626102e-05,
      "loss": 0.5737,
      "step": 5561
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7429351806640625,
      "learning_rate": 1.8919385824889405e-05,
      "loss": 0.4616,
      "step": 5562
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9960626363754272,
      "learning_rate": 1.8918925297644418e-05,
      "loss": 0.6068,
      "step": 5563
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.724228858947754,
      "learning_rate": 1.8918464677895907e-05,
      "loss": 0.4495,
      "step": 5564
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8209513425827026,
      "learning_rate": 1.891800396564866e-05,
      "loss": 0.5209,
      "step": 5565
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8447126150131226,
      "learning_rate": 1.891754316090745e-05,
      "loss": 0.4977,
      "step": 5566
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8989044427871704,
      "learning_rate": 1.891708226367706e-05,
      "loss": 0.5777,
      "step": 5567
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7484537363052368,
      "learning_rate": 1.8916621273962265e-05,
      "loss": 0.4597,
      "step": 5568
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7642675638198853,
      "learning_rate": 1.891616019176785e-05,
      "loss": 0.519,
      "step": 5569
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.742773175239563,
      "learning_rate": 1.8915699017098596e-05,
      "loss": 0.4557,
      "step": 5570
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.80543851852417,
      "learning_rate": 1.8915237749959288e-05,
      "loss": 0.4964,
      "step": 5571
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.80709969997406,
      "learning_rate": 1.8914776390354705e-05,
      "loss": 0.4881,
      "step": 5572
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8069430589675903,
      "learning_rate": 1.891431493828964e-05,
      "loss": 0.5455,
      "step": 5573
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7210248708724976,
      "learning_rate": 1.891385339376887e-05,
      "loss": 0.4495,
      "step": 5574
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.941636085510254,
      "learning_rate": 1.891339175679719e-05,
      "loss": 0.5815,
      "step": 5575
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8038582801818848,
      "learning_rate": 1.8912930027379383e-05,
      "loss": 0.4658,
      "step": 5576
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7559332847595215,
      "learning_rate": 1.8912468205520237e-05,
      "loss": 0.5285,
      "step": 5577
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8300505876541138,
      "learning_rate": 1.891200629122454e-05,
      "loss": 0.5107,
      "step": 5578
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8646862506866455,
      "learning_rate": 1.8911544284497093e-05,
      "loss": 0.5467,
      "step": 5579
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6807893514633179,
      "learning_rate": 1.891108218534268e-05,
      "loss": 0.4494,
      "step": 5580
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8649183511734009,
      "learning_rate": 1.8910619993766098e-05,
      "loss": 0.5468,
      "step": 5581
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.772172212600708,
      "learning_rate": 1.8910157709772135e-05,
      "loss": 0.4841,
      "step": 5582
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8628849983215332,
      "learning_rate": 1.890969533336559e-05,
      "loss": 0.5158,
      "step": 5583
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8127068281173706,
      "learning_rate": 1.8909232864551258e-05,
      "loss": 0.4978,
      "step": 5584
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.876526951789856,
      "learning_rate": 1.890877030333393e-05,
      "loss": 0.5812,
      "step": 5585
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7679328918457031,
      "learning_rate": 1.890830764971841e-05,
      "loss": 0.468,
      "step": 5586
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8891290426254272,
      "learning_rate": 1.8907844903709496e-05,
      "loss": 0.5329,
      "step": 5587
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9786304235458374,
      "learning_rate": 1.890738206531199e-05,
      "loss": 0.4895,
      "step": 5588
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.866564154624939,
      "learning_rate": 1.8906919134530678e-05,
      "loss": 0.5101,
      "step": 5589
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6261546611785889,
      "learning_rate": 1.8906456111370378e-05,
      "loss": 0.4429,
      "step": 5590
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7875738143920898,
      "learning_rate": 1.8905992995835887e-05,
      "loss": 0.5388,
      "step": 5591
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.864341378211975,
      "learning_rate": 1.8905529787932006e-05,
      "loss": 0.4865,
      "step": 5592
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0416669845581055,
      "learning_rate": 1.890506648766354e-05,
      "loss": 0.6131,
      "step": 5593
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7364989519119263,
      "learning_rate": 1.8904603095035294e-05,
      "loss": 0.437,
      "step": 5594
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9439276456832886,
      "learning_rate": 1.8904139610052076e-05,
      "loss": 0.604,
      "step": 5595
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7664743661880493,
      "learning_rate": 1.8903676032718688e-05,
      "loss": 0.4651,
      "step": 5596
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8590630292892456,
      "learning_rate": 1.890321236303995e-05,
      "loss": 0.5166,
      "step": 5597
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8795560598373413,
      "learning_rate": 1.8902748601020656e-05,
      "loss": 0.5051,
      "step": 5598
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7840722799301147,
      "learning_rate": 1.8902284746665625e-05,
      "loss": 0.5605,
      "step": 5599
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7387042045593262,
      "learning_rate": 1.8901820799979665e-05,
      "loss": 0.4595,
      "step": 5600
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9014461040496826,
      "learning_rate": 1.8901356760967588e-05,
      "loss": 0.585,
      "step": 5601
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7604143619537354,
      "learning_rate": 1.8900892629634208e-05,
      "loss": 0.4918,
      "step": 5602
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8149094581604004,
      "learning_rate": 1.8900428405984338e-05,
      "loss": 0.5315,
      "step": 5603
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7622712850570679,
      "learning_rate": 1.8899964090022794e-05,
      "loss": 0.5428,
      "step": 5604
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.812845230102539,
      "learning_rate": 1.889949968175439e-05,
      "loss": 0.5352,
      "step": 5605
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8662569522857666,
      "learning_rate": 1.8899035181183944e-05,
      "loss": 0.5068,
      "step": 5606
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7920535802841187,
      "learning_rate": 1.889857058831627e-05,
      "loss": 0.5154,
      "step": 5607
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.783766269683838,
      "learning_rate": 1.8898105903156193e-05,
      "loss": 0.5472,
      "step": 5608
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8459839820861816,
      "learning_rate": 1.8897641125708528e-05,
      "loss": 0.5311,
      "step": 5609
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7992980480194092,
      "learning_rate": 1.8897176255978097e-05,
      "loss": 0.5216,
      "step": 5610
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8898111581802368,
      "learning_rate": 1.8896711293969718e-05,
      "loss": 0.5601,
      "step": 5611
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.89938485622406,
      "learning_rate": 1.889624623968822e-05,
      "loss": 0.4743,
      "step": 5612
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8493279218673706,
      "learning_rate": 1.8895781093138422e-05,
      "loss": 0.4906,
      "step": 5613
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9563671350479126,
      "learning_rate": 1.8895315854325148e-05,
      "loss": 0.5023,
      "step": 5614
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7824805974960327,
      "learning_rate": 1.889485052325322e-05,
      "loss": 0.5484,
      "step": 5615
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.849058985710144,
      "learning_rate": 1.8894385099927478e-05,
      "loss": 0.4964,
      "step": 5616
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.883791208267212,
      "learning_rate": 1.8893919584352734e-05,
      "loss": 0.5629,
      "step": 5617
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.058209180831909,
      "learning_rate": 1.889345397653382e-05,
      "loss": 0.534,
      "step": 5618
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9735742807388306,
      "learning_rate": 1.889298827647557e-05,
      "loss": 0.5807,
      "step": 5619
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8418422937393188,
      "learning_rate": 1.8892522484182812e-05,
      "loss": 0.5372,
      "step": 5620
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8140307664871216,
      "learning_rate": 1.889205659966037e-05,
      "loss": 0.5135,
      "step": 5621
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7665488719940186,
      "learning_rate": 1.889159062291309e-05,
      "loss": 0.498,
      "step": 5622
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.683244228363037,
      "learning_rate": 1.889112455394579e-05,
      "loss": 0.4674,
      "step": 5623
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8415478467941284,
      "learning_rate": 1.8890658392763313e-05,
      "loss": 0.5258,
      "step": 5624
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.807438850402832,
      "learning_rate": 1.8890192139370495e-05,
      "loss": 0.5765,
      "step": 5625
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6315252780914307,
      "learning_rate": 1.8889725793772164e-05,
      "loss": 0.4508,
      "step": 5626
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.742017149925232,
      "learning_rate": 1.8889259355973165e-05,
      "loss": 0.486,
      "step": 5627
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7204158306121826,
      "learning_rate": 1.888879282597833e-05,
      "loss": 0.4813,
      "step": 5628
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9268581867218018,
      "learning_rate": 1.8888326203792496e-05,
      "loss": 0.6228,
      "step": 5629
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6944938898086548,
      "learning_rate": 1.888785948942051e-05,
      "loss": 0.4315,
      "step": 5630
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8414212465286255,
      "learning_rate": 1.8887392682867208e-05,
      "loss": 0.5653,
      "step": 5631
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8140182495117188,
      "learning_rate": 1.8886925784137433e-05,
      "loss": 0.5107,
      "step": 5632
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7374037504196167,
      "learning_rate": 1.8886458793236026e-05,
      "loss": 0.5357,
      "step": 5633
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6720271110534668,
      "learning_rate": 1.888599171016783e-05,
      "loss": 0.4401,
      "step": 5634
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.726636290550232,
      "learning_rate": 1.888552453493769e-05,
      "loss": 0.5632,
      "step": 5635
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.779981255531311,
      "learning_rate": 1.8885057267550456e-05,
      "loss": 0.4836,
      "step": 5636
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8351110219955444,
      "learning_rate": 1.888458990801097e-05,
      "loss": 0.5305,
      "step": 5637
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7529523372650146,
      "learning_rate": 1.8884122456324073e-05,
      "loss": 0.4655,
      "step": 5638
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8651154041290283,
      "learning_rate": 1.8883654912494626e-05,
      "loss": 0.5774,
      "step": 5639
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9179186820983887,
      "learning_rate": 1.888318727652747e-05,
      "loss": 0.5185,
      "step": 5640
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.104360580444336,
      "learning_rate": 1.8882719548427457e-05,
      "loss": 0.5789,
      "step": 5641
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.848496437072754,
      "learning_rate": 1.8882251728199435e-05,
      "loss": 0.4871,
      "step": 5642
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.949234962463379,
      "learning_rate": 1.8881783815848263e-05,
      "loss": 0.5701,
      "step": 5643
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7816693782806396,
      "learning_rate": 1.888131581137879e-05,
      "loss": 0.4895,
      "step": 5644
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7944644689559937,
      "learning_rate": 1.8880847714795867e-05,
      "loss": 0.5462,
      "step": 5645
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8154696226119995,
      "learning_rate": 1.8880379526104356e-05,
      "loss": 0.5203,
      "step": 5646
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.829471230506897,
      "learning_rate": 1.8879911245309106e-05,
      "loss": 0.5615,
      "step": 5647
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8019007444381714,
      "learning_rate": 1.887944287241498e-05,
      "loss": 0.4734,
      "step": 5648
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7765324115753174,
      "learning_rate": 1.8878974407426828e-05,
      "loss": 0.5342,
      "step": 5649
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8196488618850708,
      "learning_rate": 1.8878505850349514e-05,
      "loss": 0.5395,
      "step": 5650
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7746634483337402,
      "learning_rate": 1.8878037201187898e-05,
      "loss": 0.564,
      "step": 5651
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7592978477478027,
      "learning_rate": 1.8877568459946842e-05,
      "loss": 0.4578,
      "step": 5652
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8902723789215088,
      "learning_rate": 1.88770996266312e-05,
      "loss": 0.6923,
      "step": 5653
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.880585789680481,
      "learning_rate": 1.8876630701245845e-05,
      "loss": 0.4651,
      "step": 5654
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7891125679016113,
      "learning_rate": 1.887616168379563e-05,
      "loss": 0.5239,
      "step": 5655
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.777953028678894,
      "learning_rate": 1.887569257428543e-05,
      "loss": 0.5155,
      "step": 5656
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0116384029388428,
      "learning_rate": 1.8875223372720102e-05,
      "loss": 0.5818,
      "step": 5657
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7893669605255127,
      "learning_rate": 1.8874754079104517e-05,
      "loss": 0.4527,
      "step": 5658
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8577675819396973,
      "learning_rate": 1.8874284693443535e-05,
      "loss": 0.5476,
      "step": 5659
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7285716533660889,
      "learning_rate": 1.8873815215742036e-05,
      "loss": 0.4499,
      "step": 5660
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9170758724212646,
      "learning_rate": 1.8873345646004884e-05,
      "loss": 0.5624,
      "step": 5661
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.783929705619812,
      "learning_rate": 1.8872875984236945e-05,
      "loss": 0.4578,
      "step": 5662
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9339381456375122,
      "learning_rate": 1.8872406230443095e-05,
      "loss": 0.6517,
      "step": 5663
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8998956680297852,
      "learning_rate": 1.8871936384628205e-05,
      "loss": 0.4785,
      "step": 5664
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9416934251785278,
      "learning_rate": 1.8871466446797148e-05,
      "loss": 0.616,
      "step": 5665
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7870794534683228,
      "learning_rate": 1.8870996416954796e-05,
      "loss": 0.455,
      "step": 5666
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7773336172103882,
      "learning_rate": 1.8870526295106028e-05,
      "loss": 0.513,
      "step": 5667
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.764772891998291,
      "learning_rate": 1.8870056081255715e-05,
      "loss": 0.4754,
      "step": 5668
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.052807331085205,
      "learning_rate": 1.8869585775408738e-05,
      "loss": 0.5359,
      "step": 5669
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8371734619140625,
      "learning_rate": 1.8869115377569974e-05,
      "loss": 0.5216,
      "step": 5670
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9118328094482422,
      "learning_rate": 1.88686448877443e-05,
      "loss": 0.5717,
      "step": 5671
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.695044755935669,
      "learning_rate": 1.8868174305936597e-05,
      "loss": 0.4436,
      "step": 5672
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8171180486679077,
      "learning_rate": 1.8867703632151748e-05,
      "loss": 0.5515,
      "step": 5673
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.757738471031189,
      "learning_rate": 1.886723286639463e-05,
      "loss": 0.491,
      "step": 5674
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8648581504821777,
      "learning_rate": 1.8866762008670124e-05,
      "loss": 0.5454,
      "step": 5675
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8158878087997437,
      "learning_rate": 1.886629105898312e-05,
      "loss": 0.4535,
      "step": 5676
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0670480728149414,
      "learning_rate": 1.8865820017338503e-05,
      "loss": 0.5809,
      "step": 5677
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.790245532989502,
      "learning_rate": 1.8865348883741155e-05,
      "loss": 0.4493,
      "step": 5678
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.809025764465332,
      "learning_rate": 1.886487765819596e-05,
      "loss": 0.511,
      "step": 5679
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7738041877746582,
      "learning_rate": 1.8864406340707807e-05,
      "loss": 0.4976,
      "step": 5680
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7776978015899658,
      "learning_rate": 1.886393493128159e-05,
      "loss": 0.5882,
      "step": 5681
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6543326377868652,
      "learning_rate": 1.886346342992219e-05,
      "loss": 0.456,
      "step": 5682
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.904619812965393,
      "learning_rate": 1.88629918366345e-05,
      "loss": 0.5792,
      "step": 5683
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.964901328086853,
      "learning_rate": 1.8862520151423415e-05,
      "loss": 0.4445,
      "step": 5684
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9903875589370728,
      "learning_rate": 1.886204837429382e-05,
      "loss": 0.5392,
      "step": 5685
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7992231845855713,
      "learning_rate": 1.8861576505250616e-05,
      "loss": 0.4694,
      "step": 5686
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.754201054573059,
      "learning_rate": 1.8861104544298692e-05,
      "loss": 0.5651,
      "step": 5687
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6993881464004517,
      "learning_rate": 1.8860632491442944e-05,
      "loss": 0.4637,
      "step": 5688
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.019540309906006,
      "learning_rate": 1.8860160346688268e-05,
      "loss": 0.5143,
      "step": 5689
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8141499757766724,
      "learning_rate": 1.8859688110039564e-05,
      "loss": 0.4583,
      "step": 5690
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7707585096359253,
      "learning_rate": 1.8859215781501727e-05,
      "loss": 0.5333,
      "step": 5691
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8122366666793823,
      "learning_rate": 1.885874336107965e-05,
      "loss": 0.5467,
      "step": 5692
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7159433364868164,
      "learning_rate": 1.885827084877824e-05,
      "loss": 0.5334,
      "step": 5693
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7023301124572754,
      "learning_rate": 1.8857798244602402e-05,
      "loss": 0.4362,
      "step": 5694
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.884130835533142,
      "learning_rate": 1.8857325548557024e-05,
      "loss": 0.5607,
      "step": 5695
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7242690324783325,
      "learning_rate": 1.8856852760647023e-05,
      "loss": 0.4504,
      "step": 5696
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7927688360214233,
      "learning_rate": 1.885637988087729e-05,
      "loss": 0.5311,
      "step": 5697
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6575956344604492,
      "learning_rate": 1.885590690925274e-05,
      "loss": 0.4473,
      "step": 5698
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7968361377716064,
      "learning_rate": 1.8855433845778274e-05,
      "loss": 0.5589,
      "step": 5699
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9322409629821777,
      "learning_rate": 1.8854960690458797e-05,
      "loss": 0.5205,
      "step": 5700
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7731939554214478,
      "learning_rate": 1.8854487443299218e-05,
      "loss": 0.4996,
      "step": 5701
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8089336156845093,
      "learning_rate": 1.8854014104304448e-05,
      "loss": 0.5362,
      "step": 5702
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9044286012649536,
      "learning_rate": 1.885354067347939e-05,
      "loss": 0.5553,
      "step": 5703
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8347463607788086,
      "learning_rate": 1.885306715082896e-05,
      "loss": 0.4976,
      "step": 5704
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.756345272064209,
      "learning_rate": 1.8852593536358065e-05,
      "loss": 0.4968,
      "step": 5705
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.03415584564209,
      "learning_rate": 1.8852119830071618e-05,
      "loss": 0.5475,
      "step": 5706
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8515682220458984,
      "learning_rate": 1.8851646031974535e-05,
      "loss": 0.4875,
      "step": 5707
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7933900356292725,
      "learning_rate": 1.885117214207173e-05,
      "loss": 0.5664,
      "step": 5708
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7823551893234253,
      "learning_rate": 1.885069816036811e-05,
      "loss": 0.5111,
      "step": 5709
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8274720907211304,
      "learning_rate": 1.8850224086868603e-05,
      "loss": 0.4568,
      "step": 5710
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9151774644851685,
      "learning_rate": 1.884974992157812e-05,
      "loss": 0.6117,
      "step": 5711
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7734878063201904,
      "learning_rate": 1.8849275664501573e-05,
      "loss": 0.4247,
      "step": 5712
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8093661069869995,
      "learning_rate": 1.884880131564389e-05,
      "loss": 0.5618,
      "step": 5713
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.786041021347046,
      "learning_rate": 1.884832687500999e-05,
      "loss": 0.4776,
      "step": 5714
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3250932693481445,
      "learning_rate": 1.8847852342604788e-05,
      "loss": 0.9157,
      "step": 5715
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.055985927581787,
      "learning_rate": 1.884737771843321e-05,
      "loss": 0.9213,
      "step": 5716
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9685933589935303,
      "learning_rate": 1.884690300250018e-05,
      "loss": 0.5585,
      "step": 5717
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0500333309173584,
      "learning_rate": 1.884642819481061e-05,
      "loss": 0.475,
      "step": 5718
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0293009281158447,
      "learning_rate": 1.8845953295369442e-05,
      "loss": 0.6099,
      "step": 5719
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2695693969726562,
      "learning_rate": 1.8845478304181594e-05,
      "loss": 0.4853,
      "step": 5720
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.122175931930542,
      "learning_rate": 1.8845003221251987e-05,
      "loss": 0.6796,
      "step": 5721
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8247361183166504,
      "learning_rate": 1.8844528046585554e-05,
      "loss": 0.4261,
      "step": 5722
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9351186752319336,
      "learning_rate": 1.884405278018722e-05,
      "loss": 0.5647,
      "step": 5723
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8346902132034302,
      "learning_rate": 1.884357742206192e-05,
      "loss": 0.4682,
      "step": 5724
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.037449836730957,
      "learning_rate": 1.8843101972214576e-05,
      "loss": 0.5563,
      "step": 5725
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8234184980392456,
      "learning_rate": 1.8842626430650127e-05,
      "loss": 0.4578,
      "step": 5726
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.794607162475586,
      "learning_rate": 1.88421507973735e-05,
      "loss": 0.5536,
      "step": 5727
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7755858898162842,
      "learning_rate": 1.8841675072389632e-05,
      "loss": 0.466,
      "step": 5728
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7669942378997803,
      "learning_rate": 1.8841199255703454e-05,
      "loss": 0.5397,
      "step": 5729
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7985455989837646,
      "learning_rate": 1.88407233473199e-05,
      "loss": 0.4373,
      "step": 5730
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.6249701976776123,
      "learning_rate": 1.8840247347243913e-05,
      "loss": 0.9241,
      "step": 5731
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.076148748397827,
      "learning_rate": 1.883977125548042e-05,
      "loss": 0.9241,
      "step": 5732
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2100908756256104,
      "learning_rate": 1.8839295072034366e-05,
      "loss": 0.6051,
      "step": 5733
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0160984992980957,
      "learning_rate": 1.8838818796910687e-05,
      "loss": 0.4801,
      "step": 5734
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2378344535827637,
      "learning_rate": 1.883834243011432e-05,
      "loss": 0.6317,
      "step": 5735
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8087067604064941,
      "learning_rate": 1.8837865971650212e-05,
      "loss": 0.4388,
      "step": 5736
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8924955129623413,
      "learning_rate": 1.88373894215233e-05,
      "loss": 0.603,
      "step": 5737
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7748818397521973,
      "learning_rate": 1.8836912779738526e-05,
      "loss": 0.4552,
      "step": 5738
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.853394865989685,
      "learning_rate": 1.883643604630084e-05,
      "loss": 0.5009,
      "step": 5739
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8569166660308838,
      "learning_rate": 1.8835959221215182e-05,
      "loss": 0.5157,
      "step": 5740
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8619194030761719,
      "learning_rate": 1.8835482304486495e-05,
      "loss": 0.5818,
      "step": 5741
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9671427011489868,
      "learning_rate": 1.883500529611973e-05,
      "loss": 0.4659,
      "step": 5742
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8172495365142822,
      "learning_rate": 1.883452819611983e-05,
      "loss": 0.5193,
      "step": 5743
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7868670225143433,
      "learning_rate": 1.8834051004491747e-05,
      "loss": 0.4494,
      "step": 5744
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8090187311172485,
      "learning_rate": 1.8833573721240425e-05,
      "loss": 0.5262,
      "step": 5745
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8549675941467285,
      "learning_rate": 1.8833096346370824e-05,
      "loss": 0.5557,
      "step": 5746
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.084433078765869,
      "learning_rate": 1.8832618879887884e-05,
      "loss": 0.6078,
      "step": 5747
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8700205087661743,
      "learning_rate": 1.8832141321796566e-05,
      "loss": 0.5168,
      "step": 5748
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9538806676864624,
      "learning_rate": 1.8831663672101816e-05,
      "loss": 0.5734,
      "step": 5749
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8422502279281616,
      "learning_rate": 1.883118593080859e-05,
      "loss": 0.4988,
      "step": 5750
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9568746089935303,
      "learning_rate": 1.8830708097921848e-05,
      "loss": 0.5478,
      "step": 5751
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7717528343200684,
      "learning_rate": 1.883023017344654e-05,
      "loss": 0.4792,
      "step": 5752
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7414205074310303,
      "learning_rate": 1.8829752157387625e-05,
      "loss": 0.5068,
      "step": 5753
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8397473096847534,
      "learning_rate": 1.8829274049750062e-05,
      "loss": 0.5216,
      "step": 5754
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8596465587615967,
      "learning_rate": 1.8828795850538804e-05,
      "loss": 0.6208,
      "step": 5755
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7810307741165161,
      "learning_rate": 1.882831755975882e-05,
      "loss": 0.4685,
      "step": 5756
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8114144802093506,
      "learning_rate": 1.8827839177415065e-05,
      "loss": 0.5681,
      "step": 5757
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8005945682525635,
      "learning_rate": 1.88273607035125e-05,
      "loss": 0.4687,
      "step": 5758
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8428611755371094,
      "learning_rate": 1.882688213805609e-05,
      "loss": 0.5477,
      "step": 5759
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7533912658691406,
      "learning_rate": 1.88264034810508e-05,
      "loss": 0.5013,
      "step": 5760
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7269587516784668,
      "learning_rate": 1.8825924732501585e-05,
      "loss": 0.497,
      "step": 5761
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8123812675476074,
      "learning_rate": 1.882544589241342e-05,
      "loss": 0.4883,
      "step": 5762
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.803391456604004,
      "learning_rate": 1.8824966960791268e-05,
      "loss": 0.5478,
      "step": 5763
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.744994044303894,
      "learning_rate": 1.88244879376401e-05,
      "loss": 0.4758,
      "step": 5764
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7712492942810059,
      "learning_rate": 1.8824008822964875e-05,
      "loss": 0.5464,
      "step": 5765
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8080852031707764,
      "learning_rate": 1.8823529616770574e-05,
      "loss": 0.4799,
      "step": 5766
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8647069931030273,
      "learning_rate": 1.8823050319062162e-05,
      "loss": 0.5639,
      "step": 5767
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7374473810195923,
      "learning_rate": 1.8822570929844604e-05,
      "loss": 0.4504,
      "step": 5768
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8549389839172363,
      "learning_rate": 1.8822091449122882e-05,
      "loss": 0.6001,
      "step": 5769
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6557739973068237,
      "learning_rate": 1.882161187690196e-05,
      "loss": 0.4689,
      "step": 5770
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7647266387939453,
      "learning_rate": 1.882113221318682e-05,
      "loss": 0.5041,
      "step": 5771
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7514806985855103,
      "learning_rate": 1.882065245798243e-05,
      "loss": 0.4926,
      "step": 5772
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8784263134002686,
      "learning_rate": 1.8820172611293774e-05,
      "loss": 0.5077,
      "step": 5773
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6360387802124023,
      "learning_rate": 1.881969267312582e-05,
      "loss": 0.4528,
      "step": 5774
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6794023513793945,
      "learning_rate": 1.881921264348355e-05,
      "loss": 0.5044,
      "step": 5775
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9240649938583374,
      "learning_rate": 1.8818732522371945e-05,
      "loss": 0.4856,
      "step": 5776
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.074084520339966,
      "learning_rate": 1.881825230979598e-05,
      "loss": 0.5276,
      "step": 5777
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8609083890914917,
      "learning_rate": 1.881777200576064e-05,
      "loss": 0.4976,
      "step": 5778
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8206441402435303,
      "learning_rate": 1.8817291610270896e-05,
      "loss": 0.5431,
      "step": 5779
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7951349020004272,
      "learning_rate": 1.8816811123331747e-05,
      "loss": 0.4623,
      "step": 5780
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.896865963935852,
      "learning_rate": 1.8816330544948164e-05,
      "loss": 0.5369,
      "step": 5781
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7391600608825684,
      "learning_rate": 1.8815849875125135e-05,
      "loss": 0.4412,
      "step": 5782
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8139292001724243,
      "learning_rate": 1.8815369113867647e-05,
      "loss": 0.4958,
      "step": 5783
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9425417184829712,
      "learning_rate": 1.8814888261180684e-05,
      "loss": 0.516,
      "step": 5784
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.923443078994751,
      "learning_rate": 1.8814407317069233e-05,
      "loss": 0.5829,
      "step": 5785
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7002995014190674,
      "learning_rate": 1.8813926281538286e-05,
      "loss": 0.4833,
      "step": 5786
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7636628150939941,
      "learning_rate": 1.8813445154592825e-05,
      "loss": 0.525,
      "step": 5787
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.858864665031433,
      "learning_rate": 1.881296393623785e-05,
      "loss": 0.447,
      "step": 5788
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7966524362564087,
      "learning_rate": 1.8812482626478342e-05,
      "loss": 0.5849,
      "step": 5789
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7835506200790405,
      "learning_rate": 1.8812001225319297e-05,
      "loss": 0.4565,
      "step": 5790
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.8246829509735107,
      "learning_rate": 1.8811519732765712e-05,
      "loss": 0.9954,
      "step": 5791
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.220066785812378,
      "learning_rate": 1.8811038148822575e-05,
      "loss": 0.9438,
      "step": 5792
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0461835861206055,
      "learning_rate": 1.8810556473494882e-05,
      "loss": 0.5809,
      "step": 5793
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9153872728347778,
      "learning_rate": 1.8810074706787632e-05,
      "loss": 0.4992,
      "step": 5794
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9429758787155151,
      "learning_rate": 1.8809592848705815e-05,
      "loss": 0.5195,
      "step": 5795
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7905457019805908,
      "learning_rate": 1.8809110899254436e-05,
      "loss": 0.5151,
      "step": 5796
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.003689765930176,
      "learning_rate": 1.8808628858438488e-05,
      "loss": 0.5781,
      "step": 5797
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0490620136260986,
      "learning_rate": 1.8808146726262976e-05,
      "loss": 0.4653,
      "step": 5798
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8367568254470825,
      "learning_rate": 1.8807664502732897e-05,
      "loss": 0.5279,
      "step": 5799
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.742935061454773,
      "learning_rate": 1.8807182187853252e-05,
      "loss": 0.469,
      "step": 5800
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8509695529937744,
      "learning_rate": 1.8806699781629042e-05,
      "loss": 0.5584,
      "step": 5801
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7847222089767456,
      "learning_rate": 1.880621728406528e-05,
      "loss": 0.4675,
      "step": 5802
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8377633094787598,
      "learning_rate": 1.8805734695166955e-05,
      "loss": 0.5101,
      "step": 5803
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.744134783744812,
      "learning_rate": 1.8805252014939083e-05,
      "loss": 0.4595,
      "step": 5804
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9095242023468018,
      "learning_rate": 1.880476924338667e-05,
      "loss": 0.5894,
      "step": 5805
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7681089639663696,
      "learning_rate": 1.8804286380514715e-05,
      "loss": 0.4602,
      "step": 5806
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7356680631637573,
      "learning_rate": 1.8803803426328236e-05,
      "loss": 0.5196,
      "step": 5807
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9931635856628418,
      "learning_rate": 1.8803320380832235e-05,
      "loss": 0.5203,
      "step": 5808
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9083888530731201,
      "learning_rate": 1.8802837244031727e-05,
      "loss": 0.511,
      "step": 5809
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8970425128936768,
      "learning_rate": 1.880235401593172e-05,
      "loss": 0.463,
      "step": 5810
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.764151930809021,
      "learning_rate": 1.8801870696537224e-05,
      "loss": 0.5317,
      "step": 5811
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.750167965888977,
      "learning_rate": 1.8801387285853254e-05,
      "loss": 0.4904,
      "step": 5812
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9606742858886719,
      "learning_rate": 1.8800903783884825e-05,
      "loss": 0.6115,
      "step": 5813
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9024115800857544,
      "learning_rate": 1.880042019063695e-05,
      "loss": 0.4762,
      "step": 5814
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7484921216964722,
      "learning_rate": 1.8799936506114645e-05,
      "loss": 0.5226,
      "step": 5815
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.786099910736084,
      "learning_rate": 1.8799452730322928e-05,
      "loss": 0.4879,
      "step": 5816
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9196776151657104,
      "learning_rate": 1.879896886326681e-05,
      "loss": 0.5234,
      "step": 5817
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8496472835540771,
      "learning_rate": 1.879848490495132e-05,
      "loss": 0.512,
      "step": 5818
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1514506340026855,
      "learning_rate": 1.8798000855381472e-05,
      "loss": 0.7133,
      "step": 5819
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7865078449249268,
      "learning_rate": 1.879751671456228e-05,
      "loss": 0.4465,
      "step": 5820
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.719933032989502,
      "learning_rate": 1.879703248249878e-05,
      "loss": 0.5041,
      "step": 5821
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8143550157546997,
      "learning_rate": 1.8796548159195983e-05,
      "loss": 0.5345,
      "step": 5822
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6914218664169312,
      "learning_rate": 1.8796063744658916e-05,
      "loss": 0.4959,
      "step": 5823
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8493268489837646,
      "learning_rate": 1.87955792388926e-05,
      "loss": 0.5059,
      "step": 5824
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9566996097564697,
      "learning_rate": 1.879509464190206e-05,
      "loss": 0.6049,
      "step": 5825
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7738057374954224,
      "learning_rate": 1.879460995369233e-05,
      "loss": 0.4509,
      "step": 5826
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8350720405578613,
      "learning_rate": 1.8794125174268432e-05,
      "loss": 0.5366,
      "step": 5827
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8266520500183105,
      "learning_rate": 1.879364030363539e-05,
      "loss": 0.4759,
      "step": 5828
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9470542669296265,
      "learning_rate": 1.8793155341798237e-05,
      "loss": 0.6283,
      "step": 5829
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8131558895111084,
      "learning_rate": 1.8792670288762004e-05,
      "loss": 0.4523,
      "step": 5830
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9152089357376099,
      "learning_rate": 1.8792185144531717e-05,
      "loss": 0.6456,
      "step": 5831
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7811455726623535,
      "learning_rate": 1.8791699909112414e-05,
      "loss": 0.4794,
      "step": 5832
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0422394275665283,
      "learning_rate": 1.879121458250912e-05,
      "loss": 0.6209,
      "step": 5833
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8666735887527466,
      "learning_rate": 1.8790729164726877e-05,
      "loss": 0.4917,
      "step": 5834
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9372665882110596,
      "learning_rate": 1.879024365577071e-05,
      "loss": 0.593,
      "step": 5835
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.707625389099121,
      "learning_rate": 1.8789758055645666e-05,
      "loss": 0.4245,
      "step": 5836
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9951136112213135,
      "learning_rate": 1.8789272364356775e-05,
      "loss": 0.5325,
      "step": 5837
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7534602880477905,
      "learning_rate": 1.878878658190907e-05,
      "loss": 0.4576,
      "step": 5838
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8454129695892334,
      "learning_rate": 1.87883007083076e-05,
      "loss": 0.5391,
      "step": 5839
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6736690998077393,
      "learning_rate": 1.8787814743557395e-05,
      "loss": 0.4597,
      "step": 5840
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8633486032485962,
      "learning_rate": 1.8787328687663497e-05,
      "loss": 0.5636,
      "step": 5841
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8652704954147339,
      "learning_rate": 1.878684254063095e-05,
      "loss": 0.5208,
      "step": 5842
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7495768070220947,
      "learning_rate": 1.8786356302464796e-05,
      "loss": 0.5218,
      "step": 5843
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7220975160598755,
      "learning_rate": 1.878586997317008e-05,
      "loss": 0.4641,
      "step": 5844
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.869248390197754,
      "learning_rate": 1.878538355275184e-05,
      "loss": 0.5088,
      "step": 5845
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.82390296459198,
      "learning_rate": 1.878489704121512e-05,
      "loss": 0.462,
      "step": 5846
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8321861028671265,
      "learning_rate": 1.8784410438564976e-05,
      "loss": 0.5292,
      "step": 5847
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7427700757980347,
      "learning_rate": 1.8783923744806446e-05,
      "loss": 0.4504,
      "step": 5848
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.885132074356079,
      "learning_rate": 1.8783436959944583e-05,
      "loss": 0.6152,
      "step": 5849
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7537579536437988,
      "learning_rate": 1.878295008398443e-05,
      "loss": 0.4541,
      "step": 5850
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7607876062393188,
      "learning_rate": 1.8782463116931042e-05,
      "loss": 0.5119,
      "step": 5851
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.670418381690979,
      "learning_rate": 1.878197605878947e-05,
      "loss": 0.4682,
      "step": 5852
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8312196731567383,
      "learning_rate": 1.878148890956476e-05,
      "loss": 0.5887,
      "step": 5853
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8329291343688965,
      "learning_rate": 1.878100166926197e-05,
      "loss": 0.4607,
      "step": 5854
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7780015468597412,
      "learning_rate": 1.8780514337886148e-05,
      "loss": 0.5415,
      "step": 5855
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7351869344711304,
      "learning_rate": 1.8780026915442355e-05,
      "loss": 0.4998,
      "step": 5856
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7091675996780396,
      "learning_rate": 1.8779539401935638e-05,
      "loss": 0.5411,
      "step": 5857
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8589284420013428,
      "learning_rate": 1.877905179737106e-05,
      "loss": 0.5125,
      "step": 5858
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8182141780853271,
      "learning_rate": 1.8778564101753677e-05,
      "loss": 0.5448,
      "step": 5859
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8286333084106445,
      "learning_rate": 1.877807631508855e-05,
      "loss": 0.5087,
      "step": 5860
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8161635398864746,
      "learning_rate": 1.877758843738073e-05,
      "loss": 0.5427,
      "step": 5861
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8215810060501099,
      "learning_rate": 1.8777100468635283e-05,
      "loss": 0.4498,
      "step": 5862
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.812286376953125,
      "learning_rate": 1.877661240885727e-05,
      "loss": 0.5273,
      "step": 5863
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6632637977600098,
      "learning_rate": 1.8776124258051752e-05,
      "loss": 0.4531,
      "step": 5864
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.833975911140442,
      "learning_rate": 1.877563601622379e-05,
      "loss": 0.5438,
      "step": 5865
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8620274066925049,
      "learning_rate": 1.877514768337845e-05,
      "loss": 0.547,
      "step": 5866
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7395002841949463,
      "learning_rate": 1.8774659259520798e-05,
      "loss": 0.4793,
      "step": 5867
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.701988935470581,
      "learning_rate": 1.87741707446559e-05,
      "loss": 0.4796,
      "step": 5868
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9349627494812012,
      "learning_rate": 1.8773682138788815e-05,
      "loss": 0.5716,
      "step": 5869
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.663910984992981,
      "learning_rate": 1.8773193441924616e-05,
      "loss": 0.4644,
      "step": 5870
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9541257619857788,
      "learning_rate": 1.8772704654068378e-05,
      "loss": 0.5778,
      "step": 5871
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.811086654663086,
      "learning_rate": 1.877221577522516e-05,
      "loss": 0.485,
      "step": 5872
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3503525257110596,
      "learning_rate": 1.8771726805400037e-05,
      "loss": 0.9585,
      "step": 5873
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.012693405151367,
      "learning_rate": 1.8771237744598084e-05,
      "loss": 0.9446,
      "step": 5874
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0571277141571045,
      "learning_rate": 1.8770748592824368e-05,
      "loss": 0.5678,
      "step": 5875
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9352800846099854,
      "learning_rate": 1.8770259350083962e-05,
      "loss": 0.5027,
      "step": 5876
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.854994535446167,
      "learning_rate": 1.8769770016381945e-05,
      "loss": 0.5513,
      "step": 5877
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9476934671401978,
      "learning_rate": 1.8769280591723385e-05,
      "loss": 0.5625,
      "step": 5878
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.786460280418396,
      "learning_rate": 1.876879107611337e-05,
      "loss": 0.4815,
      "step": 5879
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.874855637550354,
      "learning_rate": 1.8768301469556964e-05,
      "loss": 0.4833,
      "step": 5880
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8493824005126953,
      "learning_rate": 1.876781177205925e-05,
      "loss": 0.5273,
      "step": 5881
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9406213760375977,
      "learning_rate": 1.876732198362531e-05,
      "loss": 0.4891,
      "step": 5882
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7736116647720337,
      "learning_rate": 1.876683210426022e-05,
      "loss": 0.5204,
      "step": 5883
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8117491006851196,
      "learning_rate": 1.8766342133969064e-05,
      "loss": 0.4623,
      "step": 5884
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7625257968902588,
      "learning_rate": 1.876585207275692e-05,
      "loss": 0.5357,
      "step": 5885
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8800549507141113,
      "learning_rate": 1.8765361920628874e-05,
      "loss": 0.48,
      "step": 5886
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8238259553909302,
      "learning_rate": 1.8764871677590008e-05,
      "loss": 0.5465,
      "step": 5887
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7656348943710327,
      "learning_rate": 1.8764381343645404e-05,
      "loss": 0.4565,
      "step": 5888
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.4768850803375244,
      "learning_rate": 1.8763890918800153e-05,
      "loss": 0.5583,
      "step": 5889
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.700771689414978,
      "learning_rate": 1.876340040305934e-05,
      "loss": 0.4659,
      "step": 5890
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9988354444503784,
      "learning_rate": 1.8762909796428052e-05,
      "loss": 0.5311,
      "step": 5891
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8783143758773804,
      "learning_rate": 1.8762419098911375e-05,
      "loss": 0.4948,
      "step": 5892
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.071559429168701,
      "learning_rate": 1.87619283105144e-05,
      "loss": 0.6013,
      "step": 5893
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8281946182250977,
      "learning_rate": 1.8761437431242218e-05,
      "loss": 0.4691,
      "step": 5894
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9585216045379639,
      "learning_rate": 1.876094646109992e-05,
      "loss": 0.5612,
      "step": 5895
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7623240947723389,
      "learning_rate": 1.8760455400092595e-05,
      "loss": 0.5161,
      "step": 5896
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7645642757415771,
      "learning_rate": 1.875996424822534e-05,
      "loss": 0.5723,
      "step": 5897
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6503123044967651,
      "learning_rate": 1.8759473005503253e-05,
      "loss": 0.4535,
      "step": 5898
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7632100582122803,
      "learning_rate": 1.875898167193142e-05,
      "loss": 0.537,
      "step": 5899
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7116180658340454,
      "learning_rate": 1.875849024751494e-05,
      "loss": 0.4457,
      "step": 5900
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.790718913078308,
      "learning_rate": 1.8757998732258912e-05,
      "loss": 0.5316,
      "step": 5901
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8893723487854004,
      "learning_rate": 1.875750712616843e-05,
      "loss": 0.4836,
      "step": 5902
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.790869951248169,
      "learning_rate": 1.8757015429248598e-05,
      "loss": 0.4777,
      "step": 5903
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.185960292816162,
      "learning_rate": 1.8756523641504508e-05,
      "loss": 0.5503,
      "step": 5904
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8204907178878784,
      "learning_rate": 1.875603176294127e-05,
      "loss": 0.5284,
      "step": 5905
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6892094612121582,
      "learning_rate": 1.875553979356398e-05,
      "loss": 0.4942,
      "step": 5906
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.738155722618103,
      "learning_rate": 1.8755047733377742e-05,
      "loss": 0.5167,
      "step": 5907
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8009206056594849,
      "learning_rate": 1.8754555582387656e-05,
      "loss": 0.501,
      "step": 5908
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8435726165771484,
      "learning_rate": 1.8754063340598832e-05,
      "loss": 0.4907,
      "step": 5909
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7574915885925293,
      "learning_rate": 1.8753571008016374e-05,
      "loss": 0.4688,
      "step": 5910
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8696154356002808,
      "learning_rate": 1.8753078584645384e-05,
      "loss": 0.5499,
      "step": 5911
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7564914226531982,
      "learning_rate": 1.8752586070490975e-05,
      "loss": 0.4552,
      "step": 5912
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.132589101791382,
      "learning_rate": 1.875209346555825e-05,
      "loss": 0.6047,
      "step": 5913
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.681136131286621,
      "learning_rate": 1.8751600769852323e-05,
      "loss": 0.4409,
      "step": 5914
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9114665985107422,
      "learning_rate": 1.87511079833783e-05,
      "loss": 0.6171,
      "step": 5915
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7339993715286255,
      "learning_rate": 1.8750615106141293e-05,
      "loss": 0.4711,
      "step": 5916
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.773870825767517,
      "learning_rate": 1.8750122138146414e-05,
      "loss": 0.5231,
      "step": 5917
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7687230110168457,
      "learning_rate": 1.8749629079398778e-05,
      "loss": 0.4731,
      "step": 5918
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8260878324508667,
      "learning_rate": 1.8749135929903495e-05,
      "loss": 0.5598,
      "step": 5919
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.942844271659851,
      "learning_rate": 1.8748642689665685e-05,
      "loss": 0.5659,
      "step": 5920
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.242116928100586,
      "learning_rate": 1.874814935869046e-05,
      "loss": 0.4985,
      "step": 5921
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7766200304031372,
      "learning_rate": 1.8747655936982934e-05,
      "loss": 0.5213,
      "step": 5922
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.705066204071045,
      "learning_rate": 1.874716242454823e-05,
      "loss": 0.4698,
      "step": 5923
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7438256740570068,
      "learning_rate": 1.874666882139146e-05,
      "loss": 0.507,
      "step": 5924
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.935644268989563,
      "learning_rate": 1.8746175127517753e-05,
      "loss": 0.5941,
      "step": 5925
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7801337242126465,
      "learning_rate": 1.874568134293222e-05,
      "loss": 0.4707,
      "step": 5926
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8704211711883545,
      "learning_rate": 1.8745187467639988e-05,
      "loss": 0.6196,
      "step": 5927
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7521096467971802,
      "learning_rate": 1.8744693501646184e-05,
      "loss": 0.504,
      "step": 5928
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.969803810119629,
      "learning_rate": 1.8744199444955917e-05,
      "loss": 0.5937,
      "step": 5929
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.742769718170166,
      "learning_rate": 1.874370529757432e-05,
      "loss": 0.4612,
      "step": 5930
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7043027877807617,
      "learning_rate": 1.874321105950652e-05,
      "loss": 0.4989,
      "step": 5931
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6785482168197632,
      "learning_rate": 1.874271673075764e-05,
      "loss": 0.4147,
      "step": 5932
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.848605751991272,
      "learning_rate": 1.8742222311332806e-05,
      "loss": 0.5718,
      "step": 5933
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7888362407684326,
      "learning_rate": 1.8741727801237146e-05,
      "loss": 0.471,
      "step": 5934
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7701364755630493,
      "learning_rate": 1.874123320047579e-05,
      "loss": 0.5464,
      "step": 5935
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.823993444442749,
      "learning_rate": 1.8740738509053868e-05,
      "loss": 0.4846,
      "step": 5936
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8733503818511963,
      "learning_rate": 1.8740243726976515e-05,
      "loss": 0.5381,
      "step": 5937
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9941086769104004,
      "learning_rate": 1.8739748854248852e-05,
      "loss": 0.5094,
      "step": 5938
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8679906129837036,
      "learning_rate": 1.873925389087602e-05,
      "loss": 0.5252,
      "step": 5939
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.749732255935669,
      "learning_rate": 1.8738758836863154e-05,
      "loss": 0.4371,
      "step": 5940
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8438916206359863,
      "learning_rate": 1.873826369221538e-05,
      "loss": 0.591,
      "step": 5941
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.755820870399475,
      "learning_rate": 1.873776845693784e-05,
      "loss": 0.4468,
      "step": 5942
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7290031909942627,
      "learning_rate": 1.873727313103567e-05,
      "loss": 0.4847,
      "step": 5943
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.665920615196228,
      "learning_rate": 1.8736777714514003e-05,
      "loss": 0.4556,
      "step": 5944
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9329907894134521,
      "learning_rate": 1.873628220737798e-05,
      "loss": 0.5899,
      "step": 5945
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.743227481842041,
      "learning_rate": 1.8735786609632746e-05,
      "loss": 0.4508,
      "step": 5946
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.767630696296692,
      "learning_rate": 1.873529092128343e-05,
      "loss": 0.5748,
      "step": 5947
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7939988374710083,
      "learning_rate": 1.873479514233518e-05,
      "loss": 0.471,
      "step": 5948
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8156931400299072,
      "learning_rate": 1.873429927279314e-05,
      "loss": 0.5295,
      "step": 5949
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8163681030273438,
      "learning_rate": 1.8733803312662448e-05,
      "loss": 0.489,
      "step": 5950
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8597040176391602,
      "learning_rate": 1.8733307261948247e-05,
      "loss": 0.5792,
      "step": 5951
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7573219537734985,
      "learning_rate": 1.8732811120655687e-05,
      "loss": 0.4911,
      "step": 5952
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9001004695892334,
      "learning_rate": 1.8732314888789908e-05,
      "loss": 0.5449,
      "step": 5953
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7568947076797485,
      "learning_rate": 1.8731818566356063e-05,
      "loss": 0.4591,
      "step": 5954
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7891086339950562,
      "learning_rate": 1.8731322153359298e-05,
      "loss": 0.498,
      "step": 5955
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8848282098770142,
      "learning_rate": 1.873082564980476e-05,
      "loss": 0.5296,
      "step": 5956
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8678737878799438,
      "learning_rate": 1.8730329055697595e-05,
      "loss": 0.5474,
      "step": 5957
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7628520727157593,
      "learning_rate": 1.872983237104296e-05,
      "loss": 0.4769,
      "step": 5958
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.672659158706665,
      "learning_rate": 1.8729335595846e-05,
      "loss": 0.4906,
      "step": 5959
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7154195308685303,
      "learning_rate": 1.8728838730111877e-05,
      "loss": 0.4828,
      "step": 5960
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8231866359710693,
      "learning_rate": 1.8728341773845733e-05,
      "loss": 0.5924,
      "step": 5961
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.676512360572815,
      "learning_rate": 1.872784472705273e-05,
      "loss": 0.4819,
      "step": 5962
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9185717105865479,
      "learning_rate": 1.872734758973802e-05,
      "loss": 0.6368,
      "step": 5963
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7244971990585327,
      "learning_rate": 1.8726850361906758e-05,
      "loss": 0.4928,
      "step": 5964
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8101074695587158,
      "learning_rate": 1.8726353043564106e-05,
      "loss": 0.5417,
      "step": 5965
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7068891525268555,
      "learning_rate": 1.8725855634715215e-05,
      "loss": 0.4842,
      "step": 5966
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.4391980171203613,
      "learning_rate": 1.8725358135365254e-05,
      "loss": 0.9093,
      "step": 5967
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.104703664779663,
      "learning_rate": 1.872486054551937e-05,
      "loss": 0.8774,
      "step": 5968
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0759799480438232,
      "learning_rate": 1.872436286518273e-05,
      "loss": 0.5866,
      "step": 5969
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9400246143341064,
      "learning_rate": 1.8723865094360503e-05,
      "loss": 0.4963,
      "step": 5970
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.832597255706787,
      "learning_rate": 1.872336723305784e-05,
      "loss": 0.5491,
      "step": 5971
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7682493925094604,
      "learning_rate": 1.872286928127991e-05,
      "loss": 0.4529,
      "step": 5972
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.036231517791748,
      "learning_rate": 1.872237123903188e-05,
      "loss": 0.5149,
      "step": 5973
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8079322576522827,
      "learning_rate": 1.872187310631891e-05,
      "loss": 0.4732,
      "step": 5974
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.762858271598816,
      "learning_rate": 1.872137488314617e-05,
      "loss": 0.5585,
      "step": 5975
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7202757596969604,
      "learning_rate": 1.872087656951882e-05,
      "loss": 0.4433,
      "step": 5976
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9891753196716309,
      "learning_rate": 1.8720378165442043e-05,
      "loss": 0.5661,
      "step": 5977
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8234446048736572,
      "learning_rate": 1.8719879670920994e-05,
      "loss": 0.4883,
      "step": 5978
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7322033643722534,
      "learning_rate": 1.871938108596085e-05,
      "loss": 0.5675,
      "step": 5979
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.673210859298706,
      "learning_rate": 1.8718882410566785e-05,
      "loss": 0.4381,
      "step": 5980
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.84633207321167,
      "learning_rate": 1.8718383644743963e-05,
      "loss": 0.5205,
      "step": 5981
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8666749000549316,
      "learning_rate": 1.8717884788497565e-05,
      "loss": 0.4616,
      "step": 5982
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8354953527450562,
      "learning_rate": 1.8717385841832758e-05,
      "loss": 0.5515,
      "step": 5983
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.902246117591858,
      "learning_rate": 1.8716886804754723e-05,
      "loss": 0.4923,
      "step": 5984
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.81864595413208,
      "learning_rate": 1.871638767726863e-05,
      "loss": 0.5723,
      "step": 5985
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8831135034561157,
      "learning_rate": 1.871588845937966e-05,
      "loss": 0.5756,
      "step": 5986
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8972866535186768,
      "learning_rate": 1.871538915109299e-05,
      "loss": 0.5512,
      "step": 5987
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8005387783050537,
      "learning_rate": 1.8714889752413796e-05,
      "loss": 0.4584,
      "step": 5988
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.7801012992858887,
      "learning_rate": 1.871439026334726e-05,
      "loss": 0.9258,
      "step": 5989
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.2966675758361816,
      "learning_rate": 1.8713890683898562e-05,
      "loss": 0.9097,
      "step": 5990
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0916812419891357,
      "learning_rate": 1.8713391014072885e-05,
      "loss": 0.6247,
      "step": 5991
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.933083176612854,
      "learning_rate": 1.8712891253875406e-05,
      "loss": 0.468,
      "step": 5992
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.813148021697998,
      "learning_rate": 1.8712391403311315e-05,
      "loss": 0.598,
      "step": 5993
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7532076835632324,
      "learning_rate": 1.8711891462385794e-05,
      "loss": 0.4462,
      "step": 5994
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7118394374847412,
      "learning_rate": 1.8711391431104026e-05,
      "loss": 0.5363,
      "step": 5995
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8934892416000366,
      "learning_rate": 1.8710891309471198e-05,
      "loss": 0.4635,
      "step": 5996
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8459793329238892,
      "learning_rate": 1.87103910974925e-05,
      "loss": 0.5074,
      "step": 5997
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9113398790359497,
      "learning_rate": 1.8709890795173112e-05,
      "loss": 0.508,
      "step": 5998
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9332860708236694,
      "learning_rate": 1.8709390402518236e-05,
      "loss": 0.568,
      "step": 5999
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7493945360183716,
      "learning_rate": 1.870888991953305e-05,
      "loss": 0.4432,
      "step": 6000
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.113154411315918,
      "learning_rate": 1.8708389346222754e-05,
      "loss": 0.6039,
      "step": 6001
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9145368337631226,
      "learning_rate": 1.8707888682592527e-05,
      "loss": 0.5008,
      "step": 6002
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0319697856903076,
      "learning_rate": 1.8707387928647578e-05,
      "loss": 0.6406,
      "step": 6003
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9367079734802246,
      "learning_rate": 1.8706887084393087e-05,
      "loss": 0.5054,
      "step": 6004
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9505382776260376,
      "learning_rate": 1.8706386149834256e-05,
      "loss": 0.5641,
      "step": 6005
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8914445638656616,
      "learning_rate": 1.8705885124976278e-05,
      "loss": 0.5827,
      "step": 6006
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.817706823348999,
      "learning_rate": 1.870538400982435e-05,
      "loss": 0.5718,
      "step": 6007
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9178557395935059,
      "learning_rate": 1.8704882804383668e-05,
      "loss": 0.5116,
      "step": 6008
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0071446895599365,
      "learning_rate": 1.8704381508659434e-05,
      "loss": 0.6318,
      "step": 6009
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7707444429397583,
      "learning_rate": 1.8703880122656843e-05,
      "loss": 0.5541,
      "step": 6010
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.800685167312622,
      "learning_rate": 1.87033786463811e-05,
      "loss": 0.5833,
      "step": 6011
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.734501838684082,
      "learning_rate": 1.87028770798374e-05,
      "loss": 0.5077,
      "step": 6012
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7561177015304565,
      "learning_rate": 1.8702375423030948e-05,
      "loss": 0.5815,
      "step": 6013
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8007726669311523,
      "learning_rate": 1.8701873675966946e-05,
      "loss": 0.548,
      "step": 6014
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9118967056274414,
      "learning_rate": 1.87013718386506e-05,
      "loss": 0.6313,
      "step": 6015
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7101868391036987,
      "learning_rate": 1.8700869911087115e-05,
      "loss": 0.5191,
      "step": 6016
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9227485656738281,
      "learning_rate": 1.8700367893281696e-05,
      "loss": 0.6351,
      "step": 6017
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.782122254371643,
      "learning_rate": 1.8699865785239547e-05,
      "loss": 0.4752,
      "step": 6018
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9557548761367798,
      "learning_rate": 1.8699363586965882e-05,
      "loss": 0.5911,
      "step": 6019
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9361021518707275,
      "learning_rate": 1.86988612984659e-05,
      "loss": 0.5802,
      "step": 6020
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8939409255981445,
      "learning_rate": 1.8698358919744822e-05,
      "loss": 0.6069,
      "step": 6021
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8425999879837036,
      "learning_rate": 1.8697856450807847e-05,
      "loss": 0.5538,
      "step": 6022
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.4502811431884766,
      "learning_rate": 1.8697353891660196e-05,
      "loss": 0.9759,
      "step": 6023
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.953752040863037,
      "learning_rate": 1.8696851242307076e-05,
      "loss": 1.1399,
      "step": 6024
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1913552284240723,
      "learning_rate": 1.8696348502753703e-05,
      "loss": 0.5705,
      "step": 6025
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9793883562088013,
      "learning_rate": 1.8695845673005286e-05,
      "loss": 0.4859,
      "step": 6026
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.882990837097168,
      "learning_rate": 1.869534275306705e-05,
      "loss": 0.5875,
      "step": 6027
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7072350978851318,
      "learning_rate": 1.8694839742944202e-05,
      "loss": 0.5212,
      "step": 6028
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9873546361923218,
      "learning_rate": 1.8694336642641962e-05,
      "loss": 0.6443,
      "step": 6029
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8474547863006592,
      "learning_rate": 1.8693833452165548e-05,
      "loss": 0.5048,
      "step": 6030
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.152587413787842,
      "learning_rate": 1.8693330171520175e-05,
      "loss": 0.5796,
      "step": 6031
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8822187185287476,
      "learning_rate": 1.8692826800711073e-05,
      "loss": 0.5925,
      "step": 6032
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.851276159286499,
      "learning_rate": 1.8692323339743455e-05,
      "loss": 0.5702,
      "step": 6033
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8784621953964233,
      "learning_rate": 1.8691819788622543e-05,
      "loss": 0.617,
      "step": 6034
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9628127813339233,
      "learning_rate": 1.869131614735356e-05,
      "loss": 0.634,
      "step": 6035
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8256441354751587,
      "learning_rate": 1.8690812415941733e-05,
      "loss": 0.5744,
      "step": 6036
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.681509256362915,
      "learning_rate": 1.8690308594392284e-05,
      "loss": 0.5485,
      "step": 6037
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6945699453353882,
      "learning_rate": 1.8689804682710435e-05,
      "loss": 0.5302,
      "step": 6038
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.758583664894104,
      "learning_rate": 1.8689300680901418e-05,
      "loss": 0.5684,
      "step": 6039
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9785970449447632,
      "learning_rate": 1.8688796588970457e-05,
      "loss": 0.6115,
      "step": 6040
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8863931894302368,
      "learning_rate": 1.8688292406922783e-05,
      "loss": 0.5761,
      "step": 6041
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7147711515426636,
      "learning_rate": 1.868778813476362e-05,
      "loss": 0.502,
      "step": 6042
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8436166048049927,
      "learning_rate": 1.8687283772498205e-05,
      "loss": 0.5869,
      "step": 6043
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8012417554855347,
      "learning_rate": 1.8686779320131764e-05,
      "loss": 0.5369,
      "step": 6044
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6977790594100952,
      "learning_rate": 1.8686274777669534e-05,
      "loss": 0.6024,
      "step": 6045
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7421258687973022,
      "learning_rate": 1.8685770145116742e-05,
      "loss": 0.522,
      "step": 6046
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7574305534362793,
      "learning_rate": 1.8685265422478624e-05,
      "loss": 0.5503,
      "step": 6047
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7720136642456055,
      "learning_rate": 1.8684760609760416e-05,
      "loss": 0.5223,
      "step": 6048
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8387649059295654,
      "learning_rate": 1.8684255706967352e-05,
      "loss": 0.559,
      "step": 6049
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8571072816848755,
      "learning_rate": 1.868375071410467e-05,
      "loss": 0.4954,
      "step": 6050
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7570677995681763,
      "learning_rate": 1.8683245631177606e-05,
      "loss": 0.5766,
      "step": 6051
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7338905334472656,
      "learning_rate": 1.86827404581914e-05,
      "loss": 0.537,
      "step": 6052
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8269271850585938,
      "learning_rate": 1.8682235195151293e-05,
      "loss": 0.6459,
      "step": 6053
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.765023946762085,
      "learning_rate": 1.868172984206252e-05,
      "loss": 0.5093,
      "step": 6054
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8585858345031738,
      "learning_rate": 1.8681224398930328e-05,
      "loss": 0.6048,
      "step": 6055
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7949886322021484,
      "learning_rate": 1.8680718865759956e-05,
      "loss": 0.4842,
      "step": 6056
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9269307851791382,
      "learning_rate": 1.868021324255665e-05,
      "loss": 0.6666,
      "step": 6057
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8902770280838013,
      "learning_rate": 1.867970752932565e-05,
      "loss": 0.5134,
      "step": 6058
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.803694248199463,
      "learning_rate": 1.8679201726072206e-05,
      "loss": 0.5454,
      "step": 6059
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8233909606933594,
      "learning_rate": 1.8678695832801558e-05,
      "loss": 0.4847,
      "step": 6060
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.980847954750061,
      "learning_rate": 1.867818984951896e-05,
      "loss": 0.6097,
      "step": 6061
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.744550108909607,
      "learning_rate": 1.8677683776229652e-05,
      "loss": 0.5004,
      "step": 6062
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6987204551696777,
      "learning_rate": 1.8677177612938887e-05,
      "loss": 0.5564,
      "step": 6063
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7648470401763916,
      "learning_rate": 1.8676671359651918e-05,
      "loss": 0.5657,
      "step": 6064
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7386679649353027,
      "learning_rate": 1.8676165016373994e-05,
      "loss": 0.5501,
      "step": 6065
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7206916809082031,
      "learning_rate": 1.8675658583110362e-05,
      "loss": 0.4944,
      "step": 6066
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.840989589691162,
      "learning_rate": 1.8675152059866275e-05,
      "loss": 0.6329,
      "step": 6067
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.892154335975647,
      "learning_rate": 1.8674645446646994e-05,
      "loss": 0.5228,
      "step": 6068
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8996610641479492,
      "learning_rate": 1.8674138743457764e-05,
      "loss": 0.6427,
      "step": 6069
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8530333042144775,
      "learning_rate": 1.8673631950303846e-05,
      "loss": 0.5268,
      "step": 6070
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.887542724609375,
      "learning_rate": 1.8673125067190493e-05,
      "loss": 0.5928,
      "step": 6071
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.870450496673584,
      "learning_rate": 1.8672618094122966e-05,
      "loss": 0.5967,
      "step": 6072
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8228631019592285,
      "learning_rate": 1.867211103110652e-05,
      "loss": 0.5936,
      "step": 6073
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.767166018486023,
      "learning_rate": 1.8671603878146416e-05,
      "loss": 0.5228,
      "step": 6074
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7568747997283936,
      "learning_rate": 1.8671096635247914e-05,
      "loss": 0.5859,
      "step": 6075
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7081800699234009,
      "learning_rate": 1.8670589302416275e-05,
      "loss": 0.5431,
      "step": 6076
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0373282432556152,
      "learning_rate": 1.8670081879656757e-05,
      "loss": 0.5913,
      "step": 6077
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7271915674209595,
      "learning_rate": 1.8669574366974626e-05,
      "loss": 0.4991,
      "step": 6078
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9491620063781738,
      "learning_rate": 1.8669066764375146e-05,
      "loss": 0.644,
      "step": 6079
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0938146114349365,
      "learning_rate": 1.8668559071863584e-05,
      "loss": 0.6376,
      "step": 6080
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7198783159255981,
      "learning_rate": 1.8668051289445203e-05,
      "loss": 0.4977,
      "step": 6081
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.80216646194458,
      "learning_rate": 1.8667543417125266e-05,
      "loss": 0.5583,
      "step": 6082
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8695595264434814,
      "learning_rate": 1.8667035454909044e-05,
      "loss": 0.564,
      "step": 6083
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7005878686904907,
      "learning_rate": 1.8666527402801807e-05,
      "loss": 0.5128,
      "step": 6084
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9078221321105957,
      "learning_rate": 1.866601926080882e-05,
      "loss": 0.6471,
      "step": 6085
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.63901948928833,
      "learning_rate": 1.8665511028935353e-05,
      "loss": 0.5138,
      "step": 6086
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8400129079818726,
      "learning_rate": 1.8665002707186683e-05,
      "loss": 0.6618,
      "step": 6087
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.775724172592163,
      "learning_rate": 1.8664494295568082e-05,
      "loss": 0.4889,
      "step": 6088
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7260186672210693,
      "learning_rate": 1.8663985794084815e-05,
      "loss": 0.6267,
      "step": 6089
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7400493621826172,
      "learning_rate": 1.8663477202742163e-05,
      "loss": 0.5135,
      "step": 6090
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.720682978630066,
      "learning_rate": 1.86629685215454e-05,
      "loss": 0.5375,
      "step": 6091
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6486316919326782,
      "learning_rate": 1.8662459750499798e-05,
      "loss": 0.5055,
      "step": 6092
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8607993125915527,
      "learning_rate": 1.866195088961064e-05,
      "loss": 0.6214,
      "step": 6093
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6953799724578857,
      "learning_rate": 1.8661441938883197e-05,
      "loss": 0.533,
      "step": 6094
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.153898239135742,
      "learning_rate": 1.866093289832275e-05,
      "loss": 0.9749,
      "step": 6095
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8715088367462158,
      "learning_rate": 1.8660423767934583e-05,
      "loss": 1.0648,
      "step": 6096
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7883999347686768,
      "learning_rate": 1.865991454772397e-05,
      "loss": 0.55,
      "step": 6097
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8340328931808472,
      "learning_rate": 1.8659405237696195e-05,
      "loss": 0.584,
      "step": 6098
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8102213144302368,
      "learning_rate": 1.8658895837856543e-05,
      "loss": 0.5953,
      "step": 6099
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7668160200119019,
      "learning_rate": 1.8658386348210294e-05,
      "loss": 0.5616,
      "step": 6100
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6754801273345947,
      "learning_rate": 1.865787676876273e-05,
      "loss": 0.5119,
      "step": 6101
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7871346473693848,
      "learning_rate": 1.8657367099519143e-05,
      "loss": 0.5964,
      "step": 6102
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9647588729858398,
      "learning_rate": 1.8656857340484812e-05,
      "loss": 0.6522,
      "step": 6103
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6993962526321411,
      "learning_rate": 1.8656347491665033e-05,
      "loss": 0.5158,
      "step": 6104
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8772655725479126,
      "learning_rate": 1.865583755306508e-05,
      "loss": 0.5887,
      "step": 6105
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6720695495605469,
      "learning_rate": 1.8655327524690258e-05,
      "loss": 0.5146,
      "step": 6106
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6915881633758545,
      "learning_rate": 1.8654817406545843e-05,
      "loss": 0.5474,
      "step": 6107
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7631264925003052,
      "learning_rate": 1.865430719863714e-05,
      "loss": 0.4928,
      "step": 6108
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7614613771438599,
      "learning_rate": 1.8653796900969423e-05,
      "loss": 0.6035,
      "step": 6109
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7402310371398926,
      "learning_rate": 1.8653286513548e-05,
      "loss": 0.5347,
      "step": 6110
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.848380446434021,
      "learning_rate": 1.8652776036378157e-05,
      "loss": 0.6539,
      "step": 6111
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7335994243621826,
      "learning_rate": 1.865226546946519e-05,
      "loss": 0.5158,
      "step": 6112
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8445751667022705,
      "learning_rate": 1.8651754812814392e-05,
      "loss": 0.6414,
      "step": 6113
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8708056211471558,
      "learning_rate": 1.8651244066431065e-05,
      "loss": 0.5236,
      "step": 6114
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9014396667480469,
      "learning_rate": 1.8650733230320504e-05,
      "loss": 0.574,
      "step": 6115
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7502959966659546,
      "learning_rate": 1.8650222304488002e-05,
      "loss": 0.4858,
      "step": 6116
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8205891847610474,
      "learning_rate": 1.8649711288938865e-05,
      "loss": 0.6406,
      "step": 6117
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.798274040222168,
      "learning_rate": 1.8649200183678387e-05,
      "loss": 0.4864,
      "step": 6118
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8112376928329468,
      "learning_rate": 1.864868898871188e-05,
      "loss": 0.5818,
      "step": 6119
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.781337857246399,
      "learning_rate": 1.864817770404463e-05,
      "loss": 0.5154,
      "step": 6120
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7564818859100342,
      "learning_rate": 1.864766632968195e-05,
      "loss": 0.5611,
      "step": 6121
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9282896518707275,
      "learning_rate": 1.8647154865629143e-05,
      "loss": 0.6568,
      "step": 6122
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7732495069503784,
      "learning_rate": 1.8646643311891514e-05,
      "loss": 0.5421,
      "step": 6123
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.035830020904541,
      "learning_rate": 1.8646131668474366e-05,
      "loss": 0.5768,
      "step": 6124
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7948036193847656,
      "learning_rate": 1.864561993538301e-05,
      "loss": 0.5387,
      "step": 6125
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7564810514450073,
      "learning_rate": 1.8645108112622745e-05,
      "loss": 0.5263,
      "step": 6126
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9258143901824951,
      "learning_rate": 1.8644596200198887e-05,
      "loss": 0.619,
      "step": 6127
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7516130208969116,
      "learning_rate": 1.864408419811674e-05,
      "loss": 0.5095,
      "step": 6128
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.3103885650634766,
      "learning_rate": 1.864357210638162e-05,
      "loss": 1.0552,
      "step": 6129
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.934165120124817,
      "learning_rate": 1.8643059924998836e-05,
      "loss": 1.1317,
      "step": 6130
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.123802423477173,
      "learning_rate": 1.8642547653973696e-05,
      "loss": 0.611,
      "step": 6131
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1350061893463135,
      "learning_rate": 1.8642035293311522e-05,
      "loss": 0.6252,
      "step": 6132
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.991753339767456,
      "learning_rate": 1.864152284301762e-05,
      "loss": 0.6417,
      "step": 6133
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7420341968536377,
      "learning_rate": 1.8641010303097307e-05,
      "loss": 0.5629,
      "step": 6134
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8839223384857178,
      "learning_rate": 1.86404976735559e-05,
      "loss": 0.6094,
      "step": 6135
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9507083892822266,
      "learning_rate": 1.8639984954398714e-05,
      "loss": 0.5161,
      "step": 6136
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9956457614898682,
      "learning_rate": 1.863947214563107e-05,
      "loss": 0.5791,
      "step": 6137
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8686152696609497,
      "learning_rate": 1.8638959247258286e-05,
      "loss": 0.5089,
      "step": 6138
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.966068148612976,
      "learning_rate": 1.863844625928568e-05,
      "loss": 0.6098,
      "step": 6139
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9114097356796265,
      "learning_rate": 1.863793318171857e-05,
      "loss": 0.5299,
      "step": 6140
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8007899522781372,
      "learning_rate": 1.863742001456228e-05,
      "loss": 0.5827,
      "step": 6141
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8006367683410645,
      "learning_rate": 1.8636906757822134e-05,
      "loss": 0.5002,
      "step": 6142
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8524223566055298,
      "learning_rate": 1.8636393411503453e-05,
      "loss": 0.5409,
      "step": 6143
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.803469181060791,
      "learning_rate": 1.8635879975611564e-05,
      "loss": 0.5559,
      "step": 6144
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.815646767616272,
      "learning_rate": 1.863536645015179e-05,
      "loss": 0.5851,
      "step": 6145
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.747483730316162,
      "learning_rate": 1.8634852835129456e-05,
      "loss": 0.4781,
      "step": 6146
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8571510314941406,
      "learning_rate": 1.8634339130549887e-05,
      "loss": 0.6074,
      "step": 6147
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8174104690551758,
      "learning_rate": 1.863382533641842e-05,
      "loss": 0.5262,
      "step": 6148
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.834096074104309,
      "learning_rate": 1.8633311452740374e-05,
      "loss": 0.6,
      "step": 6149
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5873616933822632,
      "learning_rate": 1.8632797479521087e-05,
      "loss": 0.4763,
      "step": 6150
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7749415636062622,
      "learning_rate": 1.8632283416765883e-05,
      "loss": 0.5764,
      "step": 6151
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0098414421081543,
      "learning_rate": 1.8631769264480094e-05,
      "loss": 0.5886,
      "step": 6152
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9776926040649414,
      "learning_rate": 1.8631255022669057e-05,
      "loss": 0.5809,
      "step": 6153
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8069825172424316,
      "learning_rate": 1.8630740691338104e-05,
      "loss": 0.5301,
      "step": 6154
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8619366884231567,
      "learning_rate": 1.863022627049257e-05,
      "loss": 0.7176,
      "step": 6155
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6510024070739746,
      "learning_rate": 1.8629711760137788e-05,
      "loss": 0.4724,
      "step": 6156
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.847051978111267,
      "learning_rate": 1.8629197160279094e-05,
      "loss": 0.5791,
      "step": 6157
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7612749338150024,
      "learning_rate": 1.8628682470921825e-05,
      "loss": 0.4965,
      "step": 6158
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7860393524169922,
      "learning_rate": 1.8628167692071324e-05,
      "loss": 0.607,
      "step": 6159
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8203083276748657,
      "learning_rate": 1.8627652823732927e-05,
      "loss": 0.5018,
      "step": 6160
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8685702085494995,
      "learning_rate": 1.8627137865911974e-05,
      "loss": 0.5585,
      "step": 6161
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.828518033027649,
      "learning_rate": 1.8626622818613805e-05,
      "loss": 0.5112,
      "step": 6162
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8241811990737915,
      "learning_rate": 1.8626107681843762e-05,
      "loss": 0.6038,
      "step": 6163
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7210772037506104,
      "learning_rate": 1.862559245560719e-05,
      "loss": 0.5047,
      "step": 6164
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.78763747215271,
      "learning_rate": 1.862507713990943e-05,
      "loss": 0.6486,
      "step": 6165
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6759624481201172,
      "learning_rate": 1.8624561734755828e-05,
      "loss": 0.4936,
      "step": 6166
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7506308555603027,
      "learning_rate": 1.862404624015173e-05,
      "loss": 0.6086,
      "step": 6167
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7834874391555786,
      "learning_rate": 1.8623530656102482e-05,
      "loss": 0.5242,
      "step": 6168
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9356248378753662,
      "learning_rate": 1.862301498261343e-05,
      "loss": 0.6334,
      "step": 6169
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8530783653259277,
      "learning_rate": 1.8622499219689925e-05,
      "loss": 0.5181,
      "step": 6170
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.811768651008606,
      "learning_rate": 1.8621983367337314e-05,
      "loss": 0.5959,
      "step": 6171
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7424488067626953,
      "learning_rate": 1.862146742556095e-05,
      "loss": 0.5206,
      "step": 6172
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7187906503677368,
      "learning_rate": 1.862095139436618e-05,
      "loss": 0.5449,
      "step": 6173
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9237380027770996,
      "learning_rate": 1.862043527375836e-05,
      "loss": 0.5105,
      "step": 6174
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7545156478881836,
      "learning_rate": 1.861991906374284e-05,
      "loss": 0.5918,
      "step": 6175
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7566447257995605,
      "learning_rate": 1.861940276432498e-05,
      "loss": 0.5309,
      "step": 6176
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.286414623260498,
      "learning_rate": 1.8618886375510126e-05,
      "loss": 0.9548,
      "step": 6177
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8036611080169678,
      "learning_rate": 1.8618369897303637e-05,
      "loss": 1.0366,
      "step": 6178
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0823800563812256,
      "learning_rate": 1.8617853329710873e-05,
      "loss": 0.5716,
      "step": 6179
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.032773494720459,
      "learning_rate": 1.861733667273719e-05,
      "loss": 0.5461,
      "step": 6180
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.883353352546692,
      "learning_rate": 1.8616819926387943e-05,
      "loss": 0.5994,
      "step": 6181
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.721316933631897,
      "learning_rate": 1.86163030906685e-05,
      "loss": 0.5337,
      "step": 6182
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8593558073043823,
      "learning_rate": 1.8615786165584208e-05,
      "loss": 0.6051,
      "step": 6183
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8072338104248047,
      "learning_rate": 1.861526915114044e-05,
      "loss": 0.5293,
      "step": 6184
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7699674367904663,
      "learning_rate": 1.8614752047342554e-05,
      "loss": 0.5378,
      "step": 6185
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7400424480438232,
      "learning_rate": 1.8614234854195917e-05,
      "loss": 0.5121,
      "step": 6186
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7431997060775757,
      "learning_rate": 1.8613717571705885e-05,
      "loss": 0.5365,
      "step": 6187
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.878007173538208,
      "learning_rate": 1.8613200199877827e-05,
      "loss": 0.5068,
      "step": 6188
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1613266468048096,
      "learning_rate": 1.8612682738717115e-05,
      "loss": 1.0242,
      "step": 6189
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9291986227035522,
      "learning_rate": 1.8612165188229107e-05,
      "loss": 1.0485,
      "step": 6190
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1840670108795166,
      "learning_rate": 1.8611647548419175e-05,
      "loss": 0.6583,
      "step": 6191
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8945330381393433,
      "learning_rate": 1.8611129819292687e-05,
      "loss": 0.5018,
      "step": 6192
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.06709361076355,
      "learning_rate": 1.8610612000855016e-05,
      "loss": 0.6197,
      "step": 6193
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7170629501342773,
      "learning_rate": 1.8610094093111523e-05,
      "loss": 0.5114,
      "step": 6194
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.95276939868927,
      "learning_rate": 1.860957609606759e-05,
      "loss": 1.0325,
      "step": 6195
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6461613178253174,
      "learning_rate": 1.8609058009728586e-05,
      "loss": 0.9959,
      "step": 6196
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7585281133651733,
      "learning_rate": 1.8608539834099882e-05,
      "loss": 0.982,
      "step": 6197
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.542512059211731,
      "learning_rate": 1.8608021569186858e-05,
      "loss": 1.0432,
      "step": 6198
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.74835205078125,
      "learning_rate": 1.860750321499488e-05,
      "loss": 0.6156,
      "step": 6199
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.4345176219940186,
      "learning_rate": 1.8606984771529334e-05,
      "loss": 0.5167,
      "step": 6200
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1584253311157227,
      "learning_rate": 1.8606466238795592e-05,
      "loss": 0.664,
      "step": 6201
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7898378372192383,
      "learning_rate": 1.860594761679903e-05,
      "loss": 0.5192,
      "step": 6202
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.979379653930664,
      "learning_rate": 1.860542890554503e-05,
      "loss": 0.6135,
      "step": 6203
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0665571689605713,
      "learning_rate": 1.8604910105038973e-05,
      "loss": 0.5862,
      "step": 6204
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8542295694351196,
      "learning_rate": 1.8604391215286237e-05,
      "loss": 0.5348,
      "step": 6205
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8248790502548218,
      "learning_rate": 1.8603872236292207e-05,
      "loss": 0.5854,
      "step": 6206
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8632501363754272,
      "learning_rate": 1.8603353168062264e-05,
      "loss": 0.5963,
      "step": 6207
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7991528511047363,
      "learning_rate": 1.860283401060179e-05,
      "loss": 0.4858,
      "step": 6208
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.907842755317688,
      "learning_rate": 1.8602314763916167e-05,
      "loss": 0.5779,
      "step": 6209
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8089704513549805,
      "learning_rate": 1.8601795428010788e-05,
      "loss": 0.5351,
      "step": 6210
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7855148315429688,
      "learning_rate": 1.8601276002891037e-05,
      "loss": 0.6434,
      "step": 6211
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7526758909225464,
      "learning_rate": 1.8600756488562298e-05,
      "loss": 0.5469,
      "step": 6212
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8408998250961304,
      "learning_rate": 1.8600236885029958e-05,
      "loss": 0.5598,
      "step": 6213
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8063011169433594,
      "learning_rate": 1.8599717192299414e-05,
      "loss": 0.5429,
      "step": 6214
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8142634630203247,
      "learning_rate": 1.859919741037605e-05,
      "loss": 0.6537,
      "step": 6215
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.81266188621521,
      "learning_rate": 1.8598677539265253e-05,
      "loss": 0.4782,
      "step": 6216
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8101454973220825,
      "learning_rate": 1.8598157578972424e-05,
      "loss": 0.565,
      "step": 6217
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8310672044754028,
      "learning_rate": 1.8597637529502953e-05,
      "loss": 0.5534,
      "step": 6218
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.858141303062439,
      "learning_rate": 1.859711739086223e-05,
      "loss": 0.5948,
      "step": 6219
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9062087535858154,
      "learning_rate": 1.8596597163055654e-05,
      "loss": 0.5163,
      "step": 6220
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0839104652404785,
      "learning_rate": 1.8596076846088618e-05,
      "loss": 0.6652,
      "step": 6221
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8540623188018799,
      "learning_rate": 1.859555643996652e-05,
      "loss": 0.493,
      "step": 6222
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8894511461257935,
      "learning_rate": 1.859503594469475e-05,
      "loss": 0.6606,
      "step": 6223
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7578014135360718,
      "learning_rate": 1.8594515360278722e-05,
      "loss": 0.531,
      "step": 6224
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.736937403678894,
      "learning_rate": 1.8593994686723822e-05,
      "loss": 0.5796,
      "step": 6225
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8384617567062378,
      "learning_rate": 1.8593473924035456e-05,
      "loss": 0.5439,
      "step": 6226
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9034472703933716,
      "learning_rate": 1.8592953072219026e-05,
      "loss": 0.5454,
      "step": 6227
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8217248916625977,
      "learning_rate": 1.8592432131279928e-05,
      "loss": 0.547,
      "step": 6228
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.771028995513916,
      "learning_rate": 1.8591911101223572e-05,
      "loss": 0.586,
      "step": 6229
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9301679134368896,
      "learning_rate": 1.8591389982055355e-05,
      "loss": 0.6026,
      "step": 6230
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8064665794372559,
      "learning_rate": 1.859086877378069e-05,
      "loss": 0.5681,
      "step": 6231
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7372554540634155,
      "learning_rate": 1.8590347476404976e-05,
      "loss": 0.5427,
      "step": 6232
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.864417552947998,
      "learning_rate": 1.858982608993362e-05,
      "loss": 0.6141,
      "step": 6233
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6021696329116821,
      "learning_rate": 1.8589304614372034e-05,
      "loss": 0.4481,
      "step": 6234
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7213754653930664,
      "learning_rate": 1.8588783049725622e-05,
      "loss": 0.6036,
      "step": 6235
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6437538862228394,
      "learning_rate": 1.8588261395999798e-05,
      "loss": 0.4835,
      "step": 6236
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2686734199523926,
      "learning_rate": 1.858773965319997e-05,
      "loss": 1.0421,
      "step": 6237
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0227644443511963,
      "learning_rate": 1.858721782133155e-05,
      "loss": 1.1393,
      "step": 6238
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0138649940490723,
      "learning_rate": 1.8586695900399945e-05,
      "loss": 0.6097,
      "step": 6239
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9803030490875244,
      "learning_rate": 1.8586173890410576e-05,
      "loss": 0.5477,
      "step": 6240
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7851064205169678,
      "learning_rate": 1.8585651791368856e-05,
      "loss": 0.5321,
      "step": 6241
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.67594575881958,
      "learning_rate": 1.858512960328019e-05,
      "loss": 0.5091,
      "step": 6242
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.837991714477539,
      "learning_rate": 1.858460732615001e-05,
      "loss": 0.6493,
      "step": 6243
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7266008853912354,
      "learning_rate": 1.858408495998372e-05,
      "loss": 0.4948,
      "step": 6244
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8405989408493042,
      "learning_rate": 1.8583562504786746e-05,
      "loss": 0.6256,
      "step": 6245
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7914658784866333,
      "learning_rate": 1.85830399605645e-05,
      "loss": 0.5591,
      "step": 6246
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9456950426101685,
      "learning_rate": 1.8582517327322408e-05,
      "loss": 0.67,
      "step": 6247
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7901360988616943,
      "learning_rate": 1.8581994605065883e-05,
      "loss": 0.5045,
      "step": 6248
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8487807512283325,
      "learning_rate": 1.858147179380035e-05,
      "loss": 0.613,
      "step": 6249
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7863868474960327,
      "learning_rate": 1.8580948893531236e-05,
      "loss": 0.4965,
      "step": 6250
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6996850967407227,
      "learning_rate": 1.858042590426396e-05,
      "loss": 0.5875,
      "step": 6251
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.751396894454956,
      "learning_rate": 1.8579902826003945e-05,
      "loss": 0.4984,
      "step": 6252
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7676644325256348,
      "learning_rate": 1.8579379658756618e-05,
      "loss": 0.6067,
      "step": 6253
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7010776996612549,
      "learning_rate": 1.8578856402527403e-05,
      "loss": 0.5277,
      "step": 6254
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7112339735031128,
      "learning_rate": 1.8578333057321733e-05,
      "loss": 0.5536,
      "step": 6255
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7692950963974,
      "learning_rate": 1.8577809623145027e-05,
      "loss": 0.576,
      "step": 6256
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7798569202423096,
      "learning_rate": 1.8577286100002723e-05,
      "loss": 0.5645,
      "step": 6257
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8281558752059937,
      "learning_rate": 1.8576762487900245e-05,
      "loss": 0.6203,
      "step": 6258
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9016425609588623,
      "learning_rate": 1.857623878684302e-05,
      "loss": 0.5795,
      "step": 6259
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7975831031799316,
      "learning_rate": 1.8575714996836495e-05,
      "loss": 0.5077,
      "step": 6260
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.75014328956604,
      "learning_rate": 1.857519111788608e-05,
      "loss": 0.6124,
      "step": 6261
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7989873886108398,
      "learning_rate": 1.857466714999723e-05,
      "loss": 0.4951,
      "step": 6262
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9081028699874878,
      "learning_rate": 1.857414309317537e-05,
      "loss": 0.6546,
      "step": 6263
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7927343845367432,
      "learning_rate": 1.8573618947425932e-05,
      "loss": 0.5337,
      "step": 6264
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6831616163253784,
      "learning_rate": 1.8573094712754353e-05,
      "loss": 0.5561,
      "step": 6265
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7234665155410767,
      "learning_rate": 1.8572570389166076e-05,
      "loss": 0.5132,
      "step": 6266
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.16776704788208,
      "learning_rate": 1.8572045976666536e-05,
      "loss": 1.0335,
      "step": 6267
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8832054138183594,
      "learning_rate": 1.857152147526117e-05,
      "loss": 1.0819,
      "step": 6268
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9947092533111572,
      "learning_rate": 1.8570996884955423e-05,
      "loss": 0.5717,
      "step": 6269
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8936182260513306,
      "learning_rate": 1.857047220575473e-05,
      "loss": 0.5117,
      "step": 6270
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.295680522918701,
      "learning_rate": 1.8569947437664538e-05,
      "loss": 0.6244,
      "step": 6271
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6701929569244385,
      "learning_rate": 1.8569422580690283e-05,
      "loss": 0.4808,
      "step": 6272
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0559616088867188,
      "learning_rate": 1.8568897634837414e-05,
      "loss": 0.6096,
      "step": 6273
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.753788709640503,
      "learning_rate": 1.8568372600111378e-05,
      "loss": 0.5034,
      "step": 6274
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7734252214431763,
      "learning_rate": 1.8567847476517613e-05,
      "loss": 0.5281,
      "step": 6275
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7872436046600342,
      "learning_rate": 1.856732226406157e-05,
      "loss": 0.5641,
      "step": 6276
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2869274616241455,
      "learning_rate": 1.8566796962748692e-05,
      "loss": 1.0106,
      "step": 6277
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8919697999954224,
      "learning_rate": 1.8566271572584436e-05,
      "loss": 1.0926,
      "step": 6278
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.166612386703491,
      "learning_rate": 1.856574609357424e-05,
      "loss": 0.5708,
      "step": 6279
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.14943528175354,
      "learning_rate": 1.8565220525723563e-05,
      "loss": 0.6142,
      "step": 6280
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.947044849395752,
      "learning_rate": 1.8564694869037852e-05,
      "loss": 0.5543,
      "step": 6281
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8307390213012695,
      "learning_rate": 1.8564169123522563e-05,
      "loss": 0.5236,
      "step": 6282
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.812166690826416,
      "learning_rate": 1.856364328918314e-05,
      "loss": 0.6252,
      "step": 6283
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8195862770080566,
      "learning_rate": 1.8563117366025046e-05,
      "loss": 0.4986,
      "step": 6284
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0661659240722656,
      "learning_rate": 1.856259135405373e-05,
      "loss": 0.621,
      "step": 6285
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7971665859222412,
      "learning_rate": 1.856206525327465e-05,
      "loss": 0.491,
      "step": 6286
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8979161977767944,
      "learning_rate": 1.8561539063693258e-05,
      "loss": 0.6307,
      "step": 6287
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.772118330001831,
      "learning_rate": 1.8561012785315023e-05,
      "loss": 0.5417,
      "step": 6288
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.969944953918457,
      "learning_rate": 1.8560486418145388e-05,
      "loss": 0.5754,
      "step": 6289
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0173418521881104,
      "learning_rate": 1.8559959962189827e-05,
      "loss": 0.5733,
      "step": 6290
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8627017736434937,
      "learning_rate": 1.855943341745379e-05,
      "loss": 0.649,
      "step": 6291
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7333511114120483,
      "learning_rate": 1.8558906783942738e-05,
      "loss": 0.4833,
      "step": 6292
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7690061330795288,
      "learning_rate": 1.855838006166214e-05,
      "loss": 0.525,
      "step": 6293
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7605973482131958,
      "learning_rate": 1.855785325061745e-05,
      "loss": 0.503,
      "step": 6294
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.592345952987671,
      "learning_rate": 1.8557326350814145e-05,
      "loss": 0.5448,
      "step": 6295
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.755557656288147,
      "learning_rate": 1.8556799362257678e-05,
      "loss": 0.5171,
      "step": 6296
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8422647714614868,
      "learning_rate": 1.855627228495352e-05,
      "loss": 0.5786,
      "step": 6297
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6965909004211426,
      "learning_rate": 1.8555745118907135e-05,
      "loss": 0.5234,
      "step": 6298
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1848554611206055,
      "learning_rate": 1.855521786412399e-05,
      "loss": 1.0049,
      "step": 6299
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8287793397903442,
      "learning_rate": 1.8554690520609556e-05,
      "loss": 1.0873,
      "step": 6300
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0656347274780273,
      "learning_rate": 1.8554163088369302e-05,
      "loss": 0.6019,
      "step": 6301
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0930304527282715,
      "learning_rate": 1.85536355674087e-05,
      "loss": 0.5856,
      "step": 6302
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8399521112442017,
      "learning_rate": 1.8553107957733217e-05,
      "loss": 0.9876,
      "step": 6303
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5745530128479004,
      "learning_rate": 1.855258025934833e-05,
      "loss": 1.0427,
      "step": 6304
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0222907066345215,
      "learning_rate": 1.8552052472259507e-05,
      "loss": 0.6067,
      "step": 6305
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.679851770401001,
      "learning_rate": 1.8551524596472225e-05,
      "loss": 0.4728,
      "step": 6306
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.798745036125183,
      "learning_rate": 1.855099663199196e-05,
      "loss": 0.5509,
      "step": 6307
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7302219867706299,
      "learning_rate": 1.8550468578824183e-05,
      "loss": 0.5079,
      "step": 6308
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8133118152618408,
      "learning_rate": 1.854994043697438e-05,
      "loss": 0.539,
      "step": 6309
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7779028415679932,
      "learning_rate": 1.8549412206448013e-05,
      "loss": 0.5573,
      "step": 6310
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6873805522918701,
      "learning_rate": 1.8548883887250578e-05,
      "loss": 0.5394,
      "step": 6311
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8952953815460205,
      "learning_rate": 1.8548355479387546e-05,
      "loss": 0.5685,
      "step": 6312
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6266170740127563,
      "learning_rate": 1.8547826982864394e-05,
      "loss": 0.525,
      "step": 6313
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6476930379867554,
      "learning_rate": 1.854729839768661e-05,
      "loss": 0.528,
      "step": 6314
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.776459813117981,
      "learning_rate": 1.8546769723859674e-05,
      "loss": 0.5552,
      "step": 6315
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7269083261489868,
      "learning_rate": 1.8546240961389068e-05,
      "loss": 0.5117,
      "step": 6316
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9406810998916626,
      "learning_rate": 1.854571211028028e-05,
      "loss": 0.5719,
      "step": 6317
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7319661378860474,
      "learning_rate": 1.854518317053879e-05,
      "loss": 0.4897,
      "step": 6318
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9917789697647095,
      "learning_rate": 1.8544654142170085e-05,
      "loss": 0.5722,
      "step": 6319
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7569055557250977,
      "learning_rate": 1.8544125025179653e-05,
      "loss": 0.5203,
      "step": 6320
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9008660316467285,
      "learning_rate": 1.8543595819572982e-05,
      "loss": 0.649,
      "step": 6321
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6494417190551758,
      "learning_rate": 1.8543066525355562e-05,
      "loss": 0.4724,
      "step": 6322
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7492449283599854,
      "learning_rate": 1.8542537142532882e-05,
      "loss": 0.5472,
      "step": 6323
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.926477313041687,
      "learning_rate": 1.854200767111043e-05,
      "loss": 0.5706,
      "step": 6324
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8761298656463623,
      "learning_rate": 1.85414781110937e-05,
      "loss": 0.656,
      "step": 6325
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.638091802597046,
      "learning_rate": 1.8540948462488177e-05,
      "loss": 0.5188,
      "step": 6326
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8617980480194092,
      "learning_rate": 1.854041872529937e-05,
      "loss": 0.6209,
      "step": 6327
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7289457321166992,
      "learning_rate": 1.853988889953276e-05,
      "loss": 0.4893,
      "step": 6328
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7502362728118896,
      "learning_rate": 1.8539358985193846e-05,
      "loss": 0.5494,
      "step": 6329
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7260229587554932,
      "learning_rate": 1.8538828982288123e-05,
      "loss": 0.4884,
      "step": 6330
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7586424350738525,
      "learning_rate": 1.853829889082109e-05,
      "loss": 0.6062,
      "step": 6331
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.740017294883728,
      "learning_rate": 1.8537768710798245e-05,
      "loss": 0.5244,
      "step": 6332
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1030402183532715,
      "learning_rate": 1.8537238442225083e-05,
      "loss": 0.6436,
      "step": 6333
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7644076347351074,
      "learning_rate": 1.8536708085107113e-05,
      "loss": 0.493,
      "step": 6334
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.858115315437317,
      "learning_rate": 1.8536177639449825e-05,
      "loss": 0.595,
      "step": 6335
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9231899976730347,
      "learning_rate": 1.8535647105258725e-05,
      "loss": 0.5765,
      "step": 6336
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.143651008605957,
      "learning_rate": 1.8535116482539317e-05,
      "loss": 0.6354,
      "step": 6337
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8982198238372803,
      "learning_rate": 1.85345857712971e-05,
      "loss": 0.5274,
      "step": 6338
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8085030317306519,
      "learning_rate": 1.8534054971537582e-05,
      "loss": 0.5736,
      "step": 6339
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8559199571609497,
      "learning_rate": 1.853352408326627e-05,
      "loss": 0.5673,
      "step": 6340
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.869494915008545,
      "learning_rate": 1.8532993106488665e-05,
      "loss": 0.6018,
      "step": 6341
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6980243921279907,
      "learning_rate": 1.8532462041210276e-05,
      "loss": 0.5128,
      "step": 6342
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7156577110290527,
      "learning_rate": 1.8531930887436612e-05,
      "loss": 0.5606,
      "step": 6343
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7713449001312256,
      "learning_rate": 1.8531399645173182e-05,
      "loss": 0.5613,
      "step": 6344
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7207236289978027,
      "learning_rate": 1.8530868314425493e-05,
      "loss": 0.5673,
      "step": 6345
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6406935453414917,
      "learning_rate": 1.8530336895199062e-05,
      "loss": 0.5053,
      "step": 6346
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7877936363220215,
      "learning_rate": 1.8529805387499396e-05,
      "loss": 0.5565,
      "step": 6347
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.707390546798706,
      "learning_rate": 1.8529273791332005e-05,
      "loss": 0.4921,
      "step": 6348
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7516627311706543,
      "learning_rate": 1.852874210670241e-05,
      "loss": 0.5977,
      "step": 6349
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.756730318069458,
      "learning_rate": 1.8528210333616117e-05,
      "loss": 0.529,
      "step": 6350
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8110065460205078,
      "learning_rate": 1.8527678472078646e-05,
      "loss": 0.5793,
      "step": 6351
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.68502938747406,
      "learning_rate": 1.8527146522095517e-05,
      "loss": 0.508,
      "step": 6352
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.678310751914978,
      "learning_rate": 1.852661448367224e-05,
      "loss": 0.5348,
      "step": 6353
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7806305885314941,
      "learning_rate": 1.8526082356814336e-05,
      "loss": 0.53,
      "step": 6354
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9208084344863892,
      "learning_rate": 1.8525550141527324e-05,
      "loss": 0.5441,
      "step": 6355
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.697717547416687,
      "learning_rate": 1.8525017837816723e-05,
      "loss": 0.5637,
      "step": 6356
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6879686117172241,
      "learning_rate": 1.8524485445688057e-05,
      "loss": 0.564,
      "step": 6357
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8792428970336914,
      "learning_rate": 1.8523952965146846e-05,
      "loss": 0.5899,
      "step": 6358
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.743150234222412,
      "learning_rate": 1.8523420396198612e-05,
      "loss": 0.5915,
      "step": 6359
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8031363487243652,
      "learning_rate": 1.8522887738848876e-05,
      "loss": 0.5677,
      "step": 6360
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8312410116195679,
      "learning_rate": 1.8522354993103167e-05,
      "loss": 0.6017,
      "step": 6361
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6625248193740845,
      "learning_rate": 1.852182215896701e-05,
      "loss": 0.4921,
      "step": 6362
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8310765027999878,
      "learning_rate": 1.852128923644593e-05,
      "loss": 0.6043,
      "step": 6363
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7189236879348755,
      "learning_rate": 1.8520756225545453e-05,
      "loss": 0.4679,
      "step": 6364
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.702262282371521,
      "learning_rate": 1.8520223126271107e-05,
      "loss": 0.5568,
      "step": 6365
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.681397795677185,
      "learning_rate": 1.8519689938628425e-05,
      "loss": 0.4781,
      "step": 6366
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.816190481185913,
      "learning_rate": 1.8519156662622938e-05,
      "loss": 0.5362,
      "step": 6367
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.83055579662323,
      "learning_rate": 1.851862329826017e-05,
      "loss": 0.5824,
      "step": 6368
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.738305687904358,
      "learning_rate": 1.8518089845545654e-05,
      "loss": 0.5763,
      "step": 6369
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7120963335037231,
      "learning_rate": 1.851755630448493e-05,
      "loss": 0.534,
      "step": 6370
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7508610486984253,
      "learning_rate": 1.8517022675083523e-05,
      "loss": 0.6006,
      "step": 6371
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7256032228469849,
      "learning_rate": 1.8516488957346973e-05,
      "loss": 0.5082,
      "step": 6372
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9970252513885498,
      "learning_rate": 1.8515955151280813e-05,
      "loss": 0.707,
      "step": 6373
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6693636178970337,
      "learning_rate": 1.851542125689058e-05,
      "loss": 0.4905,
      "step": 6374
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6334538459777832,
      "learning_rate": 1.8514887274181816e-05,
      "loss": 0.5534,
      "step": 6375
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.698290467262268,
      "learning_rate": 1.851435320316005e-05,
      "loss": 0.5422,
      "step": 6376
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6544452905654907,
      "learning_rate": 1.851381904383083e-05,
      "loss": 0.5297,
      "step": 6377
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7931594848632812,
      "learning_rate": 1.851328479619969e-05,
      "loss": 0.5872,
      "step": 6378
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8491476774215698,
      "learning_rate": 1.8512750460272174e-05,
      "loss": 0.5683,
      "step": 6379
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6164090633392334,
      "learning_rate": 1.851221603605382e-05,
      "loss": 0.4794,
      "step": 6380
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0117552280426025,
      "learning_rate": 1.851168152355018e-05,
      "loss": 0.6125,
      "step": 6381
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.764952301979065,
      "learning_rate": 1.851114692276679e-05,
      "loss": 0.5199,
      "step": 6382
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7459321022033691,
      "learning_rate": 1.8510612233709192e-05,
      "loss": 0.5477,
      "step": 6383
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6798193454742432,
      "learning_rate": 1.851007745638294e-05,
      "loss": 0.4998,
      "step": 6384
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7383545637130737,
      "learning_rate": 1.8509542590793576e-05,
      "loss": 0.5362,
      "step": 6385
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7188235521316528,
      "learning_rate": 1.8509007636946646e-05,
      "loss": 0.5372,
      "step": 6386
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9783194065093994,
      "learning_rate": 1.8508472594847702e-05,
      "loss": 0.6632,
      "step": 6387
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.658722996711731,
      "learning_rate": 1.8507937464502288e-05,
      "loss": 0.4616,
      "step": 6388
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6897648572921753,
      "learning_rate": 1.8507402245915962e-05,
      "loss": 0.5424,
      "step": 6389
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.791247010231018,
      "learning_rate": 1.850686693909427e-05,
      "loss": 0.563,
      "step": 6390
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.710250735282898,
      "learning_rate": 1.8506331544042764e-05,
      "loss": 0.5718,
      "step": 6391
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7389613389968872,
      "learning_rate": 1.8505796060766995e-05,
      "loss": 0.5293,
      "step": 6392
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9056044816970825,
      "learning_rate": 1.8505260489272523e-05,
      "loss": 0.501,
      "step": 6393
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9370050430297852,
      "learning_rate": 1.85047248295649e-05,
      "loss": 0.5746,
      "step": 6394
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.697826623916626,
      "learning_rate": 1.8504189081649678e-05,
      "loss": 0.5407,
      "step": 6395
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8500018119812012,
      "learning_rate": 1.8503653245532417e-05,
      "loss": 0.5659,
      "step": 6396
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6615968942642212,
      "learning_rate": 1.8503117321218674e-05,
      "loss": 0.5395,
      "step": 6397
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7694281339645386,
      "learning_rate": 1.8502581308714008e-05,
      "loss": 0.5409,
      "step": 6398
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9172600507736206,
      "learning_rate": 1.8502045208023977e-05,
      "loss": 0.5853,
      "step": 6399
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9478288888931274,
      "learning_rate": 1.850150901915414e-05,
      "loss": 0.5978,
      "step": 6400
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6554663181304932,
      "learning_rate": 1.850097274211006e-05,
      "loss": 0.5647,
      "step": 6401
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6126841306686401,
      "learning_rate": 1.85004363768973e-05,
      "loss": 0.5065,
      "step": 6402
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6638680696487427,
      "learning_rate": 1.849989992352142e-05,
      "loss": 0.5318,
      "step": 6403
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8705896139144897,
      "learning_rate": 1.849936338198799e-05,
      "loss": 0.568,
      "step": 6404
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.690563440322876,
      "learning_rate": 1.849882675230257e-05,
      "loss": 0.5374,
      "step": 6405
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.795609474182129,
      "learning_rate": 1.8498290034470722e-05,
      "loss": 0.5069,
      "step": 6406
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9601757526397705,
      "learning_rate": 1.8497753228498023e-05,
      "loss": 0.6318,
      "step": 6407
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6852099895477295,
      "learning_rate": 1.849721633439003e-05,
      "loss": 0.4972,
      "step": 6408
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7476730346679688,
      "learning_rate": 1.8496679352152316e-05,
      "loss": 0.5439,
      "step": 6409
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8058645725250244,
      "learning_rate": 1.849614228179045e-05,
      "loss": 0.5586,
      "step": 6410
    },
    {
      "epoch": 0.2,
      "grad_norm": 12.144904136657715,
      "learning_rate": 1.8495605123310003e-05,
      "loss": 0.5843,
      "step": 6411
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7659814357757568,
      "learning_rate": 1.8495067876716544e-05,
      "loss": 0.5235,
      "step": 6412
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8251724243164062,
      "learning_rate": 1.849453054201565e-05,
      "loss": 0.5168,
      "step": 6413
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6953842639923096,
      "learning_rate": 1.849399311921289e-05,
      "loss": 0.4874,
      "step": 6414
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.819895625114441,
      "learning_rate": 1.849345560831384e-05,
      "loss": 0.625,
      "step": 6415
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7414053678512573,
      "learning_rate": 1.8492918009324074e-05,
      "loss": 0.5503,
      "step": 6416
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7279797792434692,
      "learning_rate": 1.8492380322249163e-05,
      "loss": 0.5029,
      "step": 6417
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.726887822151184,
      "learning_rate": 1.8491842547094694e-05,
      "loss": 0.4926,
      "step": 6418
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7461093664169312,
      "learning_rate": 1.8491304683866234e-05,
      "loss": 0.5629,
      "step": 6419
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6406604051589966,
      "learning_rate": 1.8490766732569366e-05,
      "loss": 0.5103,
      "step": 6420
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7360464334487915,
      "learning_rate": 1.849022869320967e-05,
      "loss": 0.5453,
      "step": 6421
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6743770837783813,
      "learning_rate": 1.848969056579273e-05,
      "loss": 0.5206,
      "step": 6422
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8902053833007812,
      "learning_rate": 1.848915235032412e-05,
      "loss": 0.602,
      "step": 6423
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8669424057006836,
      "learning_rate": 1.8488614046809425e-05,
      "loss": 0.5394,
      "step": 6424
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8534690141677856,
      "learning_rate": 1.848807565525423e-05,
      "loss": 0.589,
      "step": 6425
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.735037088394165,
      "learning_rate": 1.8487537175664113e-05,
      "loss": 0.5056,
      "step": 6426
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6911189556121826,
      "learning_rate": 1.8486998608044667e-05,
      "loss": 0.5485,
      "step": 6427
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.703367829322815,
      "learning_rate": 1.8486459952401472e-05,
      "loss": 0.5106,
      "step": 6428
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6342374086380005,
      "learning_rate": 1.848592120874012e-05,
      "loss": 0.5648,
      "step": 6429
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6992340087890625,
      "learning_rate": 1.8485382377066195e-05,
      "loss": 0.5219,
      "step": 6430
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6700714826583862,
      "learning_rate": 1.8484843457385283e-05,
      "loss": 0.5678,
      "step": 6431
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7192586660385132,
      "learning_rate": 1.8484304449702978e-05,
      "loss": 0.5398,
      "step": 6432
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7212634086608887,
      "learning_rate": 1.848376535402487e-05,
      "loss": 0.5517,
      "step": 6433
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7320644855499268,
      "learning_rate": 1.8483226170356548e-05,
      "loss": 0.5073,
      "step": 6434
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8222050666809082,
      "learning_rate": 1.84826868987036e-05,
      "loss": 0.5816,
      "step": 6435
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7917934656143188,
      "learning_rate": 1.848214753907163e-05,
      "loss": 0.5147,
      "step": 6436
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8074755668640137,
      "learning_rate": 1.8481608091466227e-05,
      "loss": 0.645,
      "step": 6437
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7280066013336182,
      "learning_rate": 1.8481068555892986e-05,
      "loss": 0.4987,
      "step": 6438
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8381619453430176,
      "learning_rate": 1.84805289323575e-05,
      "loss": 0.5948,
      "step": 6439
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.701295018196106,
      "learning_rate": 1.8479989220865365e-05,
      "loss": 0.5093,
      "step": 6440
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.942075252532959,
      "learning_rate": 1.8479449421422187e-05,
      "loss": 0.6327,
      "step": 6441
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7164584398269653,
      "learning_rate": 1.8478909534033556e-05,
      "loss": 0.4862,
      "step": 6442
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7551590204238892,
      "learning_rate": 1.8478369558705074e-05,
      "loss": 0.5036,
      "step": 6443
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8364499807357788,
      "learning_rate": 1.8477829495442345e-05,
      "loss": 0.5787,
      "step": 6444
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.785453200340271,
      "learning_rate": 1.8477289344250964e-05,
      "loss": 0.599,
      "step": 6445
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8127201795578003,
      "learning_rate": 1.847674910513654e-05,
      "loss": 0.5265,
      "step": 6446
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8765212297439575,
      "learning_rate": 1.8476208778104672e-05,
      "loss": 0.5659,
      "step": 6447
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.723730206489563,
      "learning_rate": 1.8475668363160964e-05,
      "loss": 0.4961,
      "step": 6448
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8127082586288452,
      "learning_rate": 1.847512786031102e-05,
      "loss": 0.5623,
      "step": 6449
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7196056842803955,
      "learning_rate": 1.847458726956045e-05,
      "loss": 0.5057,
      "step": 6450
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7591588497161865,
      "learning_rate": 1.8474046590914862e-05,
      "loss": 0.5594,
      "step": 6451
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.753045916557312,
      "learning_rate": 1.8473505824379853e-05,
      "loss": 0.4914,
      "step": 6452
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.839727520942688,
      "learning_rate": 1.8472964969961042e-05,
      "loss": 0.5982,
      "step": 6453
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.731005072593689,
      "learning_rate": 1.8472424027664037e-05,
      "loss": 0.4868,
      "step": 6454
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7707959413528442,
      "learning_rate": 1.8471882997494445e-05,
      "loss": 0.5261,
      "step": 6455
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.793653964996338,
      "learning_rate": 1.8471341879457876e-05,
      "loss": 0.5842,
      "step": 6456
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8264893293380737,
      "learning_rate": 1.847080067355995e-05,
      "loss": 0.6208,
      "step": 6457
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8075047731399536,
      "learning_rate": 1.8470259379806273e-05,
      "loss": 0.5178,
      "step": 6458
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.714978814125061,
      "learning_rate": 1.8469717998202464e-05,
      "loss": 0.5587,
      "step": 6459
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.708571434020996,
      "learning_rate": 1.8469176528754134e-05,
      "loss": 0.5316,
      "step": 6460
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7682900428771973,
      "learning_rate": 1.8468634971466903e-05,
      "loss": 0.5788,
      "step": 6461
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6856805086135864,
      "learning_rate": 1.846809332634638e-05,
      "loss": 0.4955,
      "step": 6462
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7565333843231201,
      "learning_rate": 1.8467551593398193e-05,
      "loss": 0.614,
      "step": 6463
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8144909143447876,
      "learning_rate": 1.846700977262795e-05,
      "loss": 0.5697,
      "step": 6464
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7307935953140259,
      "learning_rate": 1.846646786404128e-05,
      "loss": 0.6047,
      "step": 6465
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6515547037124634,
      "learning_rate": 1.84659258676438e-05,
      "loss": 0.5703,
      "step": 6466
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7202515602111816,
      "learning_rate": 1.8465383783441133e-05,
      "loss": 0.5682,
      "step": 6467
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6730929613113403,
      "learning_rate": 1.8464841611438897e-05,
      "loss": 0.4967,
      "step": 6468
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9696645736694336,
      "learning_rate": 1.8464299351642715e-05,
      "loss": 0.5831,
      "step": 6469
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6529699563980103,
      "learning_rate": 1.846375700405822e-05,
      "loss": 0.4922,
      "step": 6470
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8046159744262695,
      "learning_rate": 1.8463214568691024e-05,
      "loss": 0.5766,
      "step": 6471
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7021623849868774,
      "learning_rate": 1.8462672045546763e-05,
      "loss": 0.4831,
      "step": 6472
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7844606637954712,
      "learning_rate": 1.846212943463106e-05,
      "loss": 0.5568,
      "step": 6473
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8197060823440552,
      "learning_rate": 1.846158673594954e-05,
      "loss": 0.5767,
      "step": 6474
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.83323335647583,
      "learning_rate": 1.846104394950784e-05,
      "loss": 0.5811,
      "step": 6475
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7696894407272339,
      "learning_rate": 1.8460501075311583e-05,
      "loss": 0.5119,
      "step": 6476
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8479470014572144,
      "learning_rate": 1.84599581133664e-05,
      "loss": 0.5809,
      "step": 6477
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6880265474319458,
      "learning_rate": 1.845941506367792e-05,
      "loss": 0.5095,
      "step": 6478
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7330870628356934,
      "learning_rate": 1.8458871926251786e-05,
      "loss": 0.5632,
      "step": 6479
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7818632125854492,
      "learning_rate": 1.8458328701093618e-05,
      "loss": 0.5127,
      "step": 6480
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8993390798568726,
      "learning_rate": 1.8457785388209054e-05,
      "loss": 0.6249,
      "step": 6481
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.771506905555725,
      "learning_rate": 1.8457241987603735e-05,
      "loss": 0.5147,
      "step": 6482
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.707171082496643,
      "learning_rate": 1.8456698499283293e-05,
      "loss": 0.5682,
      "step": 6483
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6493343114852905,
      "learning_rate": 1.845615492325336e-05,
      "loss": 0.4916,
      "step": 6484
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7739239931106567,
      "learning_rate": 1.8455611259519582e-05,
      "loss": 0.554,
      "step": 6485
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7186279296875,
      "learning_rate": 1.8455067508087596e-05,
      "loss": 0.5611,
      "step": 6486
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8294202089309692,
      "learning_rate": 1.8454523668963037e-05,
      "loss": 0.6449,
      "step": 6487
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6830801963806152,
      "learning_rate": 1.8453979742151548e-05,
      "loss": 0.4708,
      "step": 6488
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8075965642929077,
      "learning_rate": 1.8453435727658772e-05,
      "loss": 0.5888,
      "step": 6489
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7044007778167725,
      "learning_rate": 1.8452891625490348e-05,
      "loss": 0.5449,
      "step": 6490
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7052984237670898,
      "learning_rate": 1.845234743565192e-05,
      "loss": 0.5336,
      "step": 6491
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7489858865737915,
      "learning_rate": 1.8451803158149135e-05,
      "loss": 0.5025,
      "step": 6492
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7581379413604736,
      "learning_rate": 1.8451258792987636e-05,
      "loss": 0.6074,
      "step": 6493
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6582273244857788,
      "learning_rate": 1.845071434017307e-05,
      "loss": 0.4612,
      "step": 6494
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.746847152709961,
      "learning_rate": 1.845016979971108e-05,
      "loss": 0.5384,
      "step": 6495
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.710450530052185,
      "learning_rate": 1.844962517160732e-05,
      "loss": 0.5374,
      "step": 6496
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7917805910110474,
      "learning_rate": 1.8449080455867435e-05,
      "loss": 0.6047,
      "step": 6497
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6693291664123535,
      "learning_rate": 1.8448535652497073e-05,
      "loss": 0.5591,
      "step": 6498
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5907862186431885,
      "learning_rate": 1.8447990761501887e-05,
      "loss": 0.5334,
      "step": 6499
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.937689185142517,
      "learning_rate": 1.8447445782887527e-05,
      "loss": 0.5787,
      "step": 6500
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6348073482513428,
      "learning_rate": 1.844690071665965e-05,
      "loss": 0.5628,
      "step": 6501
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.811369776725769,
      "learning_rate": 1.84463555628239e-05,
      "loss": 0.5222,
      "step": 6502
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6959880590438843,
      "learning_rate": 1.844581032138594e-05,
      "loss": 0.5492,
      "step": 6503
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.745319128036499,
      "learning_rate": 1.844526499235142e-05,
      "loss": 0.4766,
      "step": 6504
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8065972328186035,
      "learning_rate": 1.8444719575725995e-05,
      "loss": 0.5727,
      "step": 6505
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7254705429077148,
      "learning_rate": 1.8444174071515325e-05,
      "loss": 0.6016,
      "step": 6506
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9088804721832275,
      "learning_rate": 1.8443628479725072e-05,
      "loss": 0.6583,
      "step": 6507
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7426751852035522,
      "learning_rate": 1.8443082800360887e-05,
      "loss": 0.4939,
      "step": 6508
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7601665258407593,
      "learning_rate": 1.844253703342843e-05,
      "loss": 0.5348,
      "step": 6509
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.611708402633667,
      "learning_rate": 1.8441991178933365e-05,
      "loss": 0.4845,
      "step": 6510
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8306719064712524,
      "learning_rate": 1.844144523688135e-05,
      "loss": 0.6292,
      "step": 6511
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6628198623657227,
      "learning_rate": 1.8440899207278055e-05,
      "loss": 0.5108,
      "step": 6512
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8483788967132568,
      "learning_rate": 1.8440353090129132e-05,
      "loss": 0.5517,
      "step": 6513
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7323664426803589,
      "learning_rate": 1.8439806885440252e-05,
      "loss": 0.5476,
      "step": 6514
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7879835367202759,
      "learning_rate": 1.843926059321708e-05,
      "loss": 0.6002,
      "step": 6515
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7316566705703735,
      "learning_rate": 1.843871421346528e-05,
      "loss": 0.5311,
      "step": 6516
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7901597023010254,
      "learning_rate": 1.843816774619052e-05,
      "loss": 0.6688,
      "step": 6517
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6563453674316406,
      "learning_rate": 1.8437621191398464e-05,
      "loss": 0.4701,
      "step": 6518
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7723056077957153,
      "learning_rate": 1.843707454909479e-05,
      "loss": 0.5784,
      "step": 6519
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.704876184463501,
      "learning_rate": 1.8436527819285153e-05,
      "loss": 0.526,
      "step": 6520
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7913048267364502,
      "learning_rate": 1.843598100197524e-05,
      "loss": 0.6516,
      "step": 6521
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.74068284034729,
      "learning_rate": 1.843543409717071e-05,
      "loss": 0.5394,
      "step": 6522
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7628319263458252,
      "learning_rate": 1.843488710487724e-05,
      "loss": 0.5847,
      "step": 6523
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5842877626419067,
      "learning_rate": 1.8434340025100504e-05,
      "loss": 0.4833,
      "step": 6524
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.903956413269043,
      "learning_rate": 1.8433792857846174e-05,
      "loss": 0.6158,
      "step": 6525
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6793264150619507,
      "learning_rate": 1.8433245603119925e-05,
      "loss": 0.5215,
      "step": 6526
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8419193029403687,
      "learning_rate": 1.8432698260927434e-05,
      "loss": 0.6041,
      "step": 6527
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7045519351959229,
      "learning_rate": 1.8432150831274376e-05,
      "loss": 0.5009,
      "step": 6528
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7824981212615967,
      "learning_rate": 1.8431603314166434e-05,
      "loss": 0.5765,
      "step": 6529
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.776205062866211,
      "learning_rate": 1.843105570960928e-05,
      "loss": 0.5179,
      "step": 6530
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7646987438201904,
      "learning_rate": 1.8430508017608596e-05,
      "loss": 0.5619,
      "step": 6531
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.917668104171753,
      "learning_rate": 1.842996023817006e-05,
      "loss": 0.5657,
      "step": 6532
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9693444967269897,
      "learning_rate": 1.8429412371299362e-05,
      "loss": 0.6739,
      "step": 6533
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7830584049224854,
      "learning_rate": 1.8428864417002178e-05,
      "loss": 0.5286,
      "step": 6534
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7228602170944214,
      "learning_rate": 1.8428316375284186e-05,
      "loss": 0.6027,
      "step": 6535
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7502853870391846,
      "learning_rate": 1.842776824615108e-05,
      "loss": 0.5237,
      "step": 6536
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7610526084899902,
      "learning_rate": 1.842722002960854e-05,
      "loss": 0.6077,
      "step": 6537
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.831488847732544,
      "learning_rate": 1.8426671725662255e-05,
      "loss": 0.5312,
      "step": 6538
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.646071434020996,
      "learning_rate": 1.8426123334317907e-05,
      "loss": 0.5629,
      "step": 6539
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.797676920890808,
      "learning_rate": 1.8425574855581184e-05,
      "loss": 0.5482,
      "step": 6540
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9832231998443604,
      "learning_rate": 1.842502628945778e-05,
      "loss": 0.6339,
      "step": 6541
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8246057033538818,
      "learning_rate": 1.8424477635953377e-05,
      "loss": 0.5254,
      "step": 6542
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.773166298866272,
      "learning_rate": 1.8423928895073673e-05,
      "loss": 0.5376,
      "step": 6543
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7106645107269287,
      "learning_rate": 1.8423380066824356e-05,
      "loss": 0.5295,
      "step": 6544
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.724334716796875,
      "learning_rate": 1.8422831151211113e-05,
      "loss": 0.5795,
      "step": 6545
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7459604740142822,
      "learning_rate": 1.842228214823965e-05,
      "loss": 0.5216,
      "step": 6546
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7604715824127197,
      "learning_rate": 1.8421733057915644e-05,
      "loss": 0.5704,
      "step": 6547
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7423046827316284,
      "learning_rate": 1.8421183880244807e-05,
      "loss": 0.5018,
      "step": 6548
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7663474082946777,
      "learning_rate": 1.8420634615232824e-05,
      "loss": 0.6411,
      "step": 6549
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.660352349281311,
      "learning_rate": 1.8420085262885395e-05,
      "loss": 0.4891,
      "step": 6550
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.834330677986145,
      "learning_rate": 1.8419535823208216e-05,
      "loss": 0.6334,
      "step": 6551
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7295935153961182,
      "learning_rate": 1.8418986296206987e-05,
      "loss": 0.5311,
      "step": 6552
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.707244634628296,
      "learning_rate": 1.841843668188741e-05,
      "loss": 0.5318,
      "step": 6553
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8376275300979614,
      "learning_rate": 1.841788698025518e-05,
      "loss": 0.551,
      "step": 6554
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.686110496520996,
      "learning_rate": 1.8417337191316003e-05,
      "loss": 0.5394,
      "step": 6555
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7798426151275635,
      "learning_rate": 1.8416787315075583e-05,
      "loss": 0.586,
      "step": 6556
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9266080856323242,
      "learning_rate": 1.8416237351539615e-05,
      "loss": 0.5405,
      "step": 6557
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8606185913085938,
      "learning_rate": 1.8415687300713805e-05,
      "loss": 0.5574,
      "step": 6558
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8100959062576294,
      "learning_rate": 1.8415137162603865e-05,
      "loss": 0.6209,
      "step": 6559
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7259345054626465,
      "learning_rate": 1.8414586937215495e-05,
      "loss": 0.4888,
      "step": 6560
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8251103162765503,
      "learning_rate": 1.8414036624554403e-05,
      "loss": 0.5598,
      "step": 6561
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7205803394317627,
      "learning_rate": 1.8413486224626294e-05,
      "loss": 0.5031,
      "step": 6562
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6793147325515747,
      "learning_rate": 1.8412935737436883e-05,
      "loss": 0.5816,
      "step": 6563
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7579127550125122,
      "learning_rate": 1.8412385162991872e-05,
      "loss": 0.537,
      "step": 6564
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.795038104057312,
      "learning_rate": 1.8411834501296973e-05,
      "loss": 0.5904,
      "step": 6565
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.763036847114563,
      "learning_rate": 1.8411283752357904e-05,
      "loss": 0.526,
      "step": 6566
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7977663278579712,
      "learning_rate": 1.841073291618037e-05,
      "loss": 0.5833,
      "step": 6567
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7081540822982788,
      "learning_rate": 1.8410181992770085e-05,
      "loss": 0.5222,
      "step": 6568
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7224520444869995,
      "learning_rate": 1.8409630982132767e-05,
      "loss": 0.5711,
      "step": 6569
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6899259090423584,
      "learning_rate": 1.8409079884274125e-05,
      "loss": 0.4693,
      "step": 6570
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7590090036392212,
      "learning_rate": 1.8408528699199882e-05,
      "loss": 0.5372,
      "step": 6571
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.739113450050354,
      "learning_rate": 1.8407977426915747e-05,
      "loss": 0.5103,
      "step": 6572
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.752634882926941,
      "learning_rate": 1.840742606742744e-05,
      "loss": 0.5776,
      "step": 6573
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7928705215454102,
      "learning_rate": 1.8406874620740686e-05,
      "loss": 0.557,
      "step": 6574
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8726251125335693,
      "learning_rate": 1.8406323086861192e-05,
      "loss": 0.577,
      "step": 6575
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.647391438484192,
      "learning_rate": 1.840577146579469e-05,
      "loss": 0.4744,
      "step": 6576
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7276010513305664,
      "learning_rate": 1.8405219757546894e-05,
      "loss": 0.5422,
      "step": 6577
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8300491571426392,
      "learning_rate": 1.8404667962123532e-05,
      "loss": 0.5068,
      "step": 6578
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7586755752563477,
      "learning_rate": 1.8404116079530318e-05,
      "loss": 0.5965,
      "step": 6579
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.791660189628601,
      "learning_rate": 1.8403564109772986e-05,
      "loss": 0.546,
      "step": 6580
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.738273024559021,
      "learning_rate": 1.840301205285725e-05,
      "loss": 0.5866,
      "step": 6581
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8387739658355713,
      "learning_rate": 1.840245990878885e-05,
      "loss": 0.5318,
      "step": 6582
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7810002565383911,
      "learning_rate": 1.84019076775735e-05,
      "loss": 0.5718,
      "step": 6583
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.678992748260498,
      "learning_rate": 1.8401355359216935e-05,
      "loss": 0.4922,
      "step": 6584
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7305890321731567,
      "learning_rate": 1.8400802953724876e-05,
      "loss": 0.5553,
      "step": 6585
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.675478219985962,
      "learning_rate": 1.840025046110306e-05,
      "loss": 0.4928,
      "step": 6586
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7423194646835327,
      "learning_rate": 1.8399697881357214e-05,
      "loss": 0.5226,
      "step": 6587
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.799012303352356,
      "learning_rate": 1.8399145214493067e-05,
      "loss": 0.5556,
      "step": 6588
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8699861764907837,
      "learning_rate": 1.8398592460516355e-05,
      "loss": 0.6197,
      "step": 6589
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7710626125335693,
      "learning_rate": 1.839803961943281e-05,
      "loss": 0.4847,
      "step": 6590
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8225879669189453,
      "learning_rate": 1.8397486691248162e-05,
      "loss": 0.5769,
      "step": 6591
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7243971824645996,
      "learning_rate": 1.8396933675968152e-05,
      "loss": 0.4747,
      "step": 6592
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7848589420318604,
      "learning_rate": 1.8396380573598512e-05,
      "loss": 0.556,
      "step": 6593
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7209832668304443,
      "learning_rate": 1.8395827384144978e-05,
      "loss": 0.5305,
      "step": 6594
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.768038272857666,
      "learning_rate": 1.8395274107613285e-05,
      "loss": 0.5228,
      "step": 6595
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8243952989578247,
      "learning_rate": 1.839472074400918e-05,
      "loss": 0.5832,
      "step": 6596
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7314460277557373,
      "learning_rate": 1.8394167293338396e-05,
      "loss": 0.5377,
      "step": 6597
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7775954008102417,
      "learning_rate": 1.8393613755606673e-05,
      "loss": 0.5287,
      "step": 6598
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8570140600204468,
      "learning_rate": 1.8393060130819752e-05,
      "loss": 0.6261,
      "step": 6599
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7726596593856812,
      "learning_rate": 1.839250641898338e-05,
      "loss": 0.5233,
      "step": 6600
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8101881742477417,
      "learning_rate": 1.839195262010329e-05,
      "loss": 0.6073,
      "step": 6601
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.734437108039856,
      "learning_rate": 1.8391398734185237e-05,
      "loss": 0.5411,
      "step": 6602
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7506648302078247,
      "learning_rate": 1.839084476123496e-05,
      "loss": 0.5992,
      "step": 6603
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8701056241989136,
      "learning_rate": 1.8390290701258206e-05,
      "loss": 0.5634,
      "step": 6604
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7878338098526,
      "learning_rate": 1.8389736554260718e-05,
      "loss": 0.6244,
      "step": 6605
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7271997928619385,
      "learning_rate": 1.838918232024825e-05,
      "loss": 0.5058,
      "step": 6606
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7019811868667603,
      "learning_rate": 1.8388627999226542e-05,
      "loss": 0.5623,
      "step": 6607
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7465838193893433,
      "learning_rate": 1.838807359120135e-05,
      "loss": 0.4731,
      "step": 6608
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7100697755813599,
      "learning_rate": 1.8387519096178424e-05,
      "loss": 0.5547,
      "step": 6609
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.664759635925293,
      "learning_rate": 1.8386964514163507e-05,
      "loss": 0.4897,
      "step": 6610
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8481210470199585,
      "learning_rate": 1.838640984516236e-05,
      "loss": 0.5775,
      "step": 6611
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.789452314376831,
      "learning_rate": 1.838585508918073e-05,
      "loss": 0.5008,
      "step": 6612
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9043893814086914,
      "learning_rate": 1.8385300246224375e-05,
      "loss": 0.5963,
      "step": 6613
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.778806209564209,
      "learning_rate": 1.838474531629905e-05,
      "loss": 0.523,
      "step": 6614
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6717808246612549,
      "learning_rate": 1.8384190299410506e-05,
      "loss": 0.5179,
      "step": 6615
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6627029180526733,
      "learning_rate": 1.83836351955645e-05,
      "loss": 0.4951,
      "step": 6616
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7018022537231445,
      "learning_rate": 1.8383080004766792e-05,
      "loss": 0.5608,
      "step": 6617
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6314257383346558,
      "learning_rate": 1.838252472702314e-05,
      "loss": 0.4941,
      "step": 6618
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8734519481658936,
      "learning_rate": 1.8381969362339302e-05,
      "loss": 0.6127,
      "step": 6619
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.627454400062561,
      "learning_rate": 1.8381413910721035e-05,
      "loss": 0.5028,
      "step": 6620
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8126524686813354,
      "learning_rate": 1.8380858372174107e-05,
      "loss": 0.6116,
      "step": 6621
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6871417760849,
      "learning_rate": 1.8380302746704273e-05,
      "loss": 0.4994,
      "step": 6622
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6737815141677856,
      "learning_rate": 1.83797470343173e-05,
      "loss": 0.5177,
      "step": 6623
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7723091840744019,
      "learning_rate": 1.8379191235018955e-05,
      "loss": 0.5472,
      "step": 6624
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7213236093521118,
      "learning_rate": 1.8378635348814994e-05,
      "loss": 0.5576,
      "step": 6625
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.680252194404602,
      "learning_rate": 1.8378079375711183e-05,
      "loss": 0.4965,
      "step": 6626
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8965065479278564,
      "learning_rate": 1.8377523315713294e-05,
      "loss": 0.6354,
      "step": 6627
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7328648567199707,
      "learning_rate": 1.8376967168827094e-05,
      "loss": 0.4968,
      "step": 6628
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7279856204986572,
      "learning_rate": 1.837641093505835e-05,
      "loss": 0.6086,
      "step": 6629
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6482200622558594,
      "learning_rate": 1.8375854614412823e-05,
      "loss": 0.4829,
      "step": 6630
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7311137914657593,
      "learning_rate": 1.8375298206896296e-05,
      "loss": 0.5625,
      "step": 6631
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6432063579559326,
      "learning_rate": 1.8374741712514535e-05,
      "loss": 0.4761,
      "step": 6632
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.886741042137146,
      "learning_rate": 1.8374185131273305e-05,
      "loss": 0.5896,
      "step": 6633
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8413093090057373,
      "learning_rate": 1.8373628463178388e-05,
      "loss": 0.5656,
      "step": 6634
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7496269941329956,
      "learning_rate": 1.8373071708235556e-05,
      "loss": 0.5705,
      "step": 6635
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.750582218170166,
      "learning_rate": 1.8372514866450573e-05,
      "loss": 0.479,
      "step": 6636
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.3024072647094727,
      "learning_rate": 1.837195793782923e-05,
      "loss": 0.599,
      "step": 6637
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8380823135375977,
      "learning_rate": 1.8371400922377295e-05,
      "loss": 0.5377,
      "step": 6638
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8269184827804565,
      "learning_rate": 1.8370843820100543e-05,
      "loss": 0.5312,
      "step": 6639
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8088831901550293,
      "learning_rate": 1.8370286631004757e-05,
      "loss": 0.5136,
      "step": 6640
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7556946277618408,
      "learning_rate": 1.836972935509571e-05,
      "loss": 0.5666,
      "step": 6641
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8279753923416138,
      "learning_rate": 1.8369171992379194e-05,
      "loss": 0.5377,
      "step": 6642
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8229484558105469,
      "learning_rate": 1.836861454286097e-05,
      "loss": 0.5893,
      "step": 6643
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6899826526641846,
      "learning_rate": 1.8368057006546843e-05,
      "loss": 0.5282,
      "step": 6644
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.80227530002594,
      "learning_rate": 1.8367499383442576e-05,
      "loss": 0.6194,
      "step": 6645
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6848881244659424,
      "learning_rate": 1.8366941673553963e-05,
      "loss": 0.48,
      "step": 6646
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.679444670677185,
      "learning_rate": 1.8366383876886784e-05,
      "loss": 0.54,
      "step": 6647
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.706578254699707,
      "learning_rate": 1.8365825993446825e-05,
      "loss": 0.5573,
      "step": 6648
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6460319757461548,
      "learning_rate": 1.8365268023239875e-05,
      "loss": 0.5731,
      "step": 6649
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6796047687530518,
      "learning_rate": 1.836470996627172e-05,
      "loss": 0.5374,
      "step": 6650
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6845426559448242,
      "learning_rate": 1.836415182254814e-05,
      "loss": 0.5277,
      "step": 6651
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8612184524536133,
      "learning_rate": 1.8363593592074938e-05,
      "loss": 0.545,
      "step": 6652
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.887265920639038,
      "learning_rate": 1.8363035274857894e-05,
      "loss": 0.606,
      "step": 6653
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.755202293395996,
      "learning_rate": 1.83624768709028e-05,
      "loss": 0.4961,
      "step": 6654
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7247744798660278,
      "learning_rate": 1.8361918380215452e-05,
      "loss": 0.5824,
      "step": 6655
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6112242937088013,
      "learning_rate": 1.8361359802801633e-05,
      "loss": 0.4854,
      "step": 6656
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.3675811290740967,
      "learning_rate": 1.8360801138667145e-05,
      "loss": 0.6062,
      "step": 6657
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6022324562072754,
      "learning_rate": 1.836024238781778e-05,
      "loss": 0.4722,
      "step": 6658
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7146071195602417,
      "learning_rate": 1.8359683550259335e-05,
      "loss": 0.5617,
      "step": 6659
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.817064642906189,
      "learning_rate": 1.8359124625997603e-05,
      "loss": 0.5516,
      "step": 6660
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.908719539642334,
      "learning_rate": 1.835856561503838e-05,
      "loss": 0.5902,
      "step": 6661
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5973654985427856,
      "learning_rate": 1.8358006517387466e-05,
      "loss": 0.5026,
      "step": 6662
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7069143056869507,
      "learning_rate": 1.8357447333050658e-05,
      "loss": 0.632,
      "step": 6663
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.609892725944519,
      "learning_rate": 1.835688806203376e-05,
      "loss": 0.4655,
      "step": 6664
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7350083589553833,
      "learning_rate": 1.835632870434257e-05,
      "loss": 0.5565,
      "step": 6665
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.66033935546875,
      "learning_rate": 1.8355769259982882e-05,
      "loss": 0.511,
      "step": 6666
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7252144813537598,
      "learning_rate": 1.835520972896051e-05,
      "loss": 0.5312,
      "step": 6667
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.779388427734375,
      "learning_rate": 1.8354650111281253e-05,
      "loss": 0.5658,
      "step": 6668
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7142388820648193,
      "learning_rate": 1.835409040695091e-05,
      "loss": 0.6014,
      "step": 6669
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7062870264053345,
      "learning_rate": 1.83535306159753e-05,
      "loss": 0.5329,
      "step": 6670
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.927544355392456,
      "learning_rate": 1.8352970738360212e-05,
      "loss": 0.6104,
      "step": 6671
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.71317720413208,
      "learning_rate": 1.8352410774111462e-05,
      "loss": 0.4877,
      "step": 6672
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7939170598983765,
      "learning_rate": 1.8351850723234856e-05,
      "loss": 0.5711,
      "step": 6673
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7232904434204102,
      "learning_rate": 1.8351290585736206e-05,
      "loss": 0.5804,
      "step": 6674
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6932218074798584,
      "learning_rate": 1.8350730361621312e-05,
      "loss": 0.5365,
      "step": 6675
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7547292709350586,
      "learning_rate": 1.8350170050895998e-05,
      "loss": 0.5521,
      "step": 6676
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8229432106018066,
      "learning_rate": 1.834960965356606e-05,
      "loss": 0.5955,
      "step": 6677
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.702181339263916,
      "learning_rate": 1.8349049169637325e-05,
      "loss": 0.4893,
      "step": 6678
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6873013973236084,
      "learning_rate": 1.8348488599115595e-05,
      "loss": 0.521,
      "step": 6679
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6814550161361694,
      "learning_rate": 1.834792794200669e-05,
      "loss": 0.508,
      "step": 6680
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7793138027191162,
      "learning_rate": 1.8347367198316423e-05,
      "loss": 0.6132,
      "step": 6681
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6726524829864502,
      "learning_rate": 1.834680636805061e-05,
      "loss": 0.5132,
      "step": 6682
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7943158149719238,
      "learning_rate": 1.8346245451215068e-05,
      "loss": 0.6052,
      "step": 6683
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6859856843948364,
      "learning_rate": 1.8345684447815615e-05,
      "loss": 0.518,
      "step": 6684
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.75267493724823,
      "learning_rate": 1.8345123357858065e-05,
      "loss": 0.5574,
      "step": 6685
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7235502004623413,
      "learning_rate": 1.8344562181348246e-05,
      "loss": 0.5224,
      "step": 6686
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8218978643417358,
      "learning_rate": 1.834400091829197e-05,
      "loss": 0.6352,
      "step": 6687
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.767118215560913,
      "learning_rate": 1.8343439568695066e-05,
      "loss": 0.5162,
      "step": 6688
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8847877979278564,
      "learning_rate": 1.834287813256335e-05,
      "loss": 0.6477,
      "step": 6689
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7237701416015625,
      "learning_rate": 1.8342316609902644e-05,
      "loss": 0.5069,
      "step": 6690
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6862244606018066,
      "learning_rate": 1.834175500071878e-05,
      "loss": 0.5652,
      "step": 6691
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.576385259628296,
      "learning_rate": 1.834119330501757e-05,
      "loss": 0.4727,
      "step": 6692
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7516801357269287,
      "learning_rate": 1.8340631522804852e-05,
      "loss": 0.6037,
      "step": 6693
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6972562074661255,
      "learning_rate": 1.834006965408645e-05,
      "loss": 0.495,
      "step": 6694
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7184346914291382,
      "learning_rate": 1.8339507698868188e-05,
      "loss": 0.567,
      "step": 6695
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6573216915130615,
      "learning_rate": 1.8338945657155894e-05,
      "loss": 0.5157,
      "step": 6696
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7644020318984985,
      "learning_rate": 1.8338383528955396e-05,
      "loss": 0.5913,
      "step": 6697
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.766808271408081,
      "learning_rate": 1.833782131427253e-05,
      "loss": 0.5351,
      "step": 6698
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9054927825927734,
      "learning_rate": 1.8337259013113126e-05,
      "loss": 0.5962,
      "step": 6699
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.641831398010254,
      "learning_rate": 1.8336696625483012e-05,
      "loss": 0.4754,
      "step": 6700
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.847326397895813,
      "learning_rate": 1.8336134151388022e-05,
      "loss": 0.6289,
      "step": 6701
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.642198920249939,
      "learning_rate": 1.8335571590833994e-05,
      "loss": 0.4549,
      "step": 6702
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6673383712768555,
      "learning_rate": 1.8335008943826758e-05,
      "loss": 0.51,
      "step": 6703
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9630138874053955,
      "learning_rate": 1.833444621037215e-05,
      "loss": 0.5171,
      "step": 6704
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.74066162109375,
      "learning_rate": 1.8333883390476006e-05,
      "loss": 0.589,
      "step": 6705
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.690168023109436,
      "learning_rate": 1.8333320484144165e-05,
      "loss": 0.4842,
      "step": 6706
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.690963864326477,
      "learning_rate": 1.833275749138247e-05,
      "loss": 0.6018,
      "step": 6707
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6254773139953613,
      "learning_rate": 1.833219441219675e-05,
      "loss": 0.5022,
      "step": 6708
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7955907583236694,
      "learning_rate": 1.833163124659285e-05,
      "loss": 0.6216,
      "step": 6709
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7173540592193604,
      "learning_rate": 1.8331067994576615e-05,
      "loss": 0.521,
      "step": 6710
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8441283702850342,
      "learning_rate": 1.8330504656153877e-05,
      "loss": 0.6109,
      "step": 6711
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.692679524421692,
      "learning_rate": 1.832994123133049e-05,
      "loss": 0.4634,
      "step": 6712
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.770379900932312,
      "learning_rate": 1.8329377720112292e-05,
      "loss": 0.6209,
      "step": 6713
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7235609292984009,
      "learning_rate": 1.8328814122505126e-05,
      "loss": 0.4989,
      "step": 6714
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6989110708236694,
      "learning_rate": 1.8328250438514837e-05,
      "loss": 0.5307,
      "step": 6715
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.84386146068573,
      "learning_rate": 1.8327686668147272e-05,
      "loss": 0.5212,
      "step": 6716
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7725857496261597,
      "learning_rate": 1.8327122811408286e-05,
      "loss": 0.5503,
      "step": 6717
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6173381805419922,
      "learning_rate": 1.8326558868303717e-05,
      "loss": 0.4934,
      "step": 6718
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7641171216964722,
      "learning_rate": 1.832599483883942e-05,
      "loss": 0.5716,
      "step": 6719
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9055490493774414,
      "learning_rate": 1.8325430723021236e-05,
      "loss": 0.5541,
      "step": 6720
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8250643014907837,
      "learning_rate": 1.8324866520855028e-05,
      "loss": 0.7137,
      "step": 6721
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6840852499008179,
      "learning_rate": 1.8324302232346642e-05,
      "loss": 0.5388,
      "step": 6722
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6635215282440186,
      "learning_rate": 1.8323737857501932e-05,
      "loss": 0.5495,
      "step": 6723
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.720100998878479,
      "learning_rate": 1.8323173396326747e-05,
      "loss": 0.4734,
      "step": 6724
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.317950963973999,
      "learning_rate": 1.8322608848826944e-05,
      "loss": 1.0531,
      "step": 6725
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.731608271598816,
      "learning_rate": 1.8322044215008382e-05,
      "loss": 1.0025,
      "step": 6726
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8948140144348145,
      "learning_rate": 1.8321479494876912e-05,
      "loss": 0.6384,
      "step": 6727
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.713168978691101,
      "learning_rate": 1.8320914688438392e-05,
      "loss": 0.4995,
      "step": 6728
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7376677989959717,
      "learning_rate": 1.8320349795698684e-05,
      "loss": 0.5678,
      "step": 6729
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7839761972427368,
      "learning_rate": 1.8319784816663638e-05,
      "loss": 0.5682,
      "step": 6730
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6964101791381836,
      "learning_rate": 1.8319219751339123e-05,
      "loss": 0.6087,
      "step": 6731
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.76161789894104,
      "learning_rate": 1.8318654599730998e-05,
      "loss": 0.5683,
      "step": 6732
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7755928039550781,
      "learning_rate": 1.831808936184512e-05,
      "loss": 0.7012,
      "step": 6733
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6471275091171265,
      "learning_rate": 1.8317524037687353e-05,
      "loss": 0.5259,
      "step": 6734
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7305424213409424,
      "learning_rate": 1.8316958627263564e-05,
      "loss": 0.5513,
      "step": 6735
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6152769327163696,
      "learning_rate": 1.8316393130579613e-05,
      "loss": 0.4806,
      "step": 6736
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6555447578430176,
      "learning_rate": 1.831582754764137e-05,
      "loss": 0.5268,
      "step": 6737
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7099583148956299,
      "learning_rate": 1.8315261878454692e-05,
      "loss": 0.478,
      "step": 6738
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.676923155784607,
      "learning_rate": 1.8314696123025456e-05,
      "loss": 0.5586,
      "step": 6739
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7023078203201294,
      "learning_rate": 1.831413028135952e-05,
      "loss": 0.5234,
      "step": 6740
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8656748533248901,
      "learning_rate": 1.8313564353462765e-05,
      "loss": 0.58,
      "step": 6741
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6219408512115479,
      "learning_rate": 1.8312998339341047e-05,
      "loss": 0.4989,
      "step": 6742
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7179003953933716,
      "learning_rate": 1.8312432239000245e-05,
      "loss": 0.5714,
      "step": 6743
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.70663583278656,
      "learning_rate": 1.831186605244623e-05,
      "loss": 0.5001,
      "step": 6744
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.735836148262024,
      "learning_rate": 1.8311299779684876e-05,
      "loss": 0.5543,
      "step": 6745
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7281609773635864,
      "learning_rate": 1.831073342072205e-05,
      "loss": 0.518,
      "step": 6746
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.654596209526062,
      "learning_rate": 1.8310166975563627e-05,
      "loss": 0.5337,
      "step": 6747
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6507742404937744,
      "learning_rate": 1.8309600444215484e-05,
      "loss": 0.5133,
      "step": 6748
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7475477457046509,
      "learning_rate": 1.8309033826683496e-05,
      "loss": 0.5637,
      "step": 6749
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8239771127700806,
      "learning_rate": 1.8308467122973543e-05,
      "loss": 0.541,
      "step": 6750
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.68038809299469,
      "learning_rate": 1.83079003330915e-05,
      "loss": 0.5549,
      "step": 6751
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.938843011856079,
      "learning_rate": 1.8307333457043244e-05,
      "loss": 0.6509,
      "step": 6752
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7055691480636597,
      "learning_rate": 1.8306766494834658e-05,
      "loss": 0.552,
      "step": 6753
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7345967292785645,
      "learning_rate": 1.8306199446471617e-05,
      "loss": 0.4707,
      "step": 6754
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8931430578231812,
      "learning_rate": 1.830563231196001e-05,
      "loss": 0.5425,
      "step": 6755
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6944940090179443,
      "learning_rate": 1.830506509130571e-05,
      "loss": 0.5045,
      "step": 6756
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8254483938217163,
      "learning_rate": 1.8304497784514607e-05,
      "loss": 0.5992,
      "step": 6757
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.613613486289978,
      "learning_rate": 1.8303930391592583e-05,
      "loss": 0.4804,
      "step": 6758
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8066927194595337,
      "learning_rate": 1.8303362912545516e-05,
      "loss": 0.5515,
      "step": 6759
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6683385372161865,
      "learning_rate": 1.8302795347379304e-05,
      "loss": 0.5111,
      "step": 6760
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.759202480316162,
      "learning_rate": 1.8302227696099824e-05,
      "loss": 0.5637,
      "step": 6761
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7467762231826782,
      "learning_rate": 1.8301659958712973e-05,
      "loss": 0.5099,
      "step": 6762
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8444823026657104,
      "learning_rate": 1.8301092135224627e-05,
      "loss": 0.5633,
      "step": 6763
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6659985780715942,
      "learning_rate": 1.8300524225640684e-05,
      "loss": 0.4974,
      "step": 6764
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7380892038345337,
      "learning_rate": 1.829995622996703e-05,
      "loss": 0.5817,
      "step": 6765
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6999317407608032,
      "learning_rate": 1.829938814820956e-05,
      "loss": 0.5192,
      "step": 6766
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.747739315032959,
      "learning_rate": 1.829881998037416e-05,
      "loss": 0.5511,
      "step": 6767
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6887271404266357,
      "learning_rate": 1.829825172646673e-05,
      "loss": 0.5013,
      "step": 6768
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.720836877822876,
      "learning_rate": 1.8297683386493158e-05,
      "loss": 0.5639,
      "step": 6769
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8643909692764282,
      "learning_rate": 1.8297114960459342e-05,
      "loss": 0.5093,
      "step": 6770
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7995834350585938,
      "learning_rate": 1.8296546448371177e-05,
      "loss": 0.5951,
      "step": 6771
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.892691969871521,
      "learning_rate": 1.829597785023456e-05,
      "loss": 0.62,
      "step": 6772
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7615755796432495,
      "learning_rate": 1.829540916605538e-05,
      "loss": 0.5636,
      "step": 6773
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.648591160774231,
      "learning_rate": 1.8294840395839546e-05,
      "loss": 0.475,
      "step": 6774
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7351516485214233,
      "learning_rate": 1.829427153959295e-05,
      "loss": 0.5818,
      "step": 6775
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6587923765182495,
      "learning_rate": 1.82937025973215e-05,
      "loss": 0.5305,
      "step": 6776
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9056328535079956,
      "learning_rate": 1.829313356903109e-05,
      "loss": 0.6128,
      "step": 6777
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.847923755645752,
      "learning_rate": 1.829256445472762e-05,
      "loss": 0.5105,
      "step": 6778
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7965248823165894,
      "learning_rate": 1.8291995254417e-05,
      "loss": 0.55,
      "step": 6779
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7030303478240967,
      "learning_rate": 1.8291425968105127e-05,
      "loss": 0.5011,
      "step": 6780
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8257933855056763,
      "learning_rate": 1.829085659579791e-05,
      "loss": 0.557,
      "step": 6781
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7270443439483643,
      "learning_rate": 1.829028713750125e-05,
      "loss": 0.5205,
      "step": 6782
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6459269523620605,
      "learning_rate": 1.8289717593221058e-05,
      "loss": 0.509,
      "step": 6783
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8220643997192383,
      "learning_rate": 1.828914796296324e-05,
      "loss": 0.5666,
      "step": 6784
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8250033855438232,
      "learning_rate": 1.82885782467337e-05,
      "loss": 0.6094,
      "step": 6785
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6522700786590576,
      "learning_rate": 1.828800844453835e-05,
      "loss": 0.4943,
      "step": 6786
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.653343915939331,
      "learning_rate": 1.82874385563831e-05,
      "loss": 0.5574,
      "step": 6787
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.672998309135437,
      "learning_rate": 1.828686858227386e-05,
      "loss": 0.509,
      "step": 6788
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7545239925384521,
      "learning_rate": 1.828629852221654e-05,
      "loss": 0.5854,
      "step": 6789
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6981306076049805,
      "learning_rate": 1.8285728376217053e-05,
      "loss": 0.4907,
      "step": 6790
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6119283437728882,
      "learning_rate": 1.8285158144281316e-05,
      "loss": 0.5234,
      "step": 6791
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7075353860855103,
      "learning_rate": 1.8284587826415243e-05,
      "loss": 0.5509,
      "step": 6792
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.954652190208435,
      "learning_rate": 1.8284017422624737e-05,
      "loss": 0.9847,
      "step": 6793
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6527248620986938,
      "learning_rate": 1.8283446932915733e-05,
      "loss": 1.0208,
      "step": 6794
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9314343929290771,
      "learning_rate": 1.8282876357294132e-05,
      "loss": 0.5729,
      "step": 6795
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8747726678848267,
      "learning_rate": 1.828230569576586e-05,
      "loss": 0.5179,
      "step": 6796
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8762292861938477,
      "learning_rate": 1.8281734948336835e-05,
      "loss": 0.5602,
      "step": 6797
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6749166250228882,
      "learning_rate": 1.8281164115012978e-05,
      "loss": 0.5081,
      "step": 6798
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0811028480529785,
      "learning_rate": 1.8280593195800204e-05,
      "loss": 1.0267,
      "step": 6799
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.725231409072876,
      "learning_rate": 1.8280022190704435e-05,
      "loss": 1.0646,
      "step": 6800
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8809891939163208,
      "learning_rate": 1.8279451099731597e-05,
      "loss": 0.5729,
      "step": 6801
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7768218517303467,
      "learning_rate": 1.8278879922887612e-05,
      "loss": 0.4811,
      "step": 6802
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8429405689239502,
      "learning_rate": 1.8278308660178403e-05,
      "loss": 0.5704,
      "step": 6803
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6304196119308472,
      "learning_rate": 1.8277737311609897e-05,
      "loss": 0.5055,
      "step": 6804
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6853975057601929,
      "learning_rate": 1.8277165877188017e-05,
      "loss": 0.515,
      "step": 6805
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.059084177017212,
      "learning_rate": 1.8276594356918692e-05,
      "loss": 0.5794,
      "step": 6806
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.823142647743225,
      "learning_rate": 1.8276022750807846e-05,
      "loss": 0.9843,
      "step": 6807
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5138819217681885,
      "learning_rate": 1.8275451058861415e-05,
      "loss": 1.0794,
      "step": 6808
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.983420491218567,
      "learning_rate": 1.827487928108532e-05,
      "loss": 0.5766,
      "step": 6809
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8601503372192383,
      "learning_rate": 1.8274307417485495e-05,
      "loss": 0.5187,
      "step": 6810
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9947965145111084,
      "learning_rate": 1.8273735468067872e-05,
      "loss": 0.6513,
      "step": 6811
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7465550899505615,
      "learning_rate": 1.827316343283838e-05,
      "loss": 0.4913,
      "step": 6812
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8509660959243774,
      "learning_rate": 1.8272591311802956e-05,
      "loss": 0.5865,
      "step": 6813
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7532384395599365,
      "learning_rate": 1.8272019104967527e-05,
      "loss": 0.5039,
      "step": 6814
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7707865238189697,
      "learning_rate": 1.827144681233804e-05,
      "loss": 0.5668,
      "step": 6815
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7599430084228516,
      "learning_rate": 1.8270874433920417e-05,
      "loss": 0.4957,
      "step": 6816
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8169389963150024,
      "learning_rate": 1.8270301969720597e-05,
      "loss": 0.6016,
      "step": 6817
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.742561936378479,
      "learning_rate": 1.826972941974453e-05,
      "loss": 0.495,
      "step": 6818
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7981293201446533,
      "learning_rate": 1.826915678399814e-05,
      "loss": 0.5283,
      "step": 6819
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6075481176376343,
      "learning_rate": 1.8268584062487368e-05,
      "loss": 0.5588,
      "step": 6820
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7979493141174316,
      "learning_rate": 1.8268011255218164e-05,
      "loss": 0.5941,
      "step": 6821
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7455825805664062,
      "learning_rate": 1.8267438362196456e-05,
      "loss": 0.4977,
      "step": 6822
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9267765283584595,
      "learning_rate": 1.8266865383428194e-05,
      "loss": 0.5717,
      "step": 6823
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.742913842201233,
      "learning_rate": 1.8266292318919322e-05,
      "loss": 0.4992,
      "step": 6824
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8446217775344849,
      "learning_rate": 1.8265719168675775e-05,
      "loss": 0.5793,
      "step": 6825
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6635915040969849,
      "learning_rate": 1.82651459327035e-05,
      "loss": 0.477,
      "step": 6826
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6498960256576538,
      "learning_rate": 1.8264572611008455e-05,
      "loss": 0.5525,
      "step": 6827
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6745933294296265,
      "learning_rate": 1.826399920359657e-05,
      "loss": 0.4612,
      "step": 6828
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7453652620315552,
      "learning_rate": 1.82634257104738e-05,
      "loss": 0.5842,
      "step": 6829
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6623464822769165,
      "learning_rate": 1.826285213164609e-05,
      "loss": 0.5027,
      "step": 6830
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7262049913406372,
      "learning_rate": 1.826227846711939e-05,
      "loss": 0.5546,
      "step": 6831
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8140671253204346,
      "learning_rate": 1.8261704716899653e-05,
      "loss": 0.5159,
      "step": 6832
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8227264881134033,
      "learning_rate": 1.8261130880992825e-05,
      "loss": 0.6324,
      "step": 6833
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7439357042312622,
      "learning_rate": 1.826055695940486e-05,
      "loss": 0.6059,
      "step": 6834
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.70749032497406,
      "learning_rate": 1.825998295214171e-05,
      "loss": 0.5493,
      "step": 6835
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.717171549797058,
      "learning_rate": 1.825940885920933e-05,
      "loss": 0.5407,
      "step": 6836
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6931853294372559,
      "learning_rate": 1.8258834680613672e-05,
      "loss": 0.5779,
      "step": 6837
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.709391713142395,
      "learning_rate": 1.825826041636069e-05,
      "loss": 0.5087,
      "step": 6838
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7244086265563965,
      "learning_rate": 1.8257686066456343e-05,
      "loss": 0.6361,
      "step": 6839
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.761250376701355,
      "learning_rate": 1.8257111630906587e-05,
      "loss": 0.4971,
      "step": 6840
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6957842111587524,
      "learning_rate": 1.8256537109717382e-05,
      "loss": 0.5686,
      "step": 6841
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6184320449829102,
      "learning_rate": 1.825596250289468e-05,
      "loss": 0.4771,
      "step": 6842
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8433061838150024,
      "learning_rate": 1.8255387810444447e-05,
      "loss": 0.6507,
      "step": 6843
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7324779033660889,
      "learning_rate": 1.825481303237264e-05,
      "loss": 0.4881,
      "step": 6844
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7861756086349487,
      "learning_rate": 1.8254238168685222e-05,
      "loss": 0.5456,
      "step": 6845
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7125892639160156,
      "learning_rate": 1.8253663219388157e-05,
      "loss": 0.5529,
      "step": 6846
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6715589761734009,
      "learning_rate": 1.8253088184487405e-05,
      "loss": 0.5275,
      "step": 6847
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7408748865127563,
      "learning_rate": 1.8252513063988932e-05,
      "loss": 0.4921,
      "step": 6848
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6712201833724976,
      "learning_rate": 1.8251937857898698e-05,
      "loss": 0.5345,
      "step": 6849
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7720648050308228,
      "learning_rate": 1.8251362566222676e-05,
      "loss": 0.5375,
      "step": 6850
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7366971969604492,
      "learning_rate": 1.8250787188966833e-05,
      "loss": 0.5753,
      "step": 6851
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.763309359550476,
      "learning_rate": 1.8250211726137128e-05,
      "loss": 0.5117,
      "step": 6852
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7901593446731567,
      "learning_rate": 1.8249636177739534e-05,
      "loss": 0.571,
      "step": 6853
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6476414203643799,
      "learning_rate": 1.8249060543780026e-05,
      "loss": 0.4881,
      "step": 6854
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7226536273956299,
      "learning_rate": 1.8248484824264566e-05,
      "loss": 0.6153,
      "step": 6855
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6920229196548462,
      "learning_rate": 1.824790901919913e-05,
      "loss": 0.4902,
      "step": 6856
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.861984372138977,
      "learning_rate": 1.8247333128589685e-05,
      "loss": 0.6602,
      "step": 6857
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.783290147781372,
      "learning_rate": 1.824675715244221e-05,
      "loss": 0.4919,
      "step": 6858
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.823341965675354,
      "learning_rate": 1.8246181090762677e-05,
      "loss": 0.5609,
      "step": 6859
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.601850986480713,
      "learning_rate": 1.824560494355706e-05,
      "loss": 0.4752,
      "step": 6860
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8668434619903564,
      "learning_rate": 1.8245028710831332e-05,
      "loss": 0.6763,
      "step": 6861
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6331709623336792,
      "learning_rate": 1.8244452392591473e-05,
      "loss": 0.4605,
      "step": 6862
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.698917031288147,
      "learning_rate": 1.8243875988843456e-05,
      "loss": 0.528,
      "step": 6863
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6475458145141602,
      "learning_rate": 1.8243299499593266e-05,
      "loss": 0.4836,
      "step": 6864
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6555795669555664,
      "learning_rate": 1.824272292484688e-05,
      "loss": 0.5704,
      "step": 6865
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7378153800964355,
      "learning_rate": 1.8242146264610272e-05,
      "loss": 0.473,
      "step": 6866
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6615159511566162,
      "learning_rate": 1.824156951888943e-05,
      "loss": 0.532,
      "step": 6867
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6555187702178955,
      "learning_rate": 1.8240992687690334e-05,
      "loss": 0.504,
      "step": 6868
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.872950792312622,
      "learning_rate": 1.8240415771018964e-05,
      "loss": 0.5866,
      "step": 6869
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7560209035873413,
      "learning_rate": 1.8239838768881305e-05,
      "loss": 0.5231,
      "step": 6870
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.670068383216858,
      "learning_rate": 1.823926168128334e-05,
      "loss": 0.5285,
      "step": 6871
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6949454545974731,
      "learning_rate": 1.8238684508231063e-05,
      "loss": 0.5133,
      "step": 6872
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8981844186782837,
      "learning_rate": 1.823810724973045e-05,
      "loss": 0.5955,
      "step": 6873
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6800698041915894,
      "learning_rate": 1.8237529905787488e-05,
      "loss": 0.479,
      "step": 6874
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.735520601272583,
      "learning_rate": 1.8236952476408173e-05,
      "loss": 0.538,
      "step": 6875
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6801432371139526,
      "learning_rate": 1.8236374961598486e-05,
      "loss": 0.4983,
      "step": 6876
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7683584690093994,
      "learning_rate": 1.823579736136442e-05,
      "loss": 0.5681,
      "step": 6877
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7231992483139038,
      "learning_rate": 1.8235219675711965e-05,
      "loss": 0.5138,
      "step": 6878
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8044434785842896,
      "learning_rate": 1.8234641904647117e-05,
      "loss": 0.5592,
      "step": 6879
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6617270708084106,
      "learning_rate": 1.8234064048175863e-05,
      "loss": 0.5286,
      "step": 6880
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8374260663986206,
      "learning_rate": 1.8233486106304196e-05,
      "loss": 0.5528,
      "step": 6881
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7958755493164062,
      "learning_rate": 1.8232908079038116e-05,
      "loss": 0.4864,
      "step": 6882
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.987051010131836,
      "learning_rate": 1.823232996638361e-05,
      "loss": 0.674,
      "step": 6883
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7530953884124756,
      "learning_rate": 1.823175176834668e-05,
      "loss": 0.4837,
      "step": 6884
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.962121844291687,
      "learning_rate": 1.8231173484933324e-05,
      "loss": 0.5514,
      "step": 6885
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7418763637542725,
      "learning_rate": 1.823059511614953e-05,
      "loss": 0.5192,
      "step": 6886
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7186869382858276,
      "learning_rate": 1.8230016662001307e-05,
      "loss": 0.5621,
      "step": 6887
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8096487522125244,
      "learning_rate": 1.822943812249465e-05,
      "loss": 0.5736,
      "step": 6888
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7315962314605713,
      "learning_rate": 1.8228859497635562e-05,
      "loss": 0.6091,
      "step": 6889
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.708322525024414,
      "learning_rate": 1.8228280787430042e-05,
      "loss": 0.4844,
      "step": 6890
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8930116891860962,
      "learning_rate": 1.822770199188409e-05,
      "loss": 0.619,
      "step": 6891
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.674507975578308,
      "learning_rate": 1.8227123111003712e-05,
      "loss": 0.4877,
      "step": 6892
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8756836652755737,
      "learning_rate": 1.822654414479491e-05,
      "loss": 0.5802,
      "step": 6893
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7879375219345093,
      "learning_rate": 1.8225965093263697e-05,
      "loss": 0.5968,
      "step": 6894
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7891093492507935,
      "learning_rate": 1.8225385956416067e-05,
      "loss": 0.6393,
      "step": 6895
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7805218696594238,
      "learning_rate": 1.8224806734258034e-05,
      "loss": 0.4885,
      "step": 6896
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1969192028045654,
      "learning_rate": 1.82242274267956e-05,
      "loss": 0.49,
      "step": 6897
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7111669778823853,
      "learning_rate": 1.8223648034034777e-05,
      "loss": 0.4886,
      "step": 6898
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8620229959487915,
      "learning_rate": 1.8223068555981575e-05,
      "loss": 0.6058,
      "step": 6899
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7458672523498535,
      "learning_rate": 1.8222488992642e-05,
      "loss": 0.4996,
      "step": 6900
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1101748943328857,
      "learning_rate": 1.822190934402207e-05,
      "loss": 1.0089,
      "step": 6901
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7267574071884155,
      "learning_rate": 1.822132961012779e-05,
      "loss": 1.0223,
      "step": 6902
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9516630172729492,
      "learning_rate": 1.8220749790965174e-05,
      "loss": 0.5367,
      "step": 6903
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9571971893310547,
      "learning_rate": 1.8220169886540242e-05,
      "loss": 0.5564,
      "step": 6904
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.804911494255066,
      "learning_rate": 1.8219589896859e-05,
      "loss": 0.544,
      "step": 6905
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7044123411178589,
      "learning_rate": 1.8219009821927468e-05,
      "loss": 0.4995,
      "step": 6906
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6949801445007324,
      "learning_rate": 1.821842966175166e-05,
      "loss": 0.5763,
      "step": 6907
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.85295569896698,
      "learning_rate": 1.8217849416337596e-05,
      "loss": 0.4858,
      "step": 6908
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.727545142173767,
      "learning_rate": 1.8217269085691292e-05,
      "loss": 0.5946,
      "step": 6909
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7417163848876953,
      "learning_rate": 1.8216688669818766e-05,
      "loss": 0.5414,
      "step": 6910
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7420357465744019,
      "learning_rate": 1.821610816872604e-05,
      "loss": 0.5695,
      "step": 6911
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8622534275054932,
      "learning_rate": 1.8215527582419135e-05,
      "loss": 0.5654,
      "step": 6912
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6634477376937866,
      "learning_rate": 1.8214946910904073e-05,
      "loss": 0.5185,
      "step": 6913
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7231920957565308,
      "learning_rate": 1.821436615418687e-05,
      "loss": 0.5387,
      "step": 6914
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8249799013137817,
      "learning_rate": 1.821378531227356e-05,
      "loss": 0.6245,
      "step": 6915
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.660165786743164,
      "learning_rate": 1.8213204385170162e-05,
      "loss": 0.5075,
      "step": 6916
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7480868101119995,
      "learning_rate": 1.82126233728827e-05,
      "loss": 0.665,
      "step": 6917
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.636494755744934,
      "learning_rate": 1.8212042275417197e-05,
      "loss": 0.5194,
      "step": 6918
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1400866508483887,
      "learning_rate": 1.8211461092779686e-05,
      "loss": 1.0331,
      "step": 6919
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.955029010772705,
      "learning_rate": 1.8210879824976194e-05,
      "loss": 1.056,
      "step": 6920
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.965972900390625,
      "learning_rate": 1.8210298472012747e-05,
      "loss": 0.6113,
      "step": 6921
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.955568552017212,
      "learning_rate": 1.8209717033895375e-05,
      "loss": 0.4952,
      "step": 6922
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7586915493011475,
      "learning_rate": 1.8209135510630113e-05,
      "loss": 0.542,
      "step": 6923
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7268226146697998,
      "learning_rate": 1.8208553902222987e-05,
      "loss": 0.5771,
      "step": 6924
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8938379287719727,
      "learning_rate": 1.820797220868003e-05,
      "loss": 0.6297,
      "step": 6925
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7662984132766724,
      "learning_rate": 1.8207390430007277e-05,
      "loss": 0.4697,
      "step": 6926
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7477065324783325,
      "learning_rate": 1.820680856621076e-05,
      "loss": 0.5861,
      "step": 6927
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8103630542755127,
      "learning_rate": 1.8206226617296514e-05,
      "loss": 0.4812,
      "step": 6928
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7566930055618286,
      "learning_rate": 1.8205644583270578e-05,
      "loss": 0.5444,
      "step": 6929
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8642405271530151,
      "learning_rate": 1.8205062464138983e-05,
      "loss": 0.5816,
      "step": 6930
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8363991975784302,
      "learning_rate": 1.8204480259907774e-05,
      "loss": 0.6094,
      "step": 6931
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.832868218421936,
      "learning_rate": 1.820389797058298e-05,
      "loss": 0.4979,
      "step": 6932
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7333784103393555,
      "learning_rate": 1.8203315596170652e-05,
      "loss": 0.5361,
      "step": 6933
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7117640972137451,
      "learning_rate": 1.820273313667682e-05,
      "loss": 0.4993,
      "step": 6934
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7887372970581055,
      "learning_rate": 1.8202150592107526e-05,
      "loss": 0.5511,
      "step": 6935
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6237115859985352,
      "learning_rate": 1.820156796246882e-05,
      "loss": 0.5038,
      "step": 6936
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.792952537536621,
      "learning_rate": 1.8200985247766735e-05,
      "loss": 0.5714,
      "step": 6937
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.642305374145508,
      "learning_rate": 1.8200402448007318e-05,
      "loss": 0.5003,
      "step": 6938
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7379871606826782,
      "learning_rate": 1.8199819563196618e-05,
      "loss": 0.5041,
      "step": 6939
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6834379434585571,
      "learning_rate": 1.8199236593340673e-05,
      "loss": 0.4782,
      "step": 6940
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7704483270645142,
      "learning_rate": 1.8198653538445537e-05,
      "loss": 0.5276,
      "step": 6941
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.723881483078003,
      "learning_rate": 1.819807039851725e-05,
      "loss": 0.5109,
      "step": 6942
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7981696128845215,
      "learning_rate": 1.8197487173561866e-05,
      "loss": 0.6061,
      "step": 6943
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.860373616218567,
      "learning_rate": 1.819690386358543e-05,
      "loss": 0.5226,
      "step": 6944
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7168909311294556,
      "learning_rate": 1.8196320468593997e-05,
      "loss": 0.5906,
      "step": 6945
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.801609992980957,
      "learning_rate": 1.8195736988593614e-05,
      "loss": 0.5079,
      "step": 6946
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7625845670700073,
      "learning_rate": 1.8195153423590328e-05,
      "loss": 0.532,
      "step": 6947
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6727179288864136,
      "learning_rate": 1.81945697735902e-05,
      "loss": 0.4807,
      "step": 6948
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7434643507003784,
      "learning_rate": 1.8193986038599275e-05,
      "loss": 0.5927,
      "step": 6949
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6851710081100464,
      "learning_rate": 1.8193402218623618e-05,
      "loss": 0.4901,
      "step": 6950
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7491799592971802,
      "learning_rate": 1.8192818313669276e-05,
      "loss": 0.5582,
      "step": 6951
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.645912766456604,
      "learning_rate": 1.8192234323742306e-05,
      "loss": 0.4973,
      "step": 6952
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.715807318687439,
      "learning_rate": 1.8191650248848767e-05,
      "loss": 0.5777,
      "step": 6953
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8382338285446167,
      "learning_rate": 1.8191066088994716e-05,
      "loss": 0.5393,
      "step": 6954
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7536066770553589,
      "learning_rate": 1.819048184418621e-05,
      "loss": 0.6564,
      "step": 6955
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6175823211669922,
      "learning_rate": 1.818989751442931e-05,
      "loss": 0.5001,
      "step": 6956
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.863675594329834,
      "learning_rate": 1.818931309973008e-05,
      "loss": 0.5878,
      "step": 6957
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6659328937530518,
      "learning_rate": 1.8188728600094573e-05,
      "loss": 0.4688,
      "step": 6958
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7669899463653564,
      "learning_rate": 1.8188144015528858e-05,
      "loss": 0.5778,
      "step": 6959
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7641137838363647,
      "learning_rate": 1.8187559346038996e-05,
      "loss": 0.5221,
      "step": 6960
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6638277769088745,
      "learning_rate": 1.8186974591631053e-05,
      "loss": 0.5054,
      "step": 6961
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6512691974639893,
      "learning_rate": 1.8186389752311088e-05,
      "loss": 0.4688,
      "step": 6962
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.414637804031372,
      "learning_rate": 1.8185804828085175e-05,
      "loss": 1.0422,
      "step": 6963
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0889852046966553,
      "learning_rate": 1.8185219818959374e-05,
      "loss": 1.0583,
      "step": 6964
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7742934226989746,
      "learning_rate": 1.8184634724939753e-05,
      "loss": 0.5331,
      "step": 6965
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9434494972229004,
      "learning_rate": 1.8184049546032384e-05,
      "loss": 0.4983,
      "step": 6966
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8173751831054688,
      "learning_rate": 1.8183464282243336e-05,
      "loss": 0.5666,
      "step": 6967
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7467677593231201,
      "learning_rate": 1.8182878933578672e-05,
      "loss": 0.5565,
      "step": 6968
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6833094358444214,
      "learning_rate": 1.8182293500044472e-05,
      "loss": 0.5896,
      "step": 6969
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6945133209228516,
      "learning_rate": 1.8181707981646803e-05,
      "loss": 0.4718,
      "step": 6970
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7625932693481445,
      "learning_rate": 1.818112237839174e-05,
      "loss": 0.537,
      "step": 6971
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7517131567001343,
      "learning_rate": 1.818053669028536e-05,
      "loss": 0.5168,
      "step": 6972
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8362637758255005,
      "learning_rate": 1.8179950917333723e-05,
      "loss": 0.5814,
      "step": 6973
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6641602516174316,
      "learning_rate": 1.817936505954292e-05,
      "loss": 0.505,
      "step": 6974
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.787078619003296,
      "learning_rate": 1.8178779116919023e-05,
      "loss": 0.5869,
      "step": 6975
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7537883520126343,
      "learning_rate": 1.8178193089468108e-05,
      "loss": 0.5112,
      "step": 6976
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9103153944015503,
      "learning_rate": 1.8177606977196253e-05,
      "loss": 0.5828,
      "step": 6977
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6954981088638306,
      "learning_rate": 1.8177020780109535e-05,
      "loss": 0.5035,
      "step": 6978
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8317898511886597,
      "learning_rate": 1.8176434498214036e-05,
      "loss": 0.5577,
      "step": 6979
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.744957685470581,
      "learning_rate": 1.817584813151584e-05,
      "loss": 0.4835,
      "step": 6980
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6810938119888306,
      "learning_rate": 1.817526168002102e-05,
      "loss": 0.5376,
      "step": 6981
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7092368602752686,
      "learning_rate": 1.8174675143735666e-05,
      "loss": 0.463,
      "step": 6982
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7385417222976685,
      "learning_rate": 1.817408852266586e-05,
      "loss": 0.5264,
      "step": 6983
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9921915531158447,
      "learning_rate": 1.8173501816817684e-05,
      "loss": 0.5777,
      "step": 6984
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6846873760223389,
      "learning_rate": 1.8172915026197226e-05,
      "loss": 0.5425,
      "step": 6985
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6850091218948364,
      "learning_rate": 1.8172328150810566e-05,
      "loss": 0.5736,
      "step": 6986
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7749944925308228,
      "learning_rate": 1.8171741190663796e-05,
      "loss": 0.5748,
      "step": 6987
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7288811206817627,
      "learning_rate": 1.8171154145763004e-05,
      "loss": 0.495,
      "step": 6988
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6961169242858887,
      "learning_rate": 1.8170567016114275e-05,
      "loss": 0.597,
      "step": 6989
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6991435289382935,
      "learning_rate": 1.81699798017237e-05,
      "loss": 0.4937,
      "step": 6990
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7412896156311035,
      "learning_rate": 1.8169392502597374e-05,
      "loss": 0.5621,
      "step": 6991
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.705622911453247,
      "learning_rate": 1.8168805118741383e-05,
      "loss": 0.492,
      "step": 6992
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7944022417068481,
      "learning_rate": 1.816821765016182e-05,
      "loss": 0.6386,
      "step": 6993
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6808662414550781,
      "learning_rate": 1.8167630096864778e-05,
      "loss": 0.5149,
      "step": 6994
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5896705389022827,
      "learning_rate": 1.816704245885635e-05,
      "loss": 0.5428,
      "step": 6995
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.597198247909546,
      "learning_rate": 1.8166454736142633e-05,
      "loss": 0.4714,
      "step": 6996
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7528258562088013,
      "learning_rate": 1.816586692872972e-05,
      "loss": 0.5782,
      "step": 6997
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6826119422912598,
      "learning_rate": 1.8165279036623712e-05,
      "loss": 0.5099,
      "step": 6998
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.695401906967163,
      "learning_rate": 1.8164691059830702e-05,
      "loss": 0.5514,
      "step": 6999
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.013052225112915,
      "learning_rate": 1.816410299835679e-05,
      "loss": 0.5615,
      "step": 7000
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7563198804855347,
      "learning_rate": 1.8163514852208076e-05,
      "loss": 0.5741,
      "step": 7001
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6623119115829468,
      "learning_rate": 1.8162926621390658e-05,
      "loss": 0.4888,
      "step": 7002
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8370062112808228,
      "learning_rate": 1.8162338305910636e-05,
      "loss": 0.5855,
      "step": 7003
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.693699836730957,
      "learning_rate": 1.8161749905774116e-05,
      "loss": 0.4912,
      "step": 7004
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.761307716369629,
      "learning_rate": 1.8161161420987198e-05,
      "loss": 0.5545,
      "step": 7005
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6896954774856567,
      "learning_rate": 1.816057285155598e-05,
      "loss": 0.5745,
      "step": 7006
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.874772071838379,
      "learning_rate": 1.8159984197486577e-05,
      "loss": 0.6723,
      "step": 7007
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.624008297920227,
      "learning_rate": 1.815939545878509e-05,
      "loss": 0.4952,
      "step": 7008
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7967644929885864,
      "learning_rate": 1.8158806635457625e-05,
      "loss": 0.6025,
      "step": 7009
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8307063579559326,
      "learning_rate": 1.8158217727510288e-05,
      "loss": 0.5227,
      "step": 7010
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.714799165725708,
      "learning_rate": 1.8157628734949188e-05,
      "loss": 0.5611,
      "step": 7011
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.669490098953247,
      "learning_rate": 1.815703965778043e-05,
      "loss": 0.4925,
      "step": 7012
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0810413360595703,
      "learning_rate": 1.815645049601013e-05,
      "loss": 1.1027,
      "step": 7013
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8563995361328125,
      "learning_rate": 1.8155861249644394e-05,
      "loss": 1.0651,
      "step": 7014
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0262420177459717,
      "learning_rate": 1.8155271918689335e-05,
      "loss": 0.572,
      "step": 7015
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.842007040977478,
      "learning_rate": 1.8154682503151067e-05,
      "loss": 0.5183,
      "step": 7016
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.760209560394287,
      "learning_rate": 1.81540930030357e-05,
      "loss": 0.5213,
      "step": 7017
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.774851679801941,
      "learning_rate": 1.815350341834935e-05,
      "loss": 0.4954,
      "step": 7018
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.782604455947876,
      "learning_rate": 1.815291374909813e-05,
      "loss": 0.6089,
      "step": 7019
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7957018613815308,
      "learning_rate": 1.815232399528816e-05,
      "loss": 0.5528,
      "step": 7020
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7560099363327026,
      "learning_rate": 1.8151734156925553e-05,
      "loss": 0.5517,
      "step": 7021
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5333878993988037,
      "learning_rate": 1.8151144234016426e-05,
      "loss": 0.4735,
      "step": 7022
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.791955828666687,
      "learning_rate": 1.81505542265669e-05,
      "loss": 0.6485,
      "step": 7023
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7002164125442505,
      "learning_rate": 1.8149964134583092e-05,
      "loss": 0.4838,
      "step": 7024
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8311148881912231,
      "learning_rate": 1.8149373958071125e-05,
      "loss": 0.5958,
      "step": 7025
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6168240308761597,
      "learning_rate": 1.814878369703712e-05,
      "loss": 0.4762,
      "step": 7026
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6867265701293945,
      "learning_rate": 1.8148193351487195e-05,
      "loss": 0.5554,
      "step": 7027
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6665265560150146,
      "learning_rate": 1.8147602921427476e-05,
      "loss": 0.4876,
      "step": 7028
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8930681943893433,
      "learning_rate": 1.8147012406864087e-05,
      "loss": 0.5659,
      "step": 7029
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6260724067687988,
      "learning_rate": 1.814642180780315e-05,
      "loss": 0.4864,
      "step": 7030
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6956552267074585,
      "learning_rate": 1.8145831124250795e-05,
      "loss": 0.4888,
      "step": 7031
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7508220672607422,
      "learning_rate": 1.814524035621314e-05,
      "loss": 0.5419,
      "step": 7032
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6935243606567383,
      "learning_rate": 1.814464950369632e-05,
      "loss": 0.5365,
      "step": 7033
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6589046716690063,
      "learning_rate": 1.8144058566706465e-05,
      "loss": 0.5021,
      "step": 7034
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7259585857391357,
      "learning_rate": 1.8143467545249694e-05,
      "loss": 0.6037,
      "step": 7035
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6989177465438843,
      "learning_rate": 1.8142876439332147e-05,
      "loss": 0.5486,
      "step": 7036
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8281054496765137,
      "learning_rate": 1.8142285248959944e-05,
      "loss": 0.6088,
      "step": 7037
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5645643472671509,
      "learning_rate": 1.814169397413923e-05,
      "loss": 0.4723,
      "step": 7038
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6371777057647705,
      "learning_rate": 1.8141102614876125e-05,
      "loss": 0.5294,
      "step": 7039
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6682215929031372,
      "learning_rate": 1.814051117117677e-05,
      "loss": 0.4957,
      "step": 7040
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7558348178863525,
      "learning_rate": 1.8139919643047296e-05,
      "loss": 0.5896,
      "step": 7041
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.763343095779419,
      "learning_rate": 1.8139328030493838e-05,
      "loss": 0.5611,
      "step": 7042
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7185598611831665,
      "learning_rate": 1.8138736333522535e-05,
      "loss": 0.5682,
      "step": 7043
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6604795455932617,
      "learning_rate": 1.8138144552139523e-05,
      "loss": 0.4911,
      "step": 7044
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.67989981174469,
      "learning_rate": 1.8137552686350935e-05,
      "loss": 0.5356,
      "step": 7045
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7093654870986938,
      "learning_rate": 1.813696073616291e-05,
      "loss": 0.4862,
      "step": 7046
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7534785270690918,
      "learning_rate": 1.81363687015816e-05,
      "loss": 0.5787,
      "step": 7047
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.801560640335083,
      "learning_rate": 1.8135776582613128e-05,
      "loss": 0.5276,
      "step": 7048
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8840361833572388,
      "learning_rate": 1.813518437926365e-05,
      "loss": 0.6374,
      "step": 7049
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6916693449020386,
      "learning_rate": 1.8134592091539293e-05,
      "loss": 0.474,
      "step": 7050
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6792455911636353,
      "learning_rate": 1.8133999719446214e-05,
      "loss": 0.5436,
      "step": 7051
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6984163522720337,
      "learning_rate": 1.8133407262990548e-05,
      "loss": 0.4997,
      "step": 7052
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6674655675888062,
      "learning_rate": 1.8132814722178443e-05,
      "loss": 0.5259,
      "step": 7053
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6233410835266113,
      "learning_rate": 1.8132222097016046e-05,
      "loss": 0.4936,
      "step": 7054
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7625553607940674,
      "learning_rate": 1.81316293875095e-05,
      "loss": 0.5433,
      "step": 7055
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7707651853561401,
      "learning_rate": 1.8131036593664954e-05,
      "loss": 0.5274,
      "step": 7056
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7857255935668945,
      "learning_rate": 1.813044371548856e-05,
      "loss": 0.5527,
      "step": 7057
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7347025871276855,
      "learning_rate": 1.812985075298646e-05,
      "loss": 0.5258,
      "step": 7058
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.69205641746521,
      "learning_rate": 1.812925770616481e-05,
      "loss": 0.5568,
      "step": 7059
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7131251096725464,
      "learning_rate": 1.8128664575029753e-05,
      "loss": 0.5143,
      "step": 7060
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8083521127700806,
      "learning_rate": 1.812807135958745e-05,
      "loss": 0.5745,
      "step": 7061
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5885400772094727,
      "learning_rate": 1.812747805984405e-05,
      "loss": 0.4748,
      "step": 7062
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7153400182724,
      "learning_rate": 1.8126884675805705e-05,
      "loss": 0.5564,
      "step": 7063
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6931978464126587,
      "learning_rate": 1.8126291207478567e-05,
      "loss": 0.5154,
      "step": 7064
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8245937824249268,
      "learning_rate": 1.81256976548688e-05,
      "loss": 0.6562,
      "step": 7065
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6644680500030518,
      "learning_rate": 1.8125104017982555e-05,
      "loss": 0.4856,
      "step": 7066
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6966149806976318,
      "learning_rate": 1.8124510296825982e-05,
      "loss": 0.5483,
      "step": 7067
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6875417232513428,
      "learning_rate": 1.812391649140525e-05,
      "loss": 0.5676,
      "step": 7068
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6671026945114136,
      "learning_rate": 1.812332260172651e-05,
      "loss": 0.5049,
      "step": 7069
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7496951818466187,
      "learning_rate": 1.812272862779593e-05,
      "loss": 0.5937,
      "step": 7070
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8170572519302368,
      "learning_rate": 1.8122134569619666e-05,
      "loss": 0.5543,
      "step": 7071
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.662964940071106,
      "learning_rate": 1.8121540427203874e-05,
      "loss": 0.4944,
      "step": 7072
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8347080945968628,
      "learning_rate": 1.8120946200554722e-05,
      "loss": 0.5743,
      "step": 7073
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6317936182022095,
      "learning_rate": 1.812035188967837e-05,
      "loss": 0.4611,
      "step": 7074
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6640173196792603,
      "learning_rate": 1.8119757494580987e-05,
      "loss": 0.5358,
      "step": 7075
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8200851678848267,
      "learning_rate": 1.8119163015268734e-05,
      "loss": 0.5539,
      "step": 7076
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7644712924957275,
      "learning_rate": 1.8118568451747776e-05,
      "loss": 0.5559,
      "step": 7077
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.843052625656128,
      "learning_rate": 1.811797380402428e-05,
      "loss": 0.5333,
      "step": 7078
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.714255928993225,
      "learning_rate": 1.8117379072104418e-05,
      "loss": 0.5611,
      "step": 7079
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6798951625823975,
      "learning_rate": 1.811678425599435e-05,
      "loss": 0.5369,
      "step": 7080
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.734866976737976,
      "learning_rate": 1.8116189355700252e-05,
      "loss": 0.6429,
      "step": 7081
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8239939212799072,
      "learning_rate": 1.8115594371228296e-05,
      "loss": 0.5873,
      "step": 7082
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7852411270141602,
      "learning_rate": 1.8114999302584644e-05,
      "loss": 0.5946,
      "step": 7083
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.67238187789917,
      "learning_rate": 1.8114404149775474e-05,
      "loss": 0.4969,
      "step": 7084
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6680035591125488,
      "learning_rate": 1.8113808912806958e-05,
      "loss": 0.5758,
      "step": 7085
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6514642238616943,
      "learning_rate": 1.8113213591685267e-05,
      "loss": 0.4948,
      "step": 7086
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8012828826904297,
      "learning_rate": 1.811261818641658e-05,
      "loss": 0.6037,
      "step": 7087
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.787521481513977,
      "learning_rate": 1.8112022697007067e-05,
      "loss": 0.5043,
      "step": 7088
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8105854988098145,
      "learning_rate": 1.8111427123462907e-05,
      "loss": 0.5614,
      "step": 7089
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.735979676246643,
      "learning_rate": 1.8110831465790283e-05,
      "loss": 0.5346,
      "step": 7090
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7047196626663208,
      "learning_rate": 1.811023572399536e-05,
      "loss": 0.5298,
      "step": 7091
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6030532121658325,
      "learning_rate": 1.8109639898084325e-05,
      "loss": 0.4744,
      "step": 7092
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7842296361923218,
      "learning_rate": 1.8109043988063357e-05,
      "loss": 0.6276,
      "step": 7093
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.685945749282837,
      "learning_rate": 1.8108447993938635e-05,
      "loss": 0.5102,
      "step": 7094
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6842679977416992,
      "learning_rate": 1.8107851915716344e-05,
      "loss": 0.529,
      "step": 7095
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6774890422821045,
      "learning_rate": 1.810725575340266e-05,
      "loss": 0.5082,
      "step": 7096
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.693118929862976,
      "learning_rate": 1.810665950700377e-05,
      "loss": 0.5606,
      "step": 7097
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6705750226974487,
      "learning_rate": 1.8106063176525862e-05,
      "loss": 0.5157,
      "step": 7098
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7712703943252563,
      "learning_rate": 1.810546676197511e-05,
      "loss": 0.5639,
      "step": 7099
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7695155143737793,
      "learning_rate": 1.8104870263357713e-05,
      "loss": 0.5146,
      "step": 7100
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.804646372795105,
      "learning_rate": 1.8104273680679845e-05,
      "loss": 0.5927,
      "step": 7101
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7208133935928345,
      "learning_rate": 1.8103677013947703e-05,
      "loss": 0.5097,
      "step": 7102
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7412513494491577,
      "learning_rate": 1.8103080263167473e-05,
      "loss": 0.5325,
      "step": 7103
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6073780059814453,
      "learning_rate": 1.810248342834534e-05,
      "loss": 0.4719,
      "step": 7104
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8435933589935303,
      "learning_rate": 1.8101886509487497e-05,
      "loss": 0.562,
      "step": 7105
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7508753538131714,
      "learning_rate": 1.810128950660014e-05,
      "loss": 0.5074,
      "step": 7106
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7144116163253784,
      "learning_rate": 1.8100692419689453e-05,
      "loss": 0.5696,
      "step": 7107
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.802199125289917,
      "learning_rate": 1.810009524876163e-05,
      "loss": 0.5697,
      "step": 7108
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7949497699737549,
      "learning_rate": 1.809949799382287e-05,
      "loss": 0.5875,
      "step": 7109
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7319084405899048,
      "learning_rate": 1.8098900654879363e-05,
      "loss": 0.5418,
      "step": 7110
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8236572742462158,
      "learning_rate": 1.8098303231937303e-05,
      "loss": 0.6004,
      "step": 7111
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7926174402236938,
      "learning_rate": 1.8097705725002894e-05,
      "loss": 0.4772,
      "step": 7112
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7889872789382935,
      "learning_rate": 1.8097108134082326e-05,
      "loss": 0.5838,
      "step": 7113
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6182122230529785,
      "learning_rate": 1.8096510459181795e-05,
      "loss": 0.4828,
      "step": 7114
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6492598056793213,
      "learning_rate": 1.8095912700307505e-05,
      "loss": 0.5625,
      "step": 7115
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7921255826950073,
      "learning_rate": 1.8095314857465657e-05,
      "loss": 0.5782,
      "step": 7116
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8114166259765625,
      "learning_rate": 1.8094716930662447e-05,
      "loss": 0.6048,
      "step": 7117
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7167772054672241,
      "learning_rate": 1.809411891990408e-05,
      "loss": 0.5585,
      "step": 7118
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6056796312332153,
      "learning_rate": 1.8093520825196753e-05,
      "loss": 0.4967,
      "step": 7119
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7209175825119019,
      "learning_rate": 1.8092922646546674e-05,
      "loss": 0.5624,
      "step": 7120
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6808031797409058,
      "learning_rate": 1.8092324383960047e-05,
      "loss": 0.5677,
      "step": 7121
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7717623710632324,
      "learning_rate": 1.8091726037443077e-05,
      "loss": 0.5711,
      "step": 7122
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7170544862747192,
      "learning_rate": 1.8091127607001964e-05,
      "loss": 0.5869,
      "step": 7123
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7302333116531372,
      "learning_rate": 1.8090529092642926e-05,
      "loss": 0.4924,
      "step": 7124
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.77972412109375,
      "learning_rate": 1.808993049437216e-05,
      "loss": 0.5408,
      "step": 7125
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.783896803855896,
      "learning_rate": 1.808933181219588e-05,
      "loss": 0.5618,
      "step": 7126
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7158797979354858,
      "learning_rate": 1.8088733046120292e-05,
      "loss": 0.5641,
      "step": 7127
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.724783182144165,
      "learning_rate": 1.8088134196151608e-05,
      "loss": 0.5398,
      "step": 7128
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5932947397232056,
      "learning_rate": 1.808753526229604e-05,
      "loss": 0.5314,
      "step": 7129
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6552890539169312,
      "learning_rate": 1.8086936244559797e-05,
      "loss": 0.4704,
      "step": 7130
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8249436616897583,
      "learning_rate": 1.8086337142949096e-05,
      "loss": 0.5296,
      "step": 7131
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6853880882263184,
      "learning_rate": 1.8085737957470143e-05,
      "loss": 0.5435,
      "step": 7132
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8044220209121704,
      "learning_rate": 1.808513868812916e-05,
      "loss": 0.5925,
      "step": 7133
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7128870487213135,
      "learning_rate": 1.808453933493236e-05,
      "loss": 0.52,
      "step": 7134
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7423633337020874,
      "learning_rate": 1.8083939897885964e-05,
      "loss": 0.6031,
      "step": 7135
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6997828483581543,
      "learning_rate": 1.808334037699618e-05,
      "loss": 0.5134,
      "step": 7136
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.723467230796814,
      "learning_rate": 1.808274077226923e-05,
      "loss": 0.5549,
      "step": 7137
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.642417311668396,
      "learning_rate": 1.8082141083711334e-05,
      "loss": 0.4972,
      "step": 7138
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7933475971221924,
      "learning_rate": 1.8081541311328713e-05,
      "loss": 0.5488,
      "step": 7139
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7770549058914185,
      "learning_rate": 1.8080941455127584e-05,
      "loss": 0.4982,
      "step": 7140
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7608222961425781,
      "learning_rate": 1.8080341515114172e-05,
      "loss": 0.5525,
      "step": 7141
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7618385553359985,
      "learning_rate": 1.8079741491294693e-05,
      "loss": 0.4778,
      "step": 7142
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7059273719787598,
      "learning_rate": 1.8079141383675378e-05,
      "loss": 0.6085,
      "step": 7143
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8016184568405151,
      "learning_rate": 1.8078541192262444e-05,
      "loss": 0.5601,
      "step": 7144
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.846828818321228,
      "learning_rate": 1.8077940917062126e-05,
      "loss": 0.5644,
      "step": 7145
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6714271306991577,
      "learning_rate": 1.8077340558080637e-05,
      "loss": 0.471,
      "step": 7146
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.832004189491272,
      "learning_rate": 1.8076740115324215e-05,
      "loss": 0.5681,
      "step": 7147
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7126344442367554,
      "learning_rate": 1.8076139588799078e-05,
      "loss": 0.4934,
      "step": 7148
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8419361114501953,
      "learning_rate": 1.807553897851146e-05,
      "loss": 0.6127,
      "step": 7149
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6446924209594727,
      "learning_rate": 1.8074938284467592e-05,
      "loss": 0.4854,
      "step": 7150
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7569959163665771,
      "learning_rate": 1.8074337506673698e-05,
      "loss": 0.5712,
      "step": 7151
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9500935077667236,
      "learning_rate": 1.807373664513602e-05,
      "loss": 0.4957,
      "step": 7152
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8284658193588257,
      "learning_rate": 1.8073135699860773e-05,
      "loss": 0.54,
      "step": 7153
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7657010555267334,
      "learning_rate": 1.8072534670854204e-05,
      "loss": 0.52,
      "step": 7154
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7858798503875732,
      "learning_rate": 1.807193355812254e-05,
      "loss": 0.5544,
      "step": 7155
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6940685510635376,
      "learning_rate": 1.8071332361672018e-05,
      "loss": 0.4896,
      "step": 7156
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0117413997650146,
      "learning_rate": 1.8070731081508874e-05,
      "loss": 0.6354,
      "step": 7157
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7409762144088745,
      "learning_rate": 1.807012971763934e-05,
      "loss": 0.4951,
      "step": 7158
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8915985822677612,
      "learning_rate": 1.806952827006966e-05,
      "loss": 0.5322,
      "step": 7159
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7064095735549927,
      "learning_rate": 1.8068926738806064e-05,
      "loss": 0.5266,
      "step": 7160
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7515641450881958,
      "learning_rate": 1.80683251238548e-05,
      "loss": 0.5558,
      "step": 7161
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7881253957748413,
      "learning_rate": 1.80677234252221e-05,
      "loss": 0.5195,
      "step": 7162
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7372241020202637,
      "learning_rate": 1.8067121642914205e-05,
      "loss": 0.5809,
      "step": 7163
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6038289070129395,
      "learning_rate": 1.806651977693736e-05,
      "loss": 0.4876,
      "step": 7164
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7986725568771362,
      "learning_rate": 1.8065917827297807e-05,
      "loss": 0.5229,
      "step": 7165
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7484527826309204,
      "learning_rate": 1.806531579400179e-05,
      "loss": 0.533,
      "step": 7166
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8577682971954346,
      "learning_rate": 1.8064713677055545e-05,
      "loss": 0.5976,
      "step": 7167
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7461285591125488,
      "learning_rate": 1.8064111476465325e-05,
      "loss": 0.5028,
      "step": 7168
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.1151623725891113,
      "learning_rate": 1.806350919223738e-05,
      "loss": 1.0375,
      "step": 7169
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.938104271888733,
      "learning_rate": 1.8062906824377942e-05,
      "loss": 1.1205,
      "step": 7170
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.911773443222046,
      "learning_rate": 1.806230437289327e-05,
      "loss": 0.5644,
      "step": 7171
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8593846559524536,
      "learning_rate": 1.8061701837789612e-05,
      "loss": 0.4882,
      "step": 7172
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.3611350059509277,
      "learning_rate": 1.806109921907321e-05,
      "loss": 0.5434,
      "step": 7173
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6238943338394165,
      "learning_rate": 1.8060496516750322e-05,
      "loss": 0.4614,
      "step": 7174
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9232375621795654,
      "learning_rate": 1.8059893730827197e-05,
      "loss": 0.5547,
      "step": 7175
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7444409132003784,
      "learning_rate": 1.8059290861310085e-05,
      "loss": 0.5166,
      "step": 7176
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7217806577682495,
      "learning_rate": 1.8058687908205238e-05,
      "loss": 0.5598,
      "step": 7177
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6516749858856201,
      "learning_rate": 1.805808487151891e-05,
      "loss": 0.5178,
      "step": 7178
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.2830810546875,
      "learning_rate": 1.805748175125736e-05,
      "loss": 1.0516,
      "step": 7179
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8924167156219482,
      "learning_rate": 1.8056878547426833e-05,
      "loss": 1.0842,
      "step": 7180
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.989048719406128,
      "learning_rate": 1.8056275260033596e-05,
      "loss": 0.6139,
      "step": 7181
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9551794528961182,
      "learning_rate": 1.8055671889083903e-05,
      "loss": 0.5069,
      "step": 7182
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8811192512512207,
      "learning_rate": 1.8055068434584008e-05,
      "loss": 0.566,
      "step": 7183
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9414292573928833,
      "learning_rate": 1.8054464896540173e-05,
      "loss": 0.5616,
      "step": 7184
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8612313270568848,
      "learning_rate": 1.805386127495866e-05,
      "loss": 0.5337,
      "step": 7185
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7926828861236572,
      "learning_rate": 1.8053257569845723e-05,
      "loss": 0.4965,
      "step": 7186
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8565012216567993,
      "learning_rate": 1.805265378120763e-05,
      "loss": 1.0228,
      "step": 7187
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7204948663711548,
      "learning_rate": 1.8052049909050643e-05,
      "loss": 1.1069,
      "step": 7188
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.6697347164154053,
      "learning_rate": 1.8051445953381018e-05,
      "loss": 0.5852,
      "step": 7189
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.5275769233703613,
      "learning_rate": 1.8050841914205027e-05,
      "loss": 0.4863,
      "step": 7190
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.058619976043701,
      "learning_rate": 1.8050237791528928e-05,
      "loss": 0.5244,
      "step": 7191
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7649286985397339,
      "learning_rate": 1.8049633585358992e-05,
      "loss": 0.4648,
      "step": 7192
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8118400573730469,
      "learning_rate": 1.8049029295701484e-05,
      "loss": 0.5824,
      "step": 7193
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9052748680114746,
      "learning_rate": 1.804842492256267e-05,
      "loss": 0.4635,
      "step": 7194
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0074665546417236,
      "learning_rate": 1.8047820465948818e-05,
      "loss": 0.5105,
      "step": 7195
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9040476083755493,
      "learning_rate": 1.8047215925866202e-05,
      "loss": 0.5294,
      "step": 7196
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.947374939918518,
      "learning_rate": 1.8046611302321086e-05,
      "loss": 0.5668,
      "step": 7197
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.910441517829895,
      "learning_rate": 1.8046006595319746e-05,
      "loss": 0.5044,
      "step": 7198
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8835421800613403,
      "learning_rate": 1.804540180486845e-05,
      "loss": 0.5412,
      "step": 7199
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7890702486038208,
      "learning_rate": 1.804479693097347e-05,
      "loss": 0.5194,
      "step": 7200
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7824723720550537,
      "learning_rate": 1.8044191973641084e-05,
      "loss": 0.5434,
      "step": 7201
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.064072847366333,
      "learning_rate": 1.8043586932877564e-05,
      "loss": 0.5695,
      "step": 7202
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7241668701171875,
      "learning_rate": 1.8042981808689187e-05,
      "loss": 0.5489,
      "step": 7203
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8091251850128174,
      "learning_rate": 1.8042376601082226e-05,
      "loss": 0.488,
      "step": 7204
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6504143476486206,
      "learning_rate": 1.8041771310062958e-05,
      "loss": 0.5369,
      "step": 7205
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.847688913345337,
      "learning_rate": 1.8041165935637662e-05,
      "loss": 0.5288,
      "step": 7206
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.908011794090271,
      "learning_rate": 1.8040560477812616e-05,
      "loss": 0.5489,
      "step": 7207
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8304756879806519,
      "learning_rate": 1.8039954936594102e-05,
      "loss": 0.5122,
      "step": 7208
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8833974599838257,
      "learning_rate": 1.80393493119884e-05,
      "loss": 0.5869,
      "step": 7209
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8225256204605103,
      "learning_rate": 1.8038743604001788e-05,
      "loss": 0.5125,
      "step": 7210
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7915904521942139,
      "learning_rate": 1.8038137812640547e-05,
      "loss": 0.5482,
      "step": 7211
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7780494689941406,
      "learning_rate": 1.803753193791097e-05,
      "loss": 0.4457,
      "step": 7212
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9734954833984375,
      "learning_rate": 1.803692597981933e-05,
      "loss": 0.5493,
      "step": 7213
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7961403131484985,
      "learning_rate": 1.8036319938371916e-05,
      "loss": 0.531,
      "step": 7214
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0072271823883057,
      "learning_rate": 1.8035713813575017e-05,
      "loss": 1.0357,
      "step": 7215
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7116308212280273,
      "learning_rate": 1.803510760543491e-05,
      "loss": 1.1085,
      "step": 7216
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9314415454864502,
      "learning_rate": 1.8034501313957892e-05,
      "loss": 0.6343,
      "step": 7217
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9575506448745728,
      "learning_rate": 1.8033894939150247e-05,
      "loss": 0.5312,
      "step": 7218
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7273327112197876,
      "learning_rate": 1.8033288481018267e-05,
      "loss": 0.5582,
      "step": 7219
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7161040306091309,
      "learning_rate": 1.8032681939568236e-05,
      "loss": 0.5306,
      "step": 7220
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8617808818817139,
      "learning_rate": 1.803207531480645e-05,
      "loss": 0.6013,
      "step": 7221
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6842962503433228,
      "learning_rate": 1.80314686067392e-05,
      "loss": 0.4732,
      "step": 7222
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7992933988571167,
      "learning_rate": 1.8030861815372777e-05,
      "loss": 0.5432,
      "step": 7223
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7151929140090942,
      "learning_rate": 1.8030254940713473e-05,
      "loss": 0.5071,
      "step": 7224
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7340244054794312,
      "learning_rate": 1.8029647982767588e-05,
      "loss": 0.534,
      "step": 7225
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8749185800552368,
      "learning_rate": 1.802904094154141e-05,
      "loss": 0.5232,
      "step": 7226
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8184667825698853,
      "learning_rate": 1.8028433817041237e-05,
      "loss": 0.5639,
      "step": 7227
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7371903657913208,
      "learning_rate": 1.8027826609273372e-05,
      "loss": 0.4892,
      "step": 7228
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6535102128982544,
      "learning_rate": 1.8027219318244103e-05,
      "loss": 0.5647,
      "step": 7229
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7621898651123047,
      "learning_rate": 1.802661194395974e-05,
      "loss": 0.4945,
      "step": 7230
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7346829175949097,
      "learning_rate": 1.802600448642657e-05,
      "loss": 0.5509,
      "step": 7231
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8042970895767212,
      "learning_rate": 1.80253969456509e-05,
      "loss": 0.5245,
      "step": 7232
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.832358479499817,
      "learning_rate": 1.8024789321639034e-05,
      "loss": 0.577,
      "step": 7233
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7618330717086792,
      "learning_rate": 1.8024181614397267e-05,
      "loss": 0.4832,
      "step": 7234
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7198312282562256,
      "learning_rate": 1.8023573823931907e-05,
      "loss": 0.51,
      "step": 7235
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6179968118667603,
      "learning_rate": 1.8022965950249256e-05,
      "loss": 0.5767,
      "step": 7236
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.64688241481781,
      "learning_rate": 1.8022357993355617e-05,
      "loss": 0.5386,
      "step": 7237
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.635763168334961,
      "learning_rate": 1.80217499532573e-05,
      "loss": 0.5369,
      "step": 7238
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.749976396560669,
      "learning_rate": 1.8021141829960607e-05,
      "loss": 0.5488,
      "step": 7239
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6185765266418457,
      "learning_rate": 1.8020533623471847e-05,
      "loss": 0.4731,
      "step": 7240
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7463833093643188,
      "learning_rate": 1.8019925333797326e-05,
      "loss": 0.5405,
      "step": 7241
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6934070587158203,
      "learning_rate": 1.8019316960943357e-05,
      "loss": 0.5053,
      "step": 7242
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6331088542938232,
      "learning_rate": 1.8018708504916248e-05,
      "loss": 0.5836,
      "step": 7243
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7830743789672852,
      "learning_rate": 1.8018099965722307e-05,
      "loss": 0.5435,
      "step": 7244
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7139325141906738,
      "learning_rate": 1.801749134336785e-05,
      "loss": 0.5471,
      "step": 7245
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6356335878372192,
      "learning_rate": 1.8016882637859185e-05,
      "loss": 0.4748,
      "step": 7246
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.82278311252594,
      "learning_rate": 1.8016273849202627e-05,
      "loss": 0.5859,
      "step": 7247
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6586874723434448,
      "learning_rate": 1.8015664977404493e-05,
      "loss": 0.487,
      "step": 7248
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.079298496246338,
      "learning_rate": 1.8015056022471092e-05,
      "loss": 0.9953,
      "step": 7249
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6542540788650513,
      "learning_rate": 1.801444698440875e-05,
      "loss": 1.079,
      "step": 7250
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9078859090805054,
      "learning_rate": 1.801383786322377e-05,
      "loss": 0.5959,
      "step": 7251
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.930836796760559,
      "learning_rate": 1.801322865892248e-05,
      "loss": 0.4817,
      "step": 7252
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7179030179977417,
      "learning_rate": 1.8012619371511195e-05,
      "loss": 0.5353,
      "step": 7253
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6994295120239258,
      "learning_rate": 1.8012010000996233e-05,
      "loss": 0.5058,
      "step": 7254
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9642499685287476,
      "learning_rate": 1.8011400547383917e-05,
      "loss": 1.037,
      "step": 7255
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5838834047317505,
      "learning_rate": 1.8010791010680566e-05,
      "loss": 1.0781,
      "step": 7256
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9232957363128662,
      "learning_rate": 1.8010181390892502e-05,
      "loss": 0.5869,
      "step": 7257
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.145225763320923,
      "learning_rate": 1.8009571688026047e-05,
      "loss": 0.4982,
      "step": 7258
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.046860456466675,
      "learning_rate": 1.8008961902087527e-05,
      "loss": 0.5859,
      "step": 7259
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.833404302597046,
      "learning_rate": 1.8008352033083264e-05,
      "loss": 0.6146,
      "step": 7260
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7585862874984741,
      "learning_rate": 1.8007742081019584e-05,
      "loss": 0.5716,
      "step": 7261
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7414778470993042,
      "learning_rate": 1.8007132045902816e-05,
      "loss": 0.4879,
      "step": 7262
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9117929935455322,
      "learning_rate": 1.8006521927739285e-05,
      "loss": 0.5625,
      "step": 7263
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6697906255722046,
      "learning_rate": 1.8005911726535317e-05,
      "loss": 0.4608,
      "step": 7264
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7055439949035645,
      "learning_rate": 1.8005301442297242e-05,
      "loss": 0.5368,
      "step": 7265
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.73048734664917,
      "learning_rate": 1.800469107503139e-05,
      "loss": 0.4947,
      "step": 7266
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7695996761322021,
      "learning_rate": 1.800408062474409e-05,
      "loss": 0.5562,
      "step": 7267
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7462931871414185,
      "learning_rate": 1.8003470091441678e-05,
      "loss": 0.4894,
      "step": 7268
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7198151350021362,
      "learning_rate": 1.8002859475130482e-05,
      "loss": 0.568,
      "step": 7269
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.708520770072937,
      "learning_rate": 1.8002248775816836e-05,
      "loss": 0.5026,
      "step": 7270
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7684382200241089,
      "learning_rate": 1.800163799350707e-05,
      "loss": 0.5705,
      "step": 7271
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7029401063919067,
      "learning_rate": 1.8001027128207527e-05,
      "loss": 0.541,
      "step": 7272
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.683203101158142,
      "learning_rate": 1.8000416179924535e-05,
      "loss": 0.5243,
      "step": 7273
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7441678047180176,
      "learning_rate": 1.799980514866444e-05,
      "loss": 0.5303,
      "step": 7274
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.70918607711792,
      "learning_rate": 1.7999194034433565e-05,
      "loss": 0.556,
      "step": 7275
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6458162069320679,
      "learning_rate": 1.7998582837238263e-05,
      "loss": 0.4843,
      "step": 7276
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.789638876914978,
      "learning_rate": 1.7997971557084865e-05,
      "loss": 0.6094,
      "step": 7277
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6918182373046875,
      "learning_rate": 1.7997360193979712e-05,
      "loss": 0.4847,
      "step": 7278
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7522245645523071,
      "learning_rate": 1.799674874792915e-05,
      "loss": 0.5698,
      "step": 7279
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.751429557800293,
      "learning_rate": 1.7996137218939507e-05,
      "loss": 0.5219,
      "step": 7280
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6231753826141357,
      "learning_rate": 1.7995525607017138e-05,
      "loss": 0.54,
      "step": 7281
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6410486698150635,
      "learning_rate": 1.7994913912168387e-05,
      "loss": 0.5253,
      "step": 7282
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.70113205909729,
      "learning_rate": 1.7994302134399592e-05,
      "loss": 0.5447,
      "step": 7283
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7763599157333374,
      "learning_rate": 1.7993690273717097e-05,
      "loss": 0.514,
      "step": 7284
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8408647775650024,
      "learning_rate": 1.7993078330127255e-05,
      "loss": 0.5848,
      "step": 7285
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6154505014419556,
      "learning_rate": 1.7992466303636407e-05,
      "loss": 0.4701,
      "step": 7286
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7152562141418457,
      "learning_rate": 1.7991854194250903e-05,
      "loss": 0.5814,
      "step": 7287
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6888678073883057,
      "learning_rate": 1.7991242001977093e-05,
      "loss": 0.5054,
      "step": 7288
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8378978967666626,
      "learning_rate": 1.7990629726821326e-05,
      "loss": 0.6159,
      "step": 7289
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6416226625442505,
      "learning_rate": 1.7990017368789948e-05,
      "loss": 0.4882,
      "step": 7290
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7131956815719604,
      "learning_rate": 1.7989404927889315e-05,
      "loss": 0.5377,
      "step": 7291
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5796077251434326,
      "learning_rate": 1.7988792404125773e-05,
      "loss": 0.4786,
      "step": 7292
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6539431810379028,
      "learning_rate": 1.7988179797505685e-05,
      "loss": 0.4984,
      "step": 7293
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7531243562698364,
      "learning_rate": 1.7987567108035397e-05,
      "loss": 0.5177,
      "step": 7294
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.196920394897461,
      "learning_rate": 1.798695433572126e-05,
      "loss": 1.0162,
      "step": 7295
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8409230709075928,
      "learning_rate": 1.7986341480569645e-05,
      "loss": 1.0599,
      "step": 7296
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8794968128204346,
      "learning_rate": 1.798572854258689e-05,
      "loss": 0.5365,
      "step": 7297
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8233870267868042,
      "learning_rate": 1.7985115521779362e-05,
      "loss": 0.5048,
      "step": 7298
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7665293216705322,
      "learning_rate": 1.798450241815342e-05,
      "loss": 0.5724,
      "step": 7299
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6816905736923218,
      "learning_rate": 1.7983889231715416e-05,
      "loss": 0.496,
      "step": 7300
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8754538297653198,
      "learning_rate": 1.798327596247172e-05,
      "loss": 0.5692,
      "step": 7301
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7452229261398315,
      "learning_rate": 1.798266261042868e-05,
      "loss": 0.5238,
      "step": 7302
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7861381769180298,
      "learning_rate": 1.7982049175592667e-05,
      "loss": 0.5812,
      "step": 7303
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6642111539840698,
      "learning_rate": 1.7981435657970036e-05,
      "loss": 0.5086,
      "step": 7304
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7201976776123047,
      "learning_rate": 1.7980822057567157e-05,
      "loss": 0.6107,
      "step": 7305
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7175607681274414,
      "learning_rate": 1.7980208374390392e-05,
      "loss": 0.489,
      "step": 7306
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7225812673568726,
      "learning_rate": 1.7979594608446104e-05,
      "loss": 0.5499,
      "step": 7307
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6205146312713623,
      "learning_rate": 1.7978980759740663e-05,
      "loss": 0.498,
      "step": 7308
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.784474492073059,
      "learning_rate": 1.7978366828280428e-05,
      "loss": 0.5557,
      "step": 7309
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7641489505767822,
      "learning_rate": 1.7977752814071773e-05,
      "loss": 0.4972,
      "step": 7310
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7552103996276855,
      "learning_rate": 1.7977138717121066e-05,
      "loss": 0.6073,
      "step": 7311
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7024019956588745,
      "learning_rate": 1.7976524537434672e-05,
      "loss": 0.4874,
      "step": 7312
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7722123861312866,
      "learning_rate": 1.7975910275018963e-05,
      "loss": 0.5868,
      "step": 7313
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7640624046325684,
      "learning_rate": 1.797529592988031e-05,
      "loss": 0.4779,
      "step": 7314
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7953423261642456,
      "learning_rate": 1.7974681502025085e-05,
      "loss": 0.5517,
      "step": 7315
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7841681241989136,
      "learning_rate": 1.7974066991459666e-05,
      "loss": 0.5268,
      "step": 7316
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.722885012626648,
      "learning_rate": 1.7973452398190418e-05,
      "loss": 0.5419,
      "step": 7317
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.84149968624115,
      "learning_rate": 1.7972837722223716e-05,
      "loss": 0.498,
      "step": 7318
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.656962275505066,
      "learning_rate": 1.797222296356594e-05,
      "loss": 0.5566,
      "step": 7319
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7045751810073853,
      "learning_rate": 1.7971608122223464e-05,
      "loss": 0.5012,
      "step": 7320
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7218148708343506,
      "learning_rate": 1.7970993198202662e-05,
      "loss": 0.5773,
      "step": 7321
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6140472888946533,
      "learning_rate": 1.7970378191509917e-05,
      "loss": 0.4722,
      "step": 7322
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7819961309432983,
      "learning_rate": 1.79697631021516e-05,
      "loss": 0.5785,
      "step": 7323
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6475201845169067,
      "learning_rate": 1.79691479301341e-05,
      "loss": 0.4836,
      "step": 7324
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8099430799484253,
      "learning_rate": 1.796853267546379e-05,
      "loss": 0.547,
      "step": 7325
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8056811094284058,
      "learning_rate": 1.796791733814706e-05,
      "loss": 0.5506,
      "step": 7326
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6750174760818481,
      "learning_rate": 1.7967301918190277e-05,
      "loss": 0.5127,
      "step": 7327
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8047205209732056,
      "learning_rate": 1.796668641559984e-05,
      "loss": 0.551,
      "step": 7328
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6804859638214111,
      "learning_rate": 1.796607083038212e-05,
      "loss": 0.5388,
      "step": 7329
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.659283995628357,
      "learning_rate": 1.796545516254351e-05,
      "loss": 0.4799,
      "step": 7330
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.884495496749878,
      "learning_rate": 1.796483941209039e-05,
      "loss": 0.6257,
      "step": 7331
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.758165955543518,
      "learning_rate": 1.7964223579029154e-05,
      "loss": 0.5066,
      "step": 7332
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8005268573760986,
      "learning_rate": 1.796360766336618e-05,
      "loss": 0.5682,
      "step": 7333
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7285511493682861,
      "learning_rate": 1.796299166510786e-05,
      "loss": 0.4777,
      "step": 7334
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6911929845809937,
      "learning_rate": 1.7962375584260583e-05,
      "loss": 0.5188,
      "step": 7335
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6168028116226196,
      "learning_rate": 1.796175942083074e-05,
      "loss": 0.4822,
      "step": 7336
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9971626996994019,
      "learning_rate": 1.796114317482472e-05,
      "loss": 0.9709,
      "step": 7337
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.791725754737854,
      "learning_rate": 1.7960526846248916e-05,
      "loss": 1.0144,
      "step": 7338
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8617130517959595,
      "learning_rate": 1.7959910435109717e-05,
      "loss": 0.5246,
      "step": 7339
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7521530389785767,
      "learning_rate": 1.7959293941413522e-05,
      "loss": 0.4913,
      "step": 7340
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.856066107749939,
      "learning_rate": 1.7958677365166717e-05,
      "loss": 0.9829,
      "step": 7341
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.677956223487854,
      "learning_rate": 1.7958060706375704e-05,
      "loss": 1.0113,
      "step": 7342
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9350097179412842,
      "learning_rate": 1.7957443965046876e-05,
      "loss": 0.5699,
      "step": 7343
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7876152992248535,
      "learning_rate": 1.7956827141186624e-05,
      "loss": 0.4701,
      "step": 7344
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7101812362670898,
      "learning_rate": 1.7956210234801358e-05,
      "loss": 0.5294,
      "step": 7345
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.721940517425537,
      "learning_rate": 1.7955593245897465e-05,
      "loss": 0.514,
      "step": 7346
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8253599405288696,
      "learning_rate": 1.7954976174481348e-05,
      "loss": 0.5333,
      "step": 7347
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.725409984588623,
      "learning_rate": 1.795435902055941e-05,
      "loss": 0.4934,
      "step": 7348
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7045860290527344,
      "learning_rate": 1.7953741784138047e-05,
      "loss": 0.5646,
      "step": 7349
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.569276213645935,
      "learning_rate": 1.7953124465223666e-05,
      "loss": 0.4938,
      "step": 7350
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8856106996536255,
      "learning_rate": 1.7952507063822662e-05,
      "loss": 0.5835,
      "step": 7351
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8020315170288086,
      "learning_rate": 1.795188957994145e-05,
      "loss": 0.549,
      "step": 7352
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7704330682754517,
      "learning_rate": 1.7951272013586424e-05,
      "loss": 0.577,
      "step": 7353
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6684162616729736,
      "learning_rate": 1.795065436476399e-05,
      "loss": 0.5203,
      "step": 7354
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8510113954544067,
      "learning_rate": 1.7950036633480557e-05,
      "loss": 0.5458,
      "step": 7355
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7242082357406616,
      "learning_rate": 1.7949418819742532e-05,
      "loss": 0.4641,
      "step": 7356
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8051338195800781,
      "learning_rate": 1.7948800923556323e-05,
      "loss": 0.5809,
      "step": 7357
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6390931606292725,
      "learning_rate": 1.794818294492834e-05,
      "loss": 0.5038,
      "step": 7358
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7888153791427612,
      "learning_rate": 1.7947564883864985e-05,
      "loss": 0.544,
      "step": 7359
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6339975595474243,
      "learning_rate": 1.7946946740372676e-05,
      "loss": 0.5192,
      "step": 7360
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6343238353729248,
      "learning_rate": 1.794632851445782e-05,
      "loss": 0.4952,
      "step": 7361
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6953468322753906,
      "learning_rate": 1.794571020612683e-05,
      "loss": 0.5376,
      "step": 7362
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8872203826904297,
      "learning_rate": 1.7945091815386123e-05,
      "loss": 0.6088,
      "step": 7363
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6352300643920898,
      "learning_rate": 1.7944473342242107e-05,
      "loss": 0.4949,
      "step": 7364
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7379218339920044,
      "learning_rate": 1.79438547867012e-05,
      "loss": 0.5745,
      "step": 7365
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7926788330078125,
      "learning_rate": 1.7943236148769814e-05,
      "loss": 0.4895,
      "step": 7366
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7568813562393188,
      "learning_rate": 1.794261742845437e-05,
      "loss": 0.5689,
      "step": 7367
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.627590298652649,
      "learning_rate": 1.794199862576128e-05,
      "loss": 0.4835,
      "step": 7368
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.884342074394226,
      "learning_rate": 1.7941379740696963e-05,
      "loss": 0.6081,
      "step": 7369
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6673105955123901,
      "learning_rate": 1.7940760773267843e-05,
      "loss": 0.4871,
      "step": 7370
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7891236543655396,
      "learning_rate": 1.794014172348033e-05,
      "loss": 0.5892,
      "step": 7371
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7468582391738892,
      "learning_rate": 1.7939522591340856e-05,
      "loss": 0.5271,
      "step": 7372
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7547633647918701,
      "learning_rate": 1.7938903376855834e-05,
      "loss": 0.5623,
      "step": 7373
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.623551845550537,
      "learning_rate": 1.793828408003169e-05,
      "loss": 0.4813,
      "step": 7374
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.678566575050354,
      "learning_rate": 1.793766470087485e-05,
      "loss": 0.5087,
      "step": 7375
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5475927591323853,
      "learning_rate": 1.7937045239391725e-05,
      "loss": 0.4908,
      "step": 7376
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5891886949539185,
      "learning_rate": 1.7936425695588754e-05,
      "loss": 0.5458,
      "step": 7377
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6139299869537354,
      "learning_rate": 1.793580606947236e-05,
      "loss": 0.4884,
      "step": 7378
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.1989786624908447,
      "learning_rate": 1.7935186361048965e-05,
      "loss": 1.0547,
      "step": 7379
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7961844205856323,
      "learning_rate": 1.7934566570324996e-05,
      "loss": 1.0281,
      "step": 7380
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6780365705490112,
      "learning_rate": 1.7933946697306884e-05,
      "loss": 0.5317,
      "step": 7381
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.732286810874939,
      "learning_rate": 1.7933326742001063e-05,
      "loss": 0.495,
      "step": 7382
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7152167558670044,
      "learning_rate": 1.7932706704413953e-05,
      "loss": 0.5353,
      "step": 7383
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7556086778640747,
      "learning_rate": 1.793208658455199e-05,
      "loss": 0.5296,
      "step": 7384
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7071278095245361,
      "learning_rate": 1.7931466382421606e-05,
      "loss": 0.5114,
      "step": 7385
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.800430417060852,
      "learning_rate": 1.7930846098029234e-05,
      "loss": 0.5144,
      "step": 7386
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.65757155418396,
      "learning_rate": 1.7930225731381303e-05,
      "loss": 0.5353,
      "step": 7387
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7456843852996826,
      "learning_rate": 1.792960528248425e-05,
      "loss": 0.5493,
      "step": 7388
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.715739369392395,
      "learning_rate": 1.7928984751344513e-05,
      "loss": 0.5626,
      "step": 7389
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6373971700668335,
      "learning_rate": 1.7928364137968522e-05,
      "loss": 0.5173,
      "step": 7390
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7392358779907227,
      "learning_rate": 1.7927743442362718e-05,
      "loss": 0.5411,
      "step": 7391
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.74623441696167,
      "learning_rate": 1.792712266453354e-05,
      "loss": 0.5117,
      "step": 7392
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9028834104537964,
      "learning_rate": 1.792650180448742e-05,
      "loss": 0.5266,
      "step": 7393
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7090932130813599,
      "learning_rate": 1.79258808622308e-05,
      "loss": 0.4865,
      "step": 7394
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7229399681091309,
      "learning_rate": 1.7925259837770127e-05,
      "loss": 0.5596,
      "step": 7395
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7706599235534668,
      "learning_rate": 1.792463873111183e-05,
      "loss": 0.5137,
      "step": 7396
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6617052555084229,
      "learning_rate": 1.792401754226236e-05,
      "loss": 0.5152,
      "step": 7397
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7060683965682983,
      "learning_rate": 1.792339627122816e-05,
      "loss": 0.5004,
      "step": 7398
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.648386001586914,
      "learning_rate": 1.7922774918015666e-05,
      "loss": 0.5564,
      "step": 7399
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.655273675918579,
      "learning_rate": 1.792215348263133e-05,
      "loss": 0.5123,
      "step": 7400
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6207945346832275,
      "learning_rate": 1.7921531965081594e-05,
      "loss": 0.507,
      "step": 7401
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6635220050811768,
      "learning_rate": 1.7920910365372904e-05,
      "loss": 0.533,
      "step": 7402
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7888197898864746,
      "learning_rate": 1.7920288683511708e-05,
      "loss": 0.6267,
      "step": 7403
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6786322593688965,
      "learning_rate": 1.791966691950445e-05,
      "loss": 0.522,
      "step": 7404
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6433659791946411,
      "learning_rate": 1.7919045073357585e-05,
      "loss": 0.5388,
      "step": 7405
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6174637079238892,
      "learning_rate": 1.7918423145077554e-05,
      "loss": 0.4901,
      "step": 7406
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6307722330093384,
      "learning_rate": 1.7917801134670818e-05,
      "loss": 0.5227,
      "step": 7407
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6097418069839478,
      "learning_rate": 1.7917179042143824e-05,
      "loss": 0.4811,
      "step": 7408
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6759740114212036,
      "learning_rate": 1.791655686750302e-05,
      "loss": 0.5282,
      "step": 7409
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6935912370681763,
      "learning_rate": 1.7915934610754864e-05,
      "loss": 0.5024,
      "step": 7410
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7807644605636597,
      "learning_rate": 1.7915312271905806e-05,
      "loss": 0.6381,
      "step": 7411
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6297465562820435,
      "learning_rate": 1.7914689850962306e-05,
      "loss": 0.4689,
      "step": 7412
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7641987800598145,
      "learning_rate": 1.7914067347930813e-05,
      "loss": 0.5467,
      "step": 7413
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6877676248550415,
      "learning_rate": 1.7913444762817788e-05,
      "loss": 0.5154,
      "step": 7414
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.594137191772461,
      "learning_rate": 1.7912822095629683e-05,
      "loss": 0.5187,
      "step": 7415
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.696722388267517,
      "learning_rate": 1.7912199346372963e-05,
      "loss": 0.4933,
      "step": 7416
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6895374059677124,
      "learning_rate": 1.791157651505408e-05,
      "loss": 0.5682,
      "step": 7417
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.620182991027832,
      "learning_rate": 1.7910953601679503e-05,
      "loss": 0.5018,
      "step": 7418
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6912615299224854,
      "learning_rate": 1.7910330606255683e-05,
      "loss": 0.6018,
      "step": 7419
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6570898294448853,
      "learning_rate": 1.7909707528789084e-05,
      "loss": 0.4881,
      "step": 7420
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7477595806121826,
      "learning_rate": 1.7909084369286172e-05,
      "loss": 0.5741,
      "step": 7421
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.72244131565094,
      "learning_rate": 1.790846112775341e-05,
      "loss": 0.4834,
      "step": 7422
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.681384563446045,
      "learning_rate": 1.7907837804197257e-05,
      "loss": 0.5579,
      "step": 7423
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.666935682296753,
      "learning_rate": 1.7907214398624183e-05,
      "loss": 0.4919,
      "step": 7424
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6404697895050049,
      "learning_rate": 1.7906590911040647e-05,
      "loss": 0.5694,
      "step": 7425
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6661800146102905,
      "learning_rate": 1.7905967341453124e-05,
      "loss": 0.5074,
      "step": 7426
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6348342895507812,
      "learning_rate": 1.7905343689868075e-05,
      "loss": 0.5641,
      "step": 7427
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.616005539894104,
      "learning_rate": 1.790471995629197e-05,
      "loss": 0.4997,
      "step": 7428
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.635675072669983,
      "learning_rate": 1.790409614073128e-05,
      "loss": 0.521,
      "step": 7429
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7004024982452393,
      "learning_rate": 1.7903472243192476e-05,
      "loss": 0.5924,
      "step": 7430
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7485755681991577,
      "learning_rate": 1.7902848263682026e-05,
      "loss": 0.5927,
      "step": 7431
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6298121213912964,
      "learning_rate": 1.79022242022064e-05,
      "loss": 0.4957,
      "step": 7432
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7466752529144287,
      "learning_rate": 1.7901600058772074e-05,
      "loss": 0.5735,
      "step": 7433
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6575160026550293,
      "learning_rate": 1.790097583338552e-05,
      "loss": 0.5107,
      "step": 7434
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.062659740447998,
      "learning_rate": 1.7900351526053212e-05,
      "loss": 0.6062,
      "step": 7435
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6738803386688232,
      "learning_rate": 1.7899727136781626e-05,
      "loss": 0.5019,
      "step": 7436
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8224345445632935,
      "learning_rate": 1.7899102665577234e-05,
      "loss": 0.5648,
      "step": 7437
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7026764154434204,
      "learning_rate": 1.789847811244652e-05,
      "loss": 0.495,
      "step": 7438
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8133118152618408,
      "learning_rate": 1.7897853477395956e-05,
      "loss": 0.6153,
      "step": 7439
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.674971103668213,
      "learning_rate": 1.7897228760432023e-05,
      "loss": 0.4794,
      "step": 7440
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8626766204833984,
      "learning_rate": 1.78966039615612e-05,
      "loss": 0.5755,
      "step": 7441
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.841097354888916,
      "learning_rate": 1.7895979080789964e-05,
      "loss": 0.5305,
      "step": 7442
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9007573127746582,
      "learning_rate": 1.78953541181248e-05,
      "loss": 0.5362,
      "step": 7443
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7614940404891968,
      "learning_rate": 1.7894729073572187e-05,
      "loss": 0.4944,
      "step": 7444
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7632864713668823,
      "learning_rate": 1.7894103947138615e-05,
      "loss": 0.5657,
      "step": 7445
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8608425855636597,
      "learning_rate": 1.7893478738830557e-05,
      "loss": 0.5291,
      "step": 7446
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9064311981201172,
      "learning_rate": 1.78928534486545e-05,
      "loss": 0.6161,
      "step": 7447
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6016329526901245,
      "learning_rate": 1.7892228076616937e-05,
      "loss": 0.4468,
      "step": 7448
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.770971417427063,
      "learning_rate": 1.7891602622724344e-05,
      "loss": 0.5482,
      "step": 7449
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.730805516242981,
      "learning_rate": 1.7890977086983215e-05,
      "loss": 0.5061,
      "step": 7450
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8445048332214355,
      "learning_rate": 1.7890351469400035e-05,
      "loss": 0.6352,
      "step": 7451
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6541236639022827,
      "learning_rate": 1.788972576998129e-05,
      "loss": 0.4782,
      "step": 7452
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6591804027557373,
      "learning_rate": 1.7889099988733477e-05,
      "loss": 0.5798,
      "step": 7453
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7116345167160034,
      "learning_rate": 1.788847412566308e-05,
      "loss": 0.5492,
      "step": 7454
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6919453144073486,
      "learning_rate": 1.788784818077659e-05,
      "loss": 0.5127,
      "step": 7455
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7167531251907349,
      "learning_rate": 1.7887222154080505e-05,
      "loss": 0.5722,
      "step": 7456
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6940674781799316,
      "learning_rate": 1.788659604558131e-05,
      "loss": 0.5836,
      "step": 7457
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.585729718208313,
      "learning_rate": 1.7885969855285504e-05,
      "loss": 0.481,
      "step": 7458
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7189239263534546,
      "learning_rate": 1.7885343583199583e-05,
      "loss": 0.5725,
      "step": 7459
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6934915781021118,
      "learning_rate": 1.788471722933004e-05,
      "loss": 0.4872,
      "step": 7460
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.956292986869812,
      "learning_rate": 1.7884090793683363e-05,
      "loss": 0.9916,
      "step": 7461
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.808242678642273,
      "learning_rate": 1.7883464276266064e-05,
      "loss": 1.0274,
      "step": 7462
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7487105131149292,
      "learning_rate": 1.7882837677084635e-05,
      "loss": 0.5396,
      "step": 7463
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.72042977809906,
      "learning_rate": 1.788221099614557e-05,
      "loss": 0.4553,
      "step": 7464
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.802042007446289,
      "learning_rate": 1.7881584233455377e-05,
      "loss": 0.5396,
      "step": 7465
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7699583768844604,
      "learning_rate": 1.7880957389020545e-05,
      "loss": 0.494,
      "step": 7466
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7255840301513672,
      "learning_rate": 1.788033046284759e-05,
      "loss": 0.5598,
      "step": 7467
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6105077266693115,
      "learning_rate": 1.7879703454943004e-05,
      "loss": 0.4655,
      "step": 7468
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7631888389587402,
      "learning_rate": 1.7879076365313293e-05,
      "loss": 0.5696,
      "step": 7469
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6660019159317017,
      "learning_rate": 1.7878449193964963e-05,
      "loss": 0.4846,
      "step": 7470
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7814967632293701,
      "learning_rate": 1.787782194090451e-05,
      "loss": 0.5738,
      "step": 7471
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.743268609046936,
      "learning_rate": 1.7877194606138453e-05,
      "loss": 0.4604,
      "step": 7472
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8266785144805908,
      "learning_rate": 1.787656718967329e-05,
      "loss": 0.5494,
      "step": 7473
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6843948364257812,
      "learning_rate": 1.7875939691515526e-05,
      "loss": 0.5041,
      "step": 7474
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7187950611114502,
      "learning_rate": 1.787531211167168e-05,
      "loss": 0.6019,
      "step": 7475
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6732938289642334,
      "learning_rate": 1.7874684450148246e-05,
      "loss": 0.507,
      "step": 7476
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6500107049942017,
      "learning_rate": 1.7874056706951748e-05,
      "loss": 0.5777,
      "step": 7477
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5922574996948242,
      "learning_rate": 1.787342888208869e-05,
      "loss": 0.4807,
      "step": 7478
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6913046836853027,
      "learning_rate": 1.7872800975565585e-05,
      "loss": 0.5352,
      "step": 7479
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7088953256607056,
      "learning_rate": 1.7872172987388944e-05,
      "loss": 0.506,
      "step": 7480
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.661765694618225,
      "learning_rate": 1.787154491756528e-05,
      "loss": 0.5133,
      "step": 7481
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6134605407714844,
      "learning_rate": 1.7870916766101106e-05,
      "loss": 0.4684,
      "step": 7482
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6779356002807617,
      "learning_rate": 1.787028853300294e-05,
      "loss": 0.5227,
      "step": 7483
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7165398597717285,
      "learning_rate": 1.7869660218277297e-05,
      "loss": 0.5511,
      "step": 7484
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1509921550750732,
      "learning_rate": 1.7869031821930695e-05,
      "loss": 0.9733,
      "step": 7485
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8416962623596191,
      "learning_rate": 1.7868403343969646e-05,
      "loss": 0.9921,
      "step": 7486
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8639782667160034,
      "learning_rate": 1.786777478440068e-05,
      "loss": 0.556,
      "step": 7487
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9916636943817139,
      "learning_rate": 1.78671461432303e-05,
      "loss": 0.544,
      "step": 7488
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7974979877471924,
      "learning_rate": 1.786651742046504e-05,
      "loss": 0.6246,
      "step": 7489
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.733309030532837,
      "learning_rate": 1.7865888616111408e-05,
      "loss": 0.491,
      "step": 7490
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8159295320510864,
      "learning_rate": 1.7865259730175937e-05,
      "loss": 0.5857,
      "step": 7491
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6686898469924927,
      "learning_rate": 1.786463076266515e-05,
      "loss": 0.4768,
      "step": 7492
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6807632446289062,
      "learning_rate": 1.786400171358556e-05,
      "loss": 0.5581,
      "step": 7493
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.769015908241272,
      "learning_rate": 1.78633725829437e-05,
      "loss": 0.523,
      "step": 7494
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.626194715499878,
      "learning_rate": 1.786274337074609e-05,
      "loss": 0.5364,
      "step": 7495
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8832250833511353,
      "learning_rate": 1.7862114076999257e-05,
      "loss": 0.5555,
      "step": 7496
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.834494948387146,
      "learning_rate": 1.7861484701709732e-05,
      "loss": 0.5303,
      "step": 7497
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.71701979637146,
      "learning_rate": 1.786085524488404e-05,
      "loss": 0.5095,
      "step": 7498
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.648768424987793,
      "learning_rate": 1.7860225706528705e-05,
      "loss": 0.5653,
      "step": 7499
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8020343780517578,
      "learning_rate": 1.785959608665026e-05,
      "loss": 0.5698,
      "step": 7500
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.558672547340393,
      "learning_rate": 1.785896638525524e-05,
      "loss": 0.4889,
      "step": 7501
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5864510536193848,
      "learning_rate": 1.7858336602350167e-05,
      "loss": 0.5019,
      "step": 7502
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8630183935165405,
      "learning_rate": 1.785770673794158e-05,
      "loss": 0.6015,
      "step": 7503
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7438517808914185,
      "learning_rate": 1.785707679203601e-05,
      "loss": 0.4903,
      "step": 7504
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7015695571899414,
      "learning_rate": 1.785644676463999e-05,
      "loss": 0.5551,
      "step": 7505
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6113576889038086,
      "learning_rate": 1.7855816655760053e-05,
      "loss": 0.4778,
      "step": 7506
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8131693601608276,
      "learning_rate": 1.785518646540273e-05,
      "loss": 0.5494,
      "step": 7507
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7072837352752686,
      "learning_rate": 1.7854556193574567e-05,
      "loss": 0.4888,
      "step": 7508
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8025201559066772,
      "learning_rate": 1.78539258402821e-05,
      "loss": 0.5538,
      "step": 7509
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6839338541030884,
      "learning_rate": 1.7853295405531857e-05,
      "loss": 0.4996,
      "step": 7510
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7634998559951782,
      "learning_rate": 1.7852664889330382e-05,
      "loss": 0.5496,
      "step": 7511
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6252893209457397,
      "learning_rate": 1.785203429168422e-05,
      "loss": 0.4714,
      "step": 7512
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8466318845748901,
      "learning_rate": 1.78514036125999e-05,
      "loss": 0.5843,
      "step": 7513
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7091585397720337,
      "learning_rate": 1.7850772852083976e-05,
      "loss": 0.4734,
      "step": 7514
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6542065143585205,
      "learning_rate": 1.785014201014298e-05,
      "loss": 0.5491,
      "step": 7515
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9779229164123535,
      "learning_rate": 1.784951108678346e-05,
      "loss": 0.5008,
      "step": 7516
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.756070613861084,
      "learning_rate": 1.7848880082011956e-05,
      "loss": 0.5274,
      "step": 7517
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7595080137252808,
      "learning_rate": 1.7848248995835017e-05,
      "loss": 0.495,
      "step": 7518
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7458992004394531,
      "learning_rate": 1.7847617828259185e-05,
      "loss": 0.5367,
      "step": 7519
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.733567476272583,
      "learning_rate": 1.784698657929101e-05,
      "loss": 0.4778,
      "step": 7520
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7557599544525146,
      "learning_rate": 1.784635524893703e-05,
      "loss": 0.616,
      "step": 7521
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6599074602127075,
      "learning_rate": 1.78457238372038e-05,
      "loss": 0.5071,
      "step": 7522
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.750756859779358,
      "learning_rate": 1.7845092344097873e-05,
      "loss": 0.5953,
      "step": 7523
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7476177215576172,
      "learning_rate": 1.784446076962579e-05,
      "loss": 0.5086,
      "step": 7524
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9687230587005615,
      "learning_rate": 1.7843829113794105e-05,
      "loss": 0.6599,
      "step": 7525
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6233060359954834,
      "learning_rate": 1.784319737660937e-05,
      "loss": 0.4695,
      "step": 7526
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7516672611236572,
      "learning_rate": 1.7842565558078132e-05,
      "loss": 0.5544,
      "step": 7527
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6882803440093994,
      "learning_rate": 1.7841933658206955e-05,
      "loss": 0.5162,
      "step": 7528
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7080843448638916,
      "learning_rate": 1.7841301677002383e-05,
      "loss": 0.5878,
      "step": 7529
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8042649030685425,
      "learning_rate": 1.7840669614470973e-05,
      "loss": 0.5045,
      "step": 7530
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6737735271453857,
      "learning_rate": 1.784003747061928e-05,
      "loss": 0.5311,
      "step": 7531
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.734143614768982,
      "learning_rate": 1.7839405245453866e-05,
      "loss": 0.5098,
      "step": 7532
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6863027811050415,
      "learning_rate": 1.783877293898128e-05,
      "loss": 0.5204,
      "step": 7533
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.715343713760376,
      "learning_rate": 1.7838140551208087e-05,
      "loss": 0.5026,
      "step": 7534
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.631859064102173,
      "learning_rate": 1.7837508082140842e-05,
      "loss": 1.0538,
      "step": 7535
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.363945484161377,
      "learning_rate": 1.7836875531786102e-05,
      "loss": 1.0881,
      "step": 7536
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9687780141830444,
      "learning_rate": 1.7836242900150433e-05,
      "loss": 0.5391,
      "step": 7537
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9048128128051758,
      "learning_rate": 1.7835610187240394e-05,
      "loss": 0.5155,
      "step": 7538
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.725845217704773,
      "learning_rate": 1.783497739306255e-05,
      "loss": 0.6157,
      "step": 7539
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6948857307434082,
      "learning_rate": 1.783434451762346e-05,
      "loss": 0.4945,
      "step": 7540
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9588145017623901,
      "learning_rate": 1.783371156092969e-05,
      "loss": 0.6611,
      "step": 7541
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8356552124023438,
      "learning_rate": 1.7833078522987805e-05,
      "loss": 0.5003,
      "step": 7542
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.72217857837677,
      "learning_rate": 1.783244540380437e-05,
      "loss": 0.5456,
      "step": 7543
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.738216757774353,
      "learning_rate": 1.7831812203385947e-05,
      "loss": 0.5185,
      "step": 7544
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6484330892562866,
      "learning_rate": 1.783117892173911e-05,
      "loss": 0.5639,
      "step": 7545
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.773231029510498,
      "learning_rate": 1.7830545558870426e-05,
      "loss": 0.551,
      "step": 7546
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.803034782409668,
      "learning_rate": 1.7829912114786462e-05,
      "loss": 0.5745,
      "step": 7547
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6485580205917358,
      "learning_rate": 1.782927858949379e-05,
      "loss": 0.4854,
      "step": 7548
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7774455547332764,
      "learning_rate": 1.782864498299898e-05,
      "loss": 0.6084,
      "step": 7549
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6693036556243896,
      "learning_rate": 1.78280112953086e-05,
      "loss": 0.4736,
      "step": 7550
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6672624349594116,
      "learning_rate": 1.7827377526429227e-05,
      "loss": 0.5327,
      "step": 7551
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6860318183898926,
      "learning_rate": 1.7826743676367434e-05,
      "loss": 0.4621,
      "step": 7552
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7732012271881104,
      "learning_rate": 1.782610974512979e-05,
      "loss": 0.6123,
      "step": 7553
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7570141553878784,
      "learning_rate": 1.7825475732722875e-05,
      "loss": 0.5149,
      "step": 7554
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.774855375289917,
      "learning_rate": 1.7824841639153265e-05,
      "loss": 0.5457,
      "step": 7555
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5781792402267456,
      "learning_rate": 1.7824207464427532e-05,
      "loss": 0.4752,
      "step": 7556
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6038569211959839,
      "learning_rate": 1.7823573208552257e-05,
      "loss": 0.5362,
      "step": 7557
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6972821950912476,
      "learning_rate": 1.7822938871534018e-05,
      "loss": 0.4682,
      "step": 7558
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6604124307632446,
      "learning_rate": 1.7822304453379394e-05,
      "loss": 0.5358,
      "step": 7559
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7361948490142822,
      "learning_rate": 1.7821669954094962e-05,
      "loss": 0.5417,
      "step": 7560
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6628799438476562,
      "learning_rate": 1.7821035373687308e-05,
      "loss": 0.5582,
      "step": 7561
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6551268100738525,
      "learning_rate": 1.7820400712163008e-05,
      "loss": 0.4816,
      "step": 7562
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6562420129776,
      "learning_rate": 1.7819765969528646e-05,
      "loss": 0.4916,
      "step": 7563
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.738741397857666,
      "learning_rate": 1.7819131145790808e-05,
      "loss": 0.5644,
      "step": 7564
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6536203622817993,
      "learning_rate": 1.7818496240956077e-05,
      "loss": 0.5754,
      "step": 7565
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.729493260383606,
      "learning_rate": 1.781786125503104e-05,
      "loss": 0.5343,
      "step": 7566
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6838487386703491,
      "learning_rate": 1.7817226188022277e-05,
      "loss": 0.5126,
      "step": 7567
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.713344693183899,
      "learning_rate": 1.7816591039936377e-05,
      "loss": 0.5213,
      "step": 7568
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8977599143981934,
      "learning_rate": 1.781595581077993e-05,
      "loss": 0.5749,
      "step": 7569
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.716088891029358,
      "learning_rate": 1.7815320500559524e-05,
      "loss": 0.4732,
      "step": 7570
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.745524287223816,
      "learning_rate": 1.7814685109281748e-05,
      "loss": 0.5726,
      "step": 7571
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.136537551879883,
      "learning_rate": 1.781404963695319e-05,
      "loss": 0.5616,
      "step": 7572
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7348567247390747,
      "learning_rate": 1.7813414083580444e-05,
      "loss": 0.5752,
      "step": 7573
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6656899452209473,
      "learning_rate": 1.7812778449170096e-05,
      "loss": 0.5028,
      "step": 7574
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8965246677398682,
      "learning_rate": 1.781214273372874e-05,
      "loss": 0.6453,
      "step": 7575
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.703410267829895,
      "learning_rate": 1.7811506937262976e-05,
      "loss": 0.4901,
      "step": 7576
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6942752599716187,
      "learning_rate": 1.7810871059779394e-05,
      "loss": 0.5748,
      "step": 7577
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.737288475036621,
      "learning_rate": 1.7810235101284588e-05,
      "loss": 0.5065,
      "step": 7578
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8692305088043213,
      "learning_rate": 1.7809599061785155e-05,
      "loss": 0.6009,
      "step": 7579
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.72127366065979,
      "learning_rate": 1.780896294128769e-05,
      "loss": 0.4599,
      "step": 7580
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.790592908859253,
      "learning_rate": 1.7808326739798794e-05,
      "loss": 0.6084,
      "step": 7581
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6248925924301147,
      "learning_rate": 1.7807690457325063e-05,
      "loss": 0.4768,
      "step": 7582
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7378449440002441,
      "learning_rate": 1.7807054093873096e-05,
      "loss": 0.5533,
      "step": 7583
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6817808151245117,
      "learning_rate": 1.7806417649449494e-05,
      "loss": 0.4796,
      "step": 7584
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7199175357818604,
      "learning_rate": 1.780578112406086e-05,
      "loss": 0.6076,
      "step": 7585
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.710183024406433,
      "learning_rate": 1.7805144517713792e-05,
      "loss": 0.4926,
      "step": 7586
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8879464864730835,
      "learning_rate": 1.7804507830414894e-05,
      "loss": 0.6275,
      "step": 7587
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.574817180633545,
      "learning_rate": 1.780387106217077e-05,
      "loss": 0.4586,
      "step": 7588
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7775541543960571,
      "learning_rate": 1.7803234212988023e-05,
      "loss": 0.6048,
      "step": 7589
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.596677303314209,
      "learning_rate": 1.7802597282873263e-05,
      "loss": 0.4732,
      "step": 7590
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.699835181236267,
      "learning_rate": 1.7801960271833087e-05,
      "loss": 0.5592,
      "step": 7591
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7293016910552979,
      "learning_rate": 1.7801323179874112e-05,
      "loss": 0.5064,
      "step": 7592
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7634824514389038,
      "learning_rate": 1.7800686007002935e-05,
      "loss": 0.521,
      "step": 7593
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7423481941223145,
      "learning_rate": 1.7800048753226176e-05,
      "loss": 0.5311,
      "step": 7594
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7957258224487305,
      "learning_rate": 1.7799411418550436e-05,
      "loss": 0.5774,
      "step": 7595
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8204394578933716,
      "learning_rate": 1.7798774002982327e-05,
      "loss": 0.5084,
      "step": 7596
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.655820369720459,
      "learning_rate": 1.7798136506528464e-05,
      "loss": 0.5383,
      "step": 7597
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6438312530517578,
      "learning_rate": 1.7797498929195453e-05,
      "loss": 0.4621,
      "step": 7598
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.606328010559082,
      "learning_rate": 1.7796861270989908e-05,
      "loss": 0.5232,
      "step": 7599
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7580829858779907,
      "learning_rate": 1.7796223531918446e-05,
      "loss": 0.5166,
      "step": 7600
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7305632829666138,
      "learning_rate": 1.7795585711987678e-05,
      "loss": 0.5633,
      "step": 7601
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7380579710006714,
      "learning_rate": 1.779494781120422e-05,
      "loss": 0.5194,
      "step": 7602
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7230398654937744,
      "learning_rate": 1.7794309829574688e-05,
      "loss": 0.5425,
      "step": 7603
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7398121356964111,
      "learning_rate": 1.77936717671057e-05,
      "loss": 0.517,
      "step": 7604
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7955303192138672,
      "learning_rate": 1.7793033623803876e-05,
      "loss": 0.6011,
      "step": 7605
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6963692903518677,
      "learning_rate": 1.779239539967583e-05,
      "loss": 0.5094,
      "step": 7606
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8059509992599487,
      "learning_rate": 1.779175709472818e-05,
      "loss": 0.6123,
      "step": 7607
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7499535083770752,
      "learning_rate": 1.779111870896755e-05,
      "loss": 0.5081,
      "step": 7608
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6494061946868896,
      "learning_rate": 1.7790480242400562e-05,
      "loss": 0.5769,
      "step": 7609
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.734052062034607,
      "learning_rate": 1.778984169503384e-05,
      "loss": 0.5175,
      "step": 7610
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7545180320739746,
      "learning_rate": 1.7789203066874e-05,
      "loss": 0.5731,
      "step": 7611
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7328712940216064,
      "learning_rate": 1.7788564357927666e-05,
      "loss": 0.5138,
      "step": 7612
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7748994827270508,
      "learning_rate": 1.7787925568201466e-05,
      "loss": 0.5333,
      "step": 7613
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6566462516784668,
      "learning_rate": 1.7787286697702028e-05,
      "loss": 0.484,
      "step": 7614
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7276939153671265,
      "learning_rate": 1.778664774643597e-05,
      "loss": 0.5543,
      "step": 7615
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7380489110946655,
      "learning_rate": 1.7786008714409923e-05,
      "loss": 0.4813,
      "step": 7616
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7562494277954102,
      "learning_rate": 1.7785369601630517e-05,
      "loss": 0.5712,
      "step": 7617
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.601535677909851,
      "learning_rate": 1.7784730408104382e-05,
      "loss": 0.514,
      "step": 7618
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.636683464050293,
      "learning_rate": 1.7784091133838138e-05,
      "loss": 0.5442,
      "step": 7619
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6389119625091553,
      "learning_rate": 1.778345177883843e-05,
      "loss": 0.5213,
      "step": 7620
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7577441930770874,
      "learning_rate": 1.7782812343111873e-05,
      "loss": 0.6111,
      "step": 7621
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8196370601654053,
      "learning_rate": 1.7782172826665106e-05,
      "loss": 0.4878,
      "step": 7622
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7992888689041138,
      "learning_rate": 1.778153322950477e-05,
      "loss": 0.5407,
      "step": 7623
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.78115975856781,
      "learning_rate": 1.7780893551637483e-05,
      "loss": 0.555,
      "step": 7624
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6749392747879028,
      "learning_rate": 1.7780253793069892e-05,
      "loss": 0.5534,
      "step": 7625
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5951368808746338,
      "learning_rate": 1.7779613953808627e-05,
      "loss": 0.4664,
      "step": 7626
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7615907192230225,
      "learning_rate": 1.7778974033860324e-05,
      "loss": 0.5573,
      "step": 7627
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7700570821762085,
      "learning_rate": 1.7778334033231622e-05,
      "loss": 0.4911,
      "step": 7628
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.683152437210083,
      "learning_rate": 1.777769395192916e-05,
      "loss": 0.5468,
      "step": 7629
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.606246829032898,
      "learning_rate": 1.7777053789959574e-05,
      "loss": 0.492,
      "step": 7630
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6159117221832275,
      "learning_rate": 1.7776413547329502e-05,
      "loss": 0.5997,
      "step": 7631
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6620440483093262,
      "learning_rate": 1.7775773224045588e-05,
      "loss": 0.4784,
      "step": 7632
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7844908237457275,
      "learning_rate": 1.777513282011447e-05,
      "loss": 0.6151,
      "step": 7633
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.65964937210083,
      "learning_rate": 1.7774492335542794e-05,
      "loss": 0.4744,
      "step": 7634
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7121803760528564,
      "learning_rate": 1.7773851770337196e-05,
      "loss": 0.6086,
      "step": 7635
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6146175861358643,
      "learning_rate": 1.7773211124504327e-05,
      "loss": 0.4479,
      "step": 7636
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8177779912948608,
      "learning_rate": 1.777257039805083e-05,
      "loss": 0.5488,
      "step": 7637
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7520800828933716,
      "learning_rate": 1.777192959098335e-05,
      "loss": 0.5176,
      "step": 7638
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.76534104347229,
      "learning_rate": 1.7771288703308526e-05,
      "loss": 0.5674,
      "step": 7639
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7117198705673218,
      "learning_rate": 1.7770647735033017e-05,
      "loss": 0.488,
      "step": 7640
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7549563646316528,
      "learning_rate": 1.777000668616346e-05,
      "loss": 0.5862,
      "step": 7641
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5678197145462036,
      "learning_rate": 1.7769365556706513e-05,
      "loss": 0.474,
      "step": 7642
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6918995380401611,
      "learning_rate": 1.776872434666882e-05,
      "loss": 0.5486,
      "step": 7643
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9509012699127197,
      "learning_rate": 1.7768083056057034e-05,
      "loss": 0.5489,
      "step": 7644
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7386877536773682,
      "learning_rate": 1.7767441684877804e-05,
      "loss": 0.5449,
      "step": 7645
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5984811782836914,
      "learning_rate": 1.7766800233137783e-05,
      "loss": 0.4912,
      "step": 7646
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7703560590744019,
      "learning_rate": 1.7766158700843627e-05,
      "loss": 0.5232,
      "step": 7647
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.76338791847229,
      "learning_rate": 1.776551708800198e-05,
      "loss": 0.4564,
      "step": 7648
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7575135231018066,
      "learning_rate": 1.7764875394619506e-05,
      "loss": 0.5643,
      "step": 7649
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6795127391815186,
      "learning_rate": 1.776423362070286e-05,
      "loss": 0.5029,
      "step": 7650
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.757118821144104,
      "learning_rate": 1.7763591766258692e-05,
      "loss": 0.5839,
      "step": 7651
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7111098766326904,
      "learning_rate": 1.7762949831293665e-05,
      "loss": 0.4867,
      "step": 7652
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6183357238769531,
      "learning_rate": 1.7762307815814433e-05,
      "loss": 0.5154,
      "step": 7653
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.734480381011963,
      "learning_rate": 1.7761665719827657e-05,
      "loss": 0.5252,
      "step": 7654
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7092993259429932,
      "learning_rate": 1.776102354334e-05,
      "loss": 0.5194,
      "step": 7655
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6451421976089478,
      "learning_rate": 1.7760381286358114e-05,
      "loss": 0.4727,
      "step": 7656
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6548765897750854,
      "learning_rate": 1.7759738948888662e-05,
      "loss": 0.5592,
      "step": 7657
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6802681684494019,
      "learning_rate": 1.775909653093831e-05,
      "loss": 0.5133,
      "step": 7658
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7663432359695435,
      "learning_rate": 1.7758454032513723e-05,
      "loss": 0.605,
      "step": 7659
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6749643087387085,
      "learning_rate": 1.775781145362156e-05,
      "loss": 0.4766,
      "step": 7660
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.676895022392273,
      "learning_rate": 1.7757168794268486e-05,
      "loss": 0.5555,
      "step": 7661
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7341703176498413,
      "learning_rate": 1.775652605446117e-05,
      "loss": 0.5386,
      "step": 7662
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6865806579589844,
      "learning_rate": 1.7755883234206274e-05,
      "loss": 0.5016,
      "step": 7663
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.688518762588501,
      "learning_rate": 1.7755240333510463e-05,
      "loss": 0.5162,
      "step": 7664
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5696642398834229,
      "learning_rate": 1.7754597352380413e-05,
      "loss": 0.4868,
      "step": 7665
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.680639624595642,
      "learning_rate": 1.7753954290822786e-05,
      "loss": 0.48,
      "step": 7666
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5952532291412354,
      "learning_rate": 1.7753311148844253e-05,
      "loss": 0.5538,
      "step": 7667
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7548058032989502,
      "learning_rate": 1.775266792645149e-05,
      "loss": 0.5184,
      "step": 7668
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6948586702346802,
      "learning_rate": 1.775202462365116e-05,
      "loss": 0.5404,
      "step": 7669
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6874703168869019,
      "learning_rate": 1.7751381240449935e-05,
      "loss": 0.5053,
      "step": 7670
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7534548044204712,
      "learning_rate": 1.77507377768545e-05,
      "loss": 0.5685,
      "step": 7671
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6358274221420288,
      "learning_rate": 1.7750094232871514e-05,
      "loss": 0.4703,
      "step": 7672
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.725029468536377,
      "learning_rate": 1.7749450608507656e-05,
      "loss": 0.5628,
      "step": 7673
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6055629253387451,
      "learning_rate": 1.7748806903769608e-05,
      "loss": 0.4979,
      "step": 7674
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7031525373458862,
      "learning_rate": 1.774816311866404e-05,
      "loss": 0.5725,
      "step": 7675
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7926326990127563,
      "learning_rate": 1.774751925319763e-05,
      "loss": 0.5385,
      "step": 7676
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7505345344543457,
      "learning_rate": 1.7746875307377055e-05,
      "loss": 0.5713,
      "step": 7677
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7101004123687744,
      "learning_rate": 1.7746231281209e-05,
      "loss": 0.513,
      "step": 7678
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7711055278778076,
      "learning_rate": 1.7745587174700135e-05,
      "loss": 0.6206,
      "step": 7679
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6878225803375244,
      "learning_rate": 1.7744942987857147e-05,
      "loss": 0.5168,
      "step": 7680
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7869473695755005,
      "learning_rate": 1.7744298720686714e-05,
      "loss": 0.6264,
      "step": 7681
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6399086713790894,
      "learning_rate": 1.7743654373195522e-05,
      "loss": 0.503,
      "step": 7682
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7138144969940186,
      "learning_rate": 1.7743009945390248e-05,
      "loss": 0.5576,
      "step": 7683
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6654367446899414,
      "learning_rate": 1.7742365437277583e-05,
      "loss": 0.4836,
      "step": 7684
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.727729082107544,
      "learning_rate": 1.7741720848864207e-05,
      "loss": 0.5829,
      "step": 7685
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.613853096961975,
      "learning_rate": 1.7741076180156806e-05,
      "loss": 0.4647,
      "step": 7686
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.999356985092163,
      "learning_rate": 1.7740431431162063e-05,
      "loss": 0.588,
      "step": 7687
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8442715406417847,
      "learning_rate": 1.773978660188667e-05,
      "loss": 0.5251,
      "step": 7688
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7972073554992676,
      "learning_rate": 1.7739141692337317e-05,
      "loss": 0.5655,
      "step": 7689
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6479847431182861,
      "learning_rate": 1.7738496702520685e-05,
      "loss": 0.4729,
      "step": 7690
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6480097770690918,
      "learning_rate": 1.773785163244347e-05,
      "loss": 0.5126,
      "step": 7691
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6375800371170044,
      "learning_rate": 1.773720648211236e-05,
      "loss": 0.5019,
      "step": 7692
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.737534761428833,
      "learning_rate": 1.7736561251534046e-05,
      "loss": 0.5669,
      "step": 7693
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7547557353973389,
      "learning_rate": 1.773591594071522e-05,
      "loss": 0.5003,
      "step": 7694
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6650067567825317,
      "learning_rate": 1.7735270549662574e-05,
      "loss": 0.5409,
      "step": 7695
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7021509408950806,
      "learning_rate": 1.773462507838281e-05,
      "loss": 0.4806,
      "step": 7696
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8028210401535034,
      "learning_rate": 1.7733979526882607e-05,
      "loss": 0.5502,
      "step": 7697
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7786049842834473,
      "learning_rate": 1.7733333895168673e-05,
      "loss": 0.5347,
      "step": 7698
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9281009435653687,
      "learning_rate": 1.77326881832477e-05,
      "loss": 0.6109,
      "step": 7699
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.770429253578186,
      "learning_rate": 1.7732042391126385e-05,
      "loss": 0.4725,
      "step": 7700
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8002324104309082,
      "learning_rate": 1.7731396518811424e-05,
      "loss": 0.6162,
      "step": 7701
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6067627668380737,
      "learning_rate": 1.773075056630952e-05,
      "loss": 0.4961,
      "step": 7702
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9370614290237427,
      "learning_rate": 1.773010453362737e-05,
      "loss": 1.0239,
      "step": 7703
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8392784595489502,
      "learning_rate": 1.7729458420771676e-05,
      "loss": 1.1084,
      "step": 7704
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9377100467681885,
      "learning_rate": 1.772881222774914e-05,
      "loss": 0.5836,
      "step": 7705
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8919674158096313,
      "learning_rate": 1.7728165954566458e-05,
      "loss": 0.4997,
      "step": 7706
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7887557744979858,
      "learning_rate": 1.772751960123034e-05,
      "loss": 0.5499,
      "step": 7707
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6577436923980713,
      "learning_rate": 1.7726873167747486e-05,
      "loss": 0.4964,
      "step": 7708
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1885359287261963,
      "learning_rate": 1.77262266541246e-05,
      "loss": 0.5802,
      "step": 7709
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7724324464797974,
      "learning_rate": 1.772558006036839e-05,
      "loss": 0.5568,
      "step": 7710
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.792711853981018,
      "learning_rate": 1.7724933386485563e-05,
      "loss": 0.5491,
      "step": 7711
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7917587757110596,
      "learning_rate": 1.772428663248282e-05,
      "loss": 0.5101,
      "step": 7712
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8338779211044312,
      "learning_rate": 1.7723639798366873e-05,
      "loss": 0.5416,
      "step": 7713
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6182920932769775,
      "learning_rate": 1.7722992884144433e-05,
      "loss": 0.4809,
      "step": 7714
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6887625455856323,
      "learning_rate": 1.7722345889822207e-05,
      "loss": 0.5249,
      "step": 7715
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7996615171432495,
      "learning_rate": 1.7721698815406907e-05,
      "loss": 0.5499,
      "step": 7716
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.738048791885376,
      "learning_rate": 1.772105166090524e-05,
      "loss": 0.5567,
      "step": 7717
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8427790403366089,
      "learning_rate": 1.7720404426323922e-05,
      "loss": 0.5188,
      "step": 7718
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7723658084869385,
      "learning_rate": 1.771975711166966e-05,
      "loss": 0.5709,
      "step": 7719
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.747556447982788,
      "learning_rate": 1.7719109716949177e-05,
      "loss": 0.4708,
      "step": 7720
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7634228467941284,
      "learning_rate": 1.7718462242169182e-05,
      "loss": 0.6017,
      "step": 7721
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6862208843231201,
      "learning_rate": 1.771781468733639e-05,
      "loss": 0.4967,
      "step": 7722
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.670141339302063,
      "learning_rate": 1.7717167052457523e-05,
      "loss": 0.5215,
      "step": 7723
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7009373903274536,
      "learning_rate": 1.771651933753929e-05,
      "loss": 0.5069,
      "step": 7724
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6857036352157593,
      "learning_rate": 1.7715871542588406e-05,
      "loss": 0.5501,
      "step": 7725
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5948654413223267,
      "learning_rate": 1.77152236676116e-05,
      "loss": 0.4766,
      "step": 7726
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7264838218688965,
      "learning_rate": 1.7714575712615595e-05,
      "loss": 0.5394,
      "step": 7727
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.77835214138031,
      "learning_rate": 1.7713927677607093e-05,
      "loss": 0.534,
      "step": 7728
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.173311948776245,
      "learning_rate": 1.7713279562592834e-05,
      "loss": 0.9924,
      "step": 7729
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7545667886734009,
      "learning_rate": 1.7712631367579524e-05,
      "loss": 1.0145,
      "step": 7730
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7833667993545532,
      "learning_rate": 1.77119830925739e-05,
      "loss": 0.5367,
      "step": 7731
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7455354928970337,
      "learning_rate": 1.7711334737582678e-05,
      "loss": 0.5256,
      "step": 7732
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6772757768630981,
      "learning_rate": 1.771068630261258e-05,
      "loss": 0.5511,
      "step": 7733
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7146748304367065,
      "learning_rate": 1.7710037787670337e-05,
      "loss": 0.5227,
      "step": 7734
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7400567531585693,
      "learning_rate": 1.7709389192762673e-05,
      "loss": 0.5506,
      "step": 7735
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.699576497077942,
      "learning_rate": 1.7708740517896318e-05,
      "loss": 0.5705,
      "step": 7736
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9262892007827759,
      "learning_rate": 1.7708091763077995e-05,
      "loss": 0.6224,
      "step": 7737
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.717915654182434,
      "learning_rate": 1.7707442928314433e-05,
      "loss": 0.4835,
      "step": 7738
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9117083549499512,
      "learning_rate": 1.7706794013612367e-05,
      "loss": 1.0516,
      "step": 7739
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7911632061004639,
      "learning_rate": 1.770614501897852e-05,
      "loss": 1.1211,
      "step": 7740
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6596208810806274,
      "learning_rate": 1.7705495944419623e-05,
      "loss": 0.5214,
      "step": 7741
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.761054277420044,
      "learning_rate": 1.7704846789942414e-05,
      "loss": 0.4929,
      "step": 7742
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8083974123001099,
      "learning_rate": 1.7704197555553627e-05,
      "loss": 0.5771,
      "step": 7743
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7676705121994019,
      "learning_rate": 1.770354824125999e-05,
      "loss": 0.5276,
      "step": 7744
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.773808240890503,
      "learning_rate": 1.770289884706824e-05,
      "loss": 0.556,
      "step": 7745
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.753588318824768,
      "learning_rate": 1.770224937298511e-05,
      "loss": 0.478,
      "step": 7746
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5906450748443604,
      "learning_rate": 1.7701599819017333e-05,
      "loss": 0.5356,
      "step": 7747
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7435475587844849,
      "learning_rate": 1.7700950185171657e-05,
      "loss": 0.5246,
      "step": 7748
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6493098735809326,
      "learning_rate": 1.770030047145481e-05,
      "loss": 0.5284,
      "step": 7749
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.782069206237793,
      "learning_rate": 1.7699650677873535e-05,
      "loss": 0.5193,
      "step": 7750
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6422046422958374,
      "learning_rate": 1.7699000804434567e-05,
      "loss": 0.5352,
      "step": 7751
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5726301670074463,
      "learning_rate": 1.7698350851144655e-05,
      "loss": 0.4665,
      "step": 7752
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7414990663528442,
      "learning_rate": 1.769770081801053e-05,
      "loss": 0.5312,
      "step": 7753
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6642166376113892,
      "learning_rate": 1.769705070503894e-05,
      "loss": 0.5008,
      "step": 7754
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6798808574676514,
      "learning_rate": 1.7696400512236624e-05,
      "loss": 0.5551,
      "step": 7755
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7110052108764648,
      "learning_rate": 1.769575023961033e-05,
      "loss": 0.4914,
      "step": 7756
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6723839044570923,
      "learning_rate": 1.76950998871668e-05,
      "loss": 0.5272,
      "step": 7757
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5605521202087402,
      "learning_rate": 1.7694449454912775e-05,
      "loss": 0.4916,
      "step": 7758
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6106576919555664,
      "learning_rate": 1.769379894285501e-05,
      "loss": 0.574,
      "step": 7759
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8050079345703125,
      "learning_rate": 1.7693148351000245e-05,
      "loss": 0.5582,
      "step": 7760
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8342493772506714,
      "learning_rate": 1.769249767935523e-05,
      "loss": 0.5536,
      "step": 7761
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7833530902862549,
      "learning_rate": 1.7691846927926713e-05,
      "loss": 0.4898,
      "step": 7762
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6968967914581299,
      "learning_rate": 1.7691196096721442e-05,
      "loss": 0.5525,
      "step": 7763
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.665793538093567,
      "learning_rate": 1.769054518574617e-05,
      "loss": 0.5458,
      "step": 7764
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6912676095962524,
      "learning_rate": 1.7689894195007647e-05,
      "loss": 0.5077,
      "step": 7765
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6537668704986572,
      "learning_rate": 1.7689243124512624e-05,
      "loss": 0.51,
      "step": 7766
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7701349258422852,
      "learning_rate": 1.7688591974267854e-05,
      "loss": 0.5277,
      "step": 7767
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9322093725204468,
      "learning_rate": 1.768794074428009e-05,
      "loss": 0.4804,
      "step": 7768
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7971514463424683,
      "learning_rate": 1.7687289434556087e-05,
      "loss": 0.5556,
      "step": 7769
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7782655954360962,
      "learning_rate": 1.76866380451026e-05,
      "loss": 0.5529,
      "step": 7770
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.725903034210205,
      "learning_rate": 1.7685986575926386e-05,
      "loss": 0.5614,
      "step": 7771
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.757988691329956,
      "learning_rate": 1.76853350270342e-05,
      "loss": 0.4914,
      "step": 7772
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7524257898330688,
      "learning_rate": 1.76846833984328e-05,
      "loss": 0.5542,
      "step": 7773
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6386055946350098,
      "learning_rate": 1.7684031690128944e-05,
      "loss": 0.4589,
      "step": 7774
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7127161026000977,
      "learning_rate": 1.7683379902129395e-05,
      "loss": 0.5738,
      "step": 7775
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6059963703155518,
      "learning_rate": 1.7682728034440904e-05,
      "loss": 0.4746,
      "step": 7776
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6640253067016602,
      "learning_rate": 1.7682076087070243e-05,
      "loss": 0.6042,
      "step": 7777
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6764945983886719,
      "learning_rate": 1.768142406002417e-05,
      "loss": 0.4947,
      "step": 7778
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.791935920715332,
      "learning_rate": 1.7680771953309442e-05,
      "loss": 0.6037,
      "step": 7779
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7026875019073486,
      "learning_rate": 1.7680119766932832e-05,
      "loss": 0.5008,
      "step": 7780
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6308670043945312,
      "learning_rate": 1.7679467500901094e-05,
      "loss": 0.5085,
      "step": 7781
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7225595712661743,
      "learning_rate": 1.7678815155221e-05,
      "loss": 0.476,
      "step": 7782
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6393804550170898,
      "learning_rate": 1.767816272989931e-05,
      "loss": 0.5201,
      "step": 7783
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.720558524131775,
      "learning_rate": 1.76775102249428e-05,
      "loss": 0.5492,
      "step": 7784
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6030656099319458,
      "learning_rate": 1.767685764035823e-05,
      "loss": 0.5446,
      "step": 7785
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7434321641921997,
      "learning_rate": 1.7676204976152372e-05,
      "loss": 0.5301,
      "step": 7786
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5320353507995605,
      "learning_rate": 1.767555223233199e-05,
      "loss": 0.4974,
      "step": 7787
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7457047700881958,
      "learning_rate": 1.7674899408903857e-05,
      "loss": 0.5577,
      "step": 7788
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.745789885520935,
      "learning_rate": 1.7674246505874748e-05,
      "loss": 0.5626,
      "step": 7789
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.525412917137146,
      "learning_rate": 1.767359352325143e-05,
      "loss": 0.4388,
      "step": 7790
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5603625774383545,
      "learning_rate": 1.7672940461040674e-05,
      "loss": 0.5105,
      "step": 7791
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6769355535507202,
      "learning_rate": 1.767228731924926e-05,
      "loss": 0.4875,
      "step": 7792
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6446545124053955,
      "learning_rate": 1.7671634097883952e-05,
      "loss": 0.504,
      "step": 7793
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7362216711044312,
      "learning_rate": 1.7670980796951537e-05,
      "loss": 0.4977,
      "step": 7794
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7091448307037354,
      "learning_rate": 1.767032741645878e-05,
      "loss": 0.5811,
      "step": 7795
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.652005910873413,
      "learning_rate": 1.7669673956412467e-05,
      "loss": 0.5168,
      "step": 7796
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8784942626953125,
      "learning_rate": 1.7669020416819368e-05,
      "loss": 0.5834,
      "step": 7797
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7004883289337158,
      "learning_rate": 1.7668366797686263e-05,
      "loss": 0.5025,
      "step": 7798
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8958462476730347,
      "learning_rate": 1.7667713099019933e-05,
      "loss": 0.6039,
      "step": 7799
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6744211912155151,
      "learning_rate": 1.7667059320827156e-05,
      "loss": 0.5049,
      "step": 7800
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8511043787002563,
      "learning_rate": 1.7666405463114716e-05,
      "loss": 0.6318,
      "step": 7801
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6537131071090698,
      "learning_rate": 1.766575152588939e-05,
      "loss": 0.498,
      "step": 7802
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.727912425994873,
      "learning_rate": 1.7665097509157962e-05,
      "loss": 0.5884,
      "step": 7803
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6622679233551025,
      "learning_rate": 1.7664443412927217e-05,
      "loss": 0.4791,
      "step": 7804
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.709258794784546,
      "learning_rate": 1.7663789237203935e-05,
      "loss": 0.5246,
      "step": 7805
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.71297025680542,
      "learning_rate": 1.7663134981994908e-05,
      "loss": 0.4981,
      "step": 7806
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8456075191497803,
      "learning_rate": 1.7662480647306916e-05,
      "loss": 0.5728,
      "step": 7807
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6791675090789795,
      "learning_rate": 1.7661826233146742e-05,
      "loss": 0.5016,
      "step": 7808
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7155888080596924,
      "learning_rate": 1.7661171739521187e-05,
      "loss": 0.5476,
      "step": 7809
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7319154739379883,
      "learning_rate": 1.7660517166437022e-05,
      "loss": 0.4854,
      "step": 7810
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9246145486831665,
      "learning_rate": 1.7659862513901048e-05,
      "loss": 0.9435,
      "step": 7811
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5813252925872803,
      "learning_rate": 1.7659207781920052e-05,
      "loss": 1.0203,
      "step": 7812
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8898911476135254,
      "learning_rate": 1.7658552970500823e-05,
      "loss": 0.5536,
      "step": 7813
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8470715284347534,
      "learning_rate": 1.765789807965015e-05,
      "loss": 0.4905,
      "step": 7814
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7172484397888184,
      "learning_rate": 1.7657243109374832e-05,
      "loss": 0.5001,
      "step": 7815
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.675577163696289,
      "learning_rate": 1.7656588059681657e-05,
      "loss": 0.4588,
      "step": 7816
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7888258695602417,
      "learning_rate": 1.765593293057742e-05,
      "loss": 0.5787,
      "step": 7817
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5867104530334473,
      "learning_rate": 1.7655277722068917e-05,
      "loss": 0.443,
      "step": 7818
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6416127681732178,
      "learning_rate": 1.765462243416294e-05,
      "loss": 0.5385,
      "step": 7819
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7335286140441895,
      "learning_rate": 1.7653967066866292e-05,
      "loss": 0.4846,
      "step": 7820
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6436207294464111,
      "learning_rate": 1.7653311620185763e-05,
      "loss": 0.4901,
      "step": 7821
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7382267713546753,
      "learning_rate": 1.7652656094128153e-05,
      "loss": 0.4979,
      "step": 7822
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7109711170196533,
      "learning_rate": 1.7652000488700266e-05,
      "loss": 0.5237,
      "step": 7823
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8407974243164062,
      "learning_rate": 1.7651344803908896e-05,
      "loss": 0.56,
      "step": 7824
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7308673858642578,
      "learning_rate": 1.7650689039760848e-05,
      "loss": 0.5275,
      "step": 7825
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6794260740280151,
      "learning_rate": 1.7650033196262913e-05,
      "loss": 0.4943,
      "step": 7826
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8538178205490112,
      "learning_rate": 1.7649377273421908e-05,
      "loss": 0.5267,
      "step": 7827
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.70258367061615,
      "learning_rate": 1.7648721271244626e-05,
      "loss": 0.4753,
      "step": 7828
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6910439729690552,
      "learning_rate": 1.7648065189737873e-05,
      "loss": 0.573,
      "step": 7829
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6178783178329468,
      "learning_rate": 1.7647409028908456e-05,
      "loss": 0.4769,
      "step": 7830
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6582225561141968,
      "learning_rate": 1.7646752788763176e-05,
      "loss": 0.5178,
      "step": 7831
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.718230128288269,
      "learning_rate": 1.7646096469308843e-05,
      "loss": 0.4914,
      "step": 7832
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7426668405532837,
      "learning_rate": 1.7645440070552266e-05,
      "loss": 0.579,
      "step": 7833
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7150551080703735,
      "learning_rate": 1.7644783592500246e-05,
      "loss": 0.4905,
      "step": 7834
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6901350021362305,
      "learning_rate": 1.7644127035159596e-05,
      "loss": 0.5109,
      "step": 7835
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6696323156356812,
      "learning_rate": 1.7643470398537126e-05,
      "loss": 0.4772,
      "step": 7836
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.810010313987732,
      "learning_rate": 1.7642813682639643e-05,
      "loss": 0.6217,
      "step": 7837
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8694696426391602,
      "learning_rate": 1.7642156887473964e-05,
      "loss": 0.5444,
      "step": 7838
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6369656324386597,
      "learning_rate": 1.7641500013046896e-05,
      "loss": 0.5239,
      "step": 7839
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6302340030670166,
      "learning_rate": 1.7640843059365256e-05,
      "loss": 0.4849,
      "step": 7840
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8801647424697876,
      "learning_rate": 1.764018602643585e-05,
      "loss": 0.5985,
      "step": 7841
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.677747130393982,
      "learning_rate": 1.7639528914265502e-05,
      "loss": 0.4936,
      "step": 7842
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.760890007019043,
      "learning_rate": 1.7638871722861023e-05,
      "loss": 0.5954,
      "step": 7843
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.750083327293396,
      "learning_rate": 1.763821445222923e-05,
      "loss": 0.494,
      "step": 7844
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6634275913238525,
      "learning_rate": 1.7637557102376933e-05,
      "loss": 0.5464,
      "step": 7845
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.723994493484497,
      "learning_rate": 1.763689967331096e-05,
      "loss": 0.5126,
      "step": 7846
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6983510255813599,
      "learning_rate": 1.7636242165038126e-05,
      "loss": 0.5376,
      "step": 7847
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7403565645217896,
      "learning_rate": 1.7635584577565245e-05,
      "loss": 0.481,
      "step": 7848
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8532838821411133,
      "learning_rate": 1.7634926910899147e-05,
      "loss": 0.592,
      "step": 7849
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.597618818283081,
      "learning_rate": 1.7634269165046647e-05,
      "loss": 0.4656,
      "step": 7850
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.75291109085083,
      "learning_rate": 1.7633611340014567e-05,
      "loss": 0.6289,
      "step": 7851
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6681909561157227,
      "learning_rate": 1.763295343580973e-05,
      "loss": 0.5031,
      "step": 7852
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6171631813049316,
      "learning_rate": 1.7632295452438963e-05,
      "loss": 0.5328,
      "step": 7853
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7260818481445312,
      "learning_rate": 1.7631637389909086e-05,
      "loss": 0.5114,
      "step": 7854
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8036401271820068,
      "learning_rate": 1.7630979248226926e-05,
      "loss": 0.6091,
      "step": 7855
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6493664979934692,
      "learning_rate": 1.7630321027399307e-05,
      "loss": 0.4854,
      "step": 7856
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6716941595077515,
      "learning_rate": 1.7629662727433058e-05,
      "loss": 0.5541,
      "step": 7857
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7896264791488647,
      "learning_rate": 1.7629004348335006e-05,
      "loss": 0.4891,
      "step": 7858
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6893419027328491,
      "learning_rate": 1.762834589011198e-05,
      "loss": 0.6183,
      "step": 7859
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.717474341392517,
      "learning_rate": 1.762768735277081e-05,
      "loss": 0.5305,
      "step": 7860
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.339488983154297,
      "learning_rate": 1.7627028736318323e-05,
      "loss": 0.5297,
      "step": 7861
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6481080055236816,
      "learning_rate": 1.7626370040761352e-05,
      "loss": 0.5115,
      "step": 7862
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.966591715812683,
      "learning_rate": 1.762571126610673e-05,
      "loss": 0.6167,
      "step": 7863
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7365936040878296,
      "learning_rate": 1.7625052412361287e-05,
      "loss": 0.49,
      "step": 7864
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7884453535079956,
      "learning_rate": 1.7624393479531858e-05,
      "loss": 0.5594,
      "step": 7865
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5872715711593628,
      "learning_rate": 1.762373446762528e-05,
      "loss": 0.4624,
      "step": 7866
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7596923112869263,
      "learning_rate": 1.7623075376648374e-05,
      "loss": 0.5674,
      "step": 7867
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.656012773513794,
      "learning_rate": 1.7622416206607998e-05,
      "loss": 0.476,
      "step": 7868
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7235736846923828,
      "learning_rate": 1.7621756957510968e-05,
      "loss": 0.5454,
      "step": 7869
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6053606271743774,
      "learning_rate": 1.7621097629364136e-05,
      "loss": 0.4775,
      "step": 7870
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8412339687347412,
      "learning_rate": 1.7620438222174335e-05,
      "loss": 0.6197,
      "step": 7871
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8260154724121094,
      "learning_rate": 1.76197787359484e-05,
      "loss": 0.4975,
      "step": 7872
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8725624084472656,
      "learning_rate": 1.761911917069318e-05,
      "loss": 0.6174,
      "step": 7873
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6654497385025024,
      "learning_rate": 1.761845952641551e-05,
      "loss": 0.4756,
      "step": 7874
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7800979614257812,
      "learning_rate": 1.7617799803122227e-05,
      "loss": 0.5425,
      "step": 7875
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.843815803527832,
      "learning_rate": 1.7617140000820183e-05,
      "loss": 0.5413,
      "step": 7876
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7681280374526978,
      "learning_rate": 1.7616480119516215e-05,
      "loss": 0.5308,
      "step": 7877
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7012280225753784,
      "learning_rate": 1.761582015921717e-05,
      "loss": 0.5523,
      "step": 7878
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5786921977996826,
      "learning_rate": 1.761516011992989e-05,
      "loss": 0.4999,
      "step": 7879
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7375556230545044,
      "learning_rate": 1.7614500001661224e-05,
      "loss": 0.5368,
      "step": 7880
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7676500082015991,
      "learning_rate": 1.7613839804418016e-05,
      "loss": 0.5833,
      "step": 7881
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6557891368865967,
      "learning_rate": 1.7613179528207115e-05,
      "loss": 0.4712,
      "step": 7882
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7449754476547241,
      "learning_rate": 1.7612519173035367e-05,
      "loss": 0.6158,
      "step": 7883
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6544978618621826,
      "learning_rate": 1.7611858738909624e-05,
      "loss": 0.5145,
      "step": 7884
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7053618431091309,
      "learning_rate": 1.7611198225836727e-05,
      "loss": 0.5354,
      "step": 7885
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7844743728637695,
      "learning_rate": 1.761053763382354e-05,
      "loss": 0.5342,
      "step": 7886
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8163299560546875,
      "learning_rate": 1.7609876962876907e-05,
      "loss": 0.6302,
      "step": 7887
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6639972925186157,
      "learning_rate": 1.760921621300368e-05,
      "loss": 0.4767,
      "step": 7888
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.74825119972229,
      "learning_rate": 1.7608555384210716e-05,
      "loss": 0.5605,
      "step": 7889
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.705566167831421,
      "learning_rate": 1.7607894476504864e-05,
      "loss": 0.4884,
      "step": 7890
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7087156772613525,
      "learning_rate": 1.760723348989298e-05,
      "loss": 0.5137,
      "step": 7891
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6860542297363281,
      "learning_rate": 1.760657242438192e-05,
      "loss": 0.48,
      "step": 7892
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.789096713066101,
      "learning_rate": 1.7605911279978542e-05,
      "loss": 0.5619,
      "step": 7893
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6309586763381958,
      "learning_rate": 1.7605250056689698e-05,
      "loss": 0.4962,
      "step": 7894
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6623352766036987,
      "learning_rate": 1.7604588754522253e-05,
      "loss": 0.5024,
      "step": 7895
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.647981882095337,
      "learning_rate": 1.760392737348306e-05,
      "loss": 0.4839,
      "step": 7896
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.605717420578003,
      "learning_rate": 1.7603265913578985e-05,
      "loss": 0.5669,
      "step": 7897
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6431583166122437,
      "learning_rate": 1.760260437481688e-05,
      "loss": 0.4902,
      "step": 7898
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6574180126190186,
      "learning_rate": 1.7601942757203612e-05,
      "loss": 0.5742,
      "step": 7899
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.691157341003418,
      "learning_rate": 1.7601281060746044e-05,
      "loss": 0.5551,
      "step": 7900
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7956533432006836,
      "learning_rate": 1.760061928545103e-05,
      "loss": 0.5652,
      "step": 7901
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6907306909561157,
      "learning_rate": 1.7599957431325446e-05,
      "loss": 0.5022,
      "step": 7902
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8512533903121948,
      "learning_rate": 1.7599295498376147e-05,
      "loss": 0.6414,
      "step": 7903
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5507663488388062,
      "learning_rate": 1.7598633486610004e-05,
      "loss": 0.4678,
      "step": 7904
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7127524614334106,
      "learning_rate": 1.759797139603388e-05,
      "loss": 0.547,
      "step": 7905
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6792521476745605,
      "learning_rate": 1.7597309226654642e-05,
      "loss": 0.4963,
      "step": 7906
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6175960302352905,
      "learning_rate": 1.7596646978479158e-05,
      "loss": 0.5277,
      "step": 7907
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7535996437072754,
      "learning_rate": 1.75959846515143e-05,
      "loss": 0.5492,
      "step": 7908
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7241382598876953,
      "learning_rate": 1.7595322245766936e-05,
      "loss": 0.5635,
      "step": 7909
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5829441547393799,
      "learning_rate": 1.759465976124393e-05,
      "loss": 0.4604,
      "step": 7910
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6677706241607666,
      "learning_rate": 1.759399719795216e-05,
      "loss": 0.5451,
      "step": 7911
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6666042804718018,
      "learning_rate": 1.7593334555898493e-05,
      "loss": 0.4873,
      "step": 7912
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6684573888778687,
      "learning_rate": 1.759267183508981e-05,
      "loss": 0.5975,
      "step": 7913
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5942367315292358,
      "learning_rate": 1.7592009035532973e-05,
      "loss": 0.5072,
      "step": 7914
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8515625,
      "learning_rate": 1.7591346157234866e-05,
      "loss": 0.5829,
      "step": 7915
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6297223567962646,
      "learning_rate": 1.7590683200202357e-05,
      "loss": 0.48,
      "step": 7916
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7897002696990967,
      "learning_rate": 1.7590020164442327e-05,
      "loss": 0.5861,
      "step": 7917
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6343544721603394,
      "learning_rate": 1.758935704996165e-05,
      "loss": 0.4801,
      "step": 7918
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7023179531097412,
      "learning_rate": 1.7588693856767206e-05,
      "loss": 0.5585,
      "step": 7919
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7026193141937256,
      "learning_rate": 1.7588030584865873e-05,
      "loss": 0.5163,
      "step": 7920
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.696205735206604,
      "learning_rate": 1.7587367234264526e-05,
      "loss": 0.5908,
      "step": 7921
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7421760559082031,
      "learning_rate": 1.7586703804970047e-05,
      "loss": 0.5102,
      "step": 7922
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6301243305206299,
      "learning_rate": 1.758604029698932e-05,
      "loss": 0.5783,
      "step": 7923
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7266526222229004,
      "learning_rate": 1.7585376710329223e-05,
      "loss": 0.4741,
      "step": 7924
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7908613681793213,
      "learning_rate": 1.758471304499664e-05,
      "loss": 0.5553,
      "step": 7925
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5089471340179443,
      "learning_rate": 1.7584049300998454e-05,
      "loss": 0.4779,
      "step": 7926
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9652432203292847,
      "learning_rate": 1.7583385478341553e-05,
      "loss": 0.5497,
      "step": 7927
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5993648767471313,
      "learning_rate": 1.7582721577032814e-05,
      "loss": 0.4824,
      "step": 7928
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.707007646560669,
      "learning_rate": 1.758205759707913e-05,
      "loss": 0.6413,
      "step": 7929
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6629836559295654,
      "learning_rate": 1.7581393538487378e-05,
      "loss": 0.5058,
      "step": 7930
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7131184339523315,
      "learning_rate": 1.758072940126446e-05,
      "loss": 0.5551,
      "step": 7931
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5855528116226196,
      "learning_rate": 1.7580065185417253e-05,
      "loss": 0.468,
      "step": 7932
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8397833108901978,
      "learning_rate": 1.7579400890952645e-05,
      "loss": 0.5888,
      "step": 7933
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6543056964874268,
      "learning_rate": 1.7578736517877533e-05,
      "loss": 0.4692,
      "step": 7934
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6603810787200928,
      "learning_rate": 1.7578072066198807e-05,
      "loss": 0.518,
      "step": 7935
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7459760904312134,
      "learning_rate": 1.757740753592335e-05,
      "loss": 0.5327,
      "step": 7936
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.638547658920288,
      "learning_rate": 1.7576742927058064e-05,
      "loss": 0.4817,
      "step": 7937
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8812097311019897,
      "learning_rate": 1.7576078239609838e-05,
      "loss": 0.5341,
      "step": 7938
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.700283169746399,
      "learning_rate": 1.7575413473585564e-05,
      "loss": 0.5135,
      "step": 7939
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6158298254013062,
      "learning_rate": 1.757474862899214e-05,
      "loss": 0.4711,
      "step": 7940
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6584221124649048,
      "learning_rate": 1.7574083705836457e-05,
      "loss": 0.5308,
      "step": 7941
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6496081352233887,
      "learning_rate": 1.7573418704125417e-05,
      "loss": 0.4714,
      "step": 7942
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6459301710128784,
      "learning_rate": 1.7572753623865916e-05,
      "loss": 0.5292,
      "step": 7943
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5896340608596802,
      "learning_rate": 1.7572088465064847e-05,
      "loss": 0.473,
      "step": 7944
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7919951677322388,
      "learning_rate": 1.7571423227729113e-05,
      "loss": 0.5256,
      "step": 7945
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6231919527053833,
      "learning_rate": 1.7570757911865615e-05,
      "loss": 0.4951,
      "step": 7946
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5683578252792358,
      "learning_rate": 1.7570092517481248e-05,
      "loss": 0.4971,
      "step": 7947
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.669437050819397,
      "learning_rate": 1.7569427044582917e-05,
      "loss": 0.5137,
      "step": 7948
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7185176610946655,
      "learning_rate": 1.7568761493177527e-05,
      "loss": 0.5919,
      "step": 7949
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7540264129638672,
      "learning_rate": 1.7568095863271977e-05,
      "loss": 0.5163,
      "step": 7950
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6735060214996338,
      "learning_rate": 1.7567430154873167e-05,
      "loss": 0.5267,
      "step": 7951
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6751110553741455,
      "learning_rate": 1.7566764367988006e-05,
      "loss": 0.4794,
      "step": 7952
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.81559419631958,
      "learning_rate": 1.7566098502623404e-05,
      "loss": 0.5876,
      "step": 7953
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5751545429229736,
      "learning_rate": 1.7565432558786257e-05,
      "loss": 0.4407,
      "step": 7954
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8330079317092896,
      "learning_rate": 1.7564766536483478e-05,
      "loss": 0.5996,
      "step": 7955
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6266958713531494,
      "learning_rate": 1.756410043572197e-05,
      "loss": 0.5052,
      "step": 7956
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8639214038848877,
      "learning_rate": 1.7563434256508646e-05,
      "loss": 0.5707,
      "step": 7957
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7153633832931519,
      "learning_rate": 1.7562767998850417e-05,
      "loss": 0.549,
      "step": 7958
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7060904502868652,
      "learning_rate": 1.756210166275419e-05,
      "loss": 0.5332,
      "step": 7959
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8693631887435913,
      "learning_rate": 1.7561435248226874e-05,
      "loss": 0.5309,
      "step": 7960
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.81082284450531,
      "learning_rate": 1.7560768755275385e-05,
      "loss": 0.5348,
      "step": 7961
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.738036870956421,
      "learning_rate": 1.756010218390663e-05,
      "loss": 0.5342,
      "step": 7962
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6771154403686523,
      "learning_rate": 1.7559435534127534e-05,
      "loss": 0.5297,
      "step": 7963
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6703753471374512,
      "learning_rate": 1.7558768805945e-05,
      "loss": 0.5289,
      "step": 7964
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5810478925704956,
      "learning_rate": 1.7558101999365944e-05,
      "loss": 0.5604,
      "step": 7965
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.907687783241272,
      "learning_rate": 1.7557435114397283e-05,
      "loss": 0.5423,
      "step": 7966
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6065609455108643,
      "learning_rate": 1.755676815104594e-05,
      "loss": 0.5272,
      "step": 7967
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7919893264770508,
      "learning_rate": 1.7556101109318823e-05,
      "loss": 0.5287,
      "step": 7968
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6077171564102173,
      "learning_rate": 1.755543398922286e-05,
      "loss": 0.512,
      "step": 7969
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.675650715827942,
      "learning_rate": 1.755476679076496e-05,
      "loss": 0.4779,
      "step": 7970
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7748266458511353,
      "learning_rate": 1.7554099513952046e-05,
      "loss": 0.5444,
      "step": 7971
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6313042640686035,
      "learning_rate": 1.7553432158791045e-05,
      "loss": 0.4845,
      "step": 7972
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6843942403793335,
      "learning_rate": 1.7552764725288872e-05,
      "loss": 0.5343,
      "step": 7973
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6315155029296875,
      "learning_rate": 1.755209721345245e-05,
      "loss": 0.4761,
      "step": 7974
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7474730014801025,
      "learning_rate": 1.755142962328871e-05,
      "loss": 0.5774,
      "step": 7975
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6137219667434692,
      "learning_rate": 1.7550761954804562e-05,
      "loss": 0.4549,
      "step": 7976
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.762566328048706,
      "learning_rate": 1.755009420800694e-05,
      "loss": 0.5428,
      "step": 7977
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.70283842086792,
      "learning_rate": 1.754942638290277e-05,
      "loss": 0.4989,
      "step": 7978
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7655278444290161,
      "learning_rate": 1.7548758479498974e-05,
      "loss": 0.5879,
      "step": 7979
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6258505582809448,
      "learning_rate": 1.754809049780248e-05,
      "loss": 0.467,
      "step": 7980
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6470180749893188,
      "learning_rate": 1.754742243782022e-05,
      "loss": 0.5486,
      "step": 7981
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6436209678649902,
      "learning_rate": 1.754675429955912e-05,
      "loss": 0.4512,
      "step": 7982
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6859800815582275,
      "learning_rate": 1.754608608302611e-05,
      "loss": 0.531,
      "step": 7983
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6034313440322876,
      "learning_rate": 1.754541778822812e-05,
      "loss": 0.4751,
      "step": 7984
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.92967689037323,
      "learning_rate": 1.7544749415172083e-05,
      "loss": 0.5726,
      "step": 7985
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6332619190216064,
      "learning_rate": 1.7544080963864932e-05,
      "loss": 0.4616,
      "step": 7986
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6320209503173828,
      "learning_rate": 1.7543412434313597e-05,
      "loss": 0.5338,
      "step": 7987
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6146067380905151,
      "learning_rate": 1.754274382652501e-05,
      "loss": 0.4895,
      "step": 7988
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.896399974822998,
      "learning_rate": 1.7542075140506107e-05,
      "loss": 0.599,
      "step": 7989
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6597024202346802,
      "learning_rate": 1.7541406376263827e-05,
      "loss": 0.4556,
      "step": 7990
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.078284740447998,
      "learning_rate": 1.7540737533805104e-05,
      "loss": 1.0401,
      "step": 7991
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5709887742996216,
      "learning_rate": 1.7540068613136875e-05,
      "loss": 1.043,
      "step": 7992
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.847017765045166,
      "learning_rate": 1.753939961426607e-05,
      "loss": 0.4838,
      "step": 7993
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8239054679870605,
      "learning_rate": 1.7538730537199643e-05,
      "loss": 0.5015,
      "step": 7994
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7315014600753784,
      "learning_rate": 1.7538061381944524e-05,
      "loss": 0.5215,
      "step": 7995
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9158875942230225,
      "learning_rate": 1.753739214850765e-05,
      "loss": 0.5288,
      "step": 7996
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.731856346130371,
      "learning_rate": 1.7536722836895973e-05,
      "loss": 0.5752,
      "step": 7997
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.707046627998352,
      "learning_rate": 1.7536053447116425e-05,
      "loss": 0.4882,
      "step": 7998
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8689223527908325,
      "learning_rate": 1.7535383979175954e-05,
      "loss": 0.5468,
      "step": 7999
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6877212524414062,
      "learning_rate": 1.75347144330815e-05,
      "loss": 0.4698,
      "step": 8000
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.637708067893982,
      "learning_rate": 1.7534044808840007e-05,
      "loss": 0.5011,
      "step": 8001
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6962339878082275,
      "learning_rate": 1.7533375106458423e-05,
      "loss": 0.5115,
      "step": 8002
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7979685068130493,
      "learning_rate": 1.7532705325943692e-05,
      "loss": 0.517,
      "step": 8003
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8378677368164062,
      "learning_rate": 1.7532035467302764e-05,
      "loss": 0.5522,
      "step": 8004
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8045302629470825,
      "learning_rate": 1.753136553054258e-05,
      "loss": 0.5655,
      "step": 8005
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7947455644607544,
      "learning_rate": 1.7530695515670095e-05,
      "loss": 0.5096,
      "step": 8006
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5897544622421265,
      "learning_rate": 1.7530025422692255e-05,
      "loss": 0.4902,
      "step": 8007
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6794970035552979,
      "learning_rate": 1.7529355251616008e-05,
      "loss": 0.4767,
      "step": 8008
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7118570804595947,
      "learning_rate": 1.752868500244831e-05,
      "loss": 0.5139,
      "step": 8009
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7571274042129517,
      "learning_rate": 1.752801467519611e-05,
      "loss": 0.5428,
      "step": 8010
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6438921689987183,
      "learning_rate": 1.7527344269866358e-05,
      "loss": 0.505,
      "step": 8011
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7365695238113403,
      "learning_rate": 1.752667378646601e-05,
      "loss": 0.5454,
      "step": 8012
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6570065021514893,
      "learning_rate": 1.752600322500202e-05,
      "loss": 0.5102,
      "step": 8013
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7457308769226074,
      "learning_rate": 1.7525332585481337e-05,
      "loss": 0.5507,
      "step": 8014
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6539816856384277,
      "learning_rate": 1.7524661867910925e-05,
      "loss": 0.5455,
      "step": 8015
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6849710941314697,
      "learning_rate": 1.7523991072297737e-05,
      "loss": 0.4748,
      "step": 8016
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6767324209213257,
      "learning_rate": 1.752332019864873e-05,
      "loss": 0.544,
      "step": 8017
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6205028295516968,
      "learning_rate": 1.752264924697086e-05,
      "loss": 0.4544,
      "step": 8018
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9267646074295044,
      "learning_rate": 1.7521978217271092e-05,
      "loss": 0.5928,
      "step": 8019
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.712898850440979,
      "learning_rate": 1.7521307109556377e-05,
      "loss": 0.492,
      "step": 8020
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6785041093826294,
      "learning_rate": 1.7520635923833684e-05,
      "loss": 0.521,
      "step": 8021
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.781839370727539,
      "learning_rate": 1.751996466010997e-05,
      "loss": 0.5031,
      "step": 8022
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6259690523147583,
      "learning_rate": 1.7519293318392198e-05,
      "loss": 0.5805,
      "step": 8023
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6371129751205444,
      "learning_rate": 1.7518621898687326e-05,
      "loss": 0.493,
      "step": 8024
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7986799478530884,
      "learning_rate": 1.7517950401002325e-05,
      "loss": 0.5821,
      "step": 8025
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7905724048614502,
      "learning_rate": 1.7517278825344156e-05,
      "loss": 0.517,
      "step": 8026
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6827161312103271,
      "learning_rate": 1.7516607171719786e-05,
      "loss": 0.5076,
      "step": 8027
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.702620029449463,
      "learning_rate": 1.751593544013618e-05,
      "loss": 0.5326,
      "step": 8028
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1340742111206055,
      "learning_rate": 1.7515263630600302e-05,
      "loss": 1.1517,
      "step": 8029
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7382235527038574,
      "learning_rate": 1.7514591743119127e-05,
      "loss": 1.0649,
      "step": 8030
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.742613434791565,
      "learning_rate": 1.7513919777699618e-05,
      "loss": 0.55,
      "step": 8031
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8400498628616333,
      "learning_rate": 1.7513247734348745e-05,
      "loss": 0.5105,
      "step": 8032
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6507099866867065,
      "learning_rate": 1.7512575613073476e-05,
      "loss": 0.604,
      "step": 8033
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6562504768371582,
      "learning_rate": 1.7511903413880785e-05,
      "loss": 0.4854,
      "step": 8034
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5928939580917358,
      "learning_rate": 1.7511231136777648e-05,
      "loss": 0.5135,
      "step": 8035
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5539662837982178,
      "learning_rate": 1.751055878177103e-05,
      "loss": 0.477,
      "step": 8036
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8075320720672607,
      "learning_rate": 1.750988634886791e-05,
      "loss": 0.6105,
      "step": 8037
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7221970558166504,
      "learning_rate": 1.7509213838075253e-05,
      "loss": 0.4902,
      "step": 8038
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7223913669586182,
      "learning_rate": 1.7508541249400045e-05,
      "loss": 0.5212,
      "step": 8039
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6879961490631104,
      "learning_rate": 1.7507868582849258e-05,
      "loss": 0.4639,
      "step": 8040
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6657804250717163,
      "learning_rate": 1.7507195838429868e-05,
      "loss": 0.5361,
      "step": 8041
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6921762228012085,
      "learning_rate": 1.7506523016148855e-05,
      "loss": 0.4731,
      "step": 8042
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7551549673080444,
      "learning_rate": 1.750585011601319e-05,
      "loss": 0.5622,
      "step": 8043
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6864839792251587,
      "learning_rate": 1.750517713802986e-05,
      "loss": 0.4966,
      "step": 8044
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6932419538497925,
      "learning_rate": 1.750450408220584e-05,
      "loss": 0.5611,
      "step": 8045
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.739759922027588,
      "learning_rate": 1.750383094854811e-05,
      "loss": 0.5294,
      "step": 8046
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7091865539550781,
      "learning_rate": 1.7503157737063657e-05,
      "loss": 0.5339,
      "step": 8047
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6747066974639893,
      "learning_rate": 1.750248444775946e-05,
      "loss": 0.4942,
      "step": 8048
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.681501030921936,
      "learning_rate": 1.7501811080642503e-05,
      "loss": 0.5411,
      "step": 8049
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7593215703964233,
      "learning_rate": 1.7501137635719767e-05,
      "loss": 0.5434,
      "step": 8050
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.830735683441162,
      "learning_rate": 1.7500464112998242e-05,
      "loss": 0.5646,
      "step": 8051
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.770992636680603,
      "learning_rate": 1.7499790512484905e-05,
      "loss": 0.4713,
      "step": 8052
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7687351703643799,
      "learning_rate": 1.749911683418675e-05,
      "loss": 0.5567,
      "step": 8053
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6092602014541626,
      "learning_rate": 1.7498443078110765e-05,
      "loss": 0.4497,
      "step": 8054
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8475451469421387,
      "learning_rate": 1.749776924426393e-05,
      "loss": 0.5946,
      "step": 8055
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6713651418685913,
      "learning_rate": 1.7497095332653242e-05,
      "loss": 0.4708,
      "step": 8056
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.46069073677063,
      "learning_rate": 1.7496421343285687e-05,
      "loss": 1.0295,
      "step": 8057
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8979376554489136,
      "learning_rate": 1.7495747276168253e-05,
      "loss": 1.0582,
      "step": 8058
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7326234579086304,
      "learning_rate": 1.7495073131307933e-05,
      "loss": 0.5577,
      "step": 8059
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9247692823410034,
      "learning_rate": 1.7494398908711722e-05,
      "loss": 0.5157,
      "step": 8060
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8350465297698975,
      "learning_rate": 1.749372460838661e-05,
      "loss": 0.5584,
      "step": 8061
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.73919677734375,
      "learning_rate": 1.749305023033959e-05,
      "loss": 0.4671,
      "step": 8062
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0702598094940186,
      "learning_rate": 1.7492375774577654e-05,
      "loss": 0.5914,
      "step": 8063
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.746734380722046,
      "learning_rate": 1.7491701241107803e-05,
      "loss": 0.4756,
      "step": 8064
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7419016361236572,
      "learning_rate": 1.7491026629937027e-05,
      "loss": 0.5774,
      "step": 8065
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6161725521087646,
      "learning_rate": 1.749035194107233e-05,
      "loss": 0.4892,
      "step": 8066
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7227057218551636,
      "learning_rate": 1.7489677174520707e-05,
      "loss": 0.6139,
      "step": 8067
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.740359902381897,
      "learning_rate": 1.7489002330289155e-05,
      "loss": 0.4782,
      "step": 8068
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6640881299972534,
      "learning_rate": 1.7488327408384668e-05,
      "loss": 0.5376,
      "step": 8069
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6666818857192993,
      "learning_rate": 1.7487652408814256e-05,
      "loss": 0.4669,
      "step": 8070
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7432897090911865,
      "learning_rate": 1.7486977331584912e-05,
      "loss": 0.5395,
      "step": 8071
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6957519054412842,
      "learning_rate": 1.7486302176703644e-05,
      "loss": 0.5102,
      "step": 8072
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6861220598220825,
      "learning_rate": 1.748562694417745e-05,
      "loss": 0.5522,
      "step": 8073
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5986875295639038,
      "learning_rate": 1.7484951634013335e-05,
      "loss": 0.4609,
      "step": 8074
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8108121156692505,
      "learning_rate": 1.7484276246218304e-05,
      "loss": 0.6142,
      "step": 8075
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7099881172180176,
      "learning_rate": 1.7483600780799358e-05,
      "loss": 0.4851,
      "step": 8076
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6616400480270386,
      "learning_rate": 1.7482925237763504e-05,
      "loss": 0.5654,
      "step": 8077
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.642082691192627,
      "learning_rate": 1.748224961711775e-05,
      "loss": 0.4947,
      "step": 8078
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6001958847045898,
      "learning_rate": 1.7481573918869105e-05,
      "loss": 0.5012,
      "step": 8079
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6221227645874023,
      "learning_rate": 1.7480898143024574e-05,
      "loss": 0.4442,
      "step": 8080
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8187873363494873,
      "learning_rate": 1.7480222289591165e-05,
      "loss": 0.5705,
      "step": 8081
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7526262998580933,
      "learning_rate": 1.747954635857589e-05,
      "loss": 0.4871,
      "step": 8082
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6632803678512573,
      "learning_rate": 1.7478870349985758e-05,
      "loss": 0.5167,
      "step": 8083
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.652687907218933,
      "learning_rate": 1.7478194263827783e-05,
      "loss": 0.5376,
      "step": 8084
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6654114723205566,
      "learning_rate": 1.7477518100108977e-05,
      "loss": 0.5461,
      "step": 8085
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8759958744049072,
      "learning_rate": 1.747684185883635e-05,
      "loss": 0.5906,
      "step": 8086
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7427457571029663,
      "learning_rate": 1.7476165540016912e-05,
      "loss": 0.5761,
      "step": 8087
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.644044041633606,
      "learning_rate": 1.7475489143657688e-05,
      "loss": 0.5141,
      "step": 8088
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5748350620269775,
      "learning_rate": 1.7474812669765682e-05,
      "loss": 0.5103,
      "step": 8089
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6497864723205566,
      "learning_rate": 1.7474136118347922e-05,
      "loss": 0.487,
      "step": 8090
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.62385094165802,
      "learning_rate": 1.7473459489411414e-05,
      "loss": 0.5618,
      "step": 8091
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6373363733291626,
      "learning_rate": 1.7472782782963183e-05,
      "loss": 0.5154,
      "step": 8092
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7009814977645874,
      "learning_rate": 1.7472105999010243e-05,
      "loss": 0.529,
      "step": 8093
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6834642887115479,
      "learning_rate": 1.747142913755962e-05,
      "loss": 0.5153,
      "step": 8094
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6876240968704224,
      "learning_rate": 1.7470752198618327e-05,
      "loss": 0.5264,
      "step": 8095
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7136279344558716,
      "learning_rate": 1.7470075182193384e-05,
      "loss": 0.558,
      "step": 8096
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6868326663970947,
      "learning_rate": 1.746939808829182e-05,
      "loss": 0.5202,
      "step": 8097
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7848531007766724,
      "learning_rate": 1.7468720916920656e-05,
      "loss": 0.533,
      "step": 8098
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7524782419204712,
      "learning_rate": 1.7468043668086907e-05,
      "loss": 0.5554,
      "step": 8099
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6445358991622925,
      "learning_rate": 1.7467366341797603e-05,
      "loss": 0.5209,
      "step": 8100
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6859537363052368,
      "learning_rate": 1.7466688938059774e-05,
      "loss": 0.587,
      "step": 8101
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6710855960845947,
      "learning_rate": 1.7466011456880436e-05,
      "loss": 0.5187,
      "step": 8102
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7270811796188354,
      "learning_rate": 1.7465333898266627e-05,
      "loss": 0.5664,
      "step": 8103
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5721523761749268,
      "learning_rate": 1.746465626222536e-05,
      "loss": 0.4786,
      "step": 8104
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7589852809906006,
      "learning_rate": 1.7463978548763675e-05,
      "loss": 0.565,
      "step": 8105
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0844790935516357,
      "learning_rate": 1.7463300757888597e-05,
      "loss": 0.486,
      "step": 8106
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5584532022476196,
      "learning_rate": 1.7462622889607156e-05,
      "loss": 0.5532,
      "step": 8107
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8562595844268799,
      "learning_rate": 1.746194494392638e-05,
      "loss": 0.497,
      "step": 8108
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7637966871261597,
      "learning_rate": 1.7461266920853307e-05,
      "loss": 0.6007,
      "step": 8109
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7168819904327393,
      "learning_rate": 1.7460588820394958e-05,
      "loss": 0.4929,
      "step": 8110
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6666316986083984,
      "learning_rate": 1.745991064255838e-05,
      "loss": 0.5223,
      "step": 8111
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7073678970336914,
      "learning_rate": 1.7459232387350597e-05,
      "loss": 0.5215,
      "step": 8112
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.75856351852417,
      "learning_rate": 1.7458554054778644e-05,
      "loss": 0.5234,
      "step": 8113
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.675075888633728,
      "learning_rate": 1.7457875644849563e-05,
      "loss": 0.4838,
      "step": 8114
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.779874324798584,
      "learning_rate": 1.7457197157570383e-05,
      "loss": 0.5273,
      "step": 8115
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.685380220413208,
      "learning_rate": 1.7456518592948143e-05,
      "loss": 0.4677,
      "step": 8116
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6662803888320923,
      "learning_rate": 1.7455839950989883e-05,
      "loss": 0.526,
      "step": 8117
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7526512145996094,
      "learning_rate": 1.7455161231702642e-05,
      "loss": 0.5285,
      "step": 8118
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7015385627746582,
      "learning_rate": 1.7454482435093453e-05,
      "loss": 0.5499,
      "step": 8119
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.713955044746399,
      "learning_rate": 1.7453803561169364e-05,
      "loss": 0.476,
      "step": 8120
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.691512942314148,
      "learning_rate": 1.7453124609937414e-05,
      "loss": 0.524,
      "step": 8121
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7229812145233154,
      "learning_rate": 1.745244558140464e-05,
      "loss": 0.5308,
      "step": 8122
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7877423763275146,
      "learning_rate": 1.745176647557809e-05,
      "loss": 0.5285,
      "step": 8123
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7619473934173584,
      "learning_rate": 1.7451087292464804e-05,
      "loss": 0.5231,
      "step": 8124
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6473978757858276,
      "learning_rate": 1.7450408032071832e-05,
      "loss": 0.5612,
      "step": 8125
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6675353050231934,
      "learning_rate": 1.7449728694406213e-05,
      "loss": 0.4844,
      "step": 8126
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6644529104232788,
      "learning_rate": 1.7449049279474993e-05,
      "loss": 0.519,
      "step": 8127
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.675300121307373,
      "learning_rate": 1.744836978728522e-05,
      "loss": 0.4976,
      "step": 8128
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.660796046257019,
      "learning_rate": 1.7447690217843947e-05,
      "loss": 0.5115,
      "step": 8129
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7036073207855225,
      "learning_rate": 1.744701057115821e-05,
      "loss": 0.5513,
      "step": 8130
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7432148456573486,
      "learning_rate": 1.744633084723507e-05,
      "loss": 0.5684,
      "step": 8131
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6150412559509277,
      "learning_rate": 1.744565104608157e-05,
      "loss": 0.4792,
      "step": 8132
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.696122407913208,
      "learning_rate": 1.7444971167704765e-05,
      "loss": 0.539,
      "step": 8133
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.654982089996338,
      "learning_rate": 1.74442912121117e-05,
      "loss": 0.4967,
      "step": 8134
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6380372047424316,
      "learning_rate": 1.7443611179309435e-05,
      "loss": 0.5365,
      "step": 8135
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6707290410995483,
      "learning_rate": 1.7442931069305015e-05,
      "loss": 0.4996,
      "step": 8136
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6355631351470947,
      "learning_rate": 1.74422508821055e-05,
      "loss": 0.568,
      "step": 8137
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7149505615234375,
      "learning_rate": 1.7441570617717944e-05,
      "loss": 0.4911,
      "step": 8138
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6740549802780151,
      "learning_rate": 1.7440890276149397e-05,
      "loss": 0.5766,
      "step": 8139
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5954651832580566,
      "learning_rate": 1.7440209857406923e-05,
      "loss": 0.4778,
      "step": 8140
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6393988132476807,
      "learning_rate": 1.7439529361497572e-05,
      "loss": 0.5373,
      "step": 8141
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6311533451080322,
      "learning_rate": 1.743884878842841e-05,
      "loss": 0.4668,
      "step": 8142
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7850934267044067,
      "learning_rate": 1.7438168138206488e-05,
      "loss": 0.6133,
      "step": 8143
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7361637353897095,
      "learning_rate": 1.7437487410838865e-05,
      "loss": 0.5072,
      "step": 8144
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7995476722717285,
      "learning_rate": 1.7436806606332606e-05,
      "loss": 0.5889,
      "step": 8145
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6093837022781372,
      "learning_rate": 1.7436125724694775e-05,
      "loss": 0.4548,
      "step": 8146
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6369210481643677,
      "learning_rate": 1.7435444765932422e-05,
      "loss": 0.5018,
      "step": 8147
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.639745831489563,
      "learning_rate": 1.7434763730052625e-05,
      "loss": 0.463,
      "step": 8148
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.681957483291626,
      "learning_rate": 1.7434082617062435e-05,
      "loss": 0.5177,
      "step": 8149
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8870460987091064,
      "learning_rate": 1.743340142696892e-05,
      "loss": 0.5793,
      "step": 8150
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.775174617767334,
      "learning_rate": 1.7432720159779147e-05,
      "loss": 0.5602,
      "step": 8151
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6551628112792969,
      "learning_rate": 1.743203881550018e-05,
      "loss": 0.4673,
      "step": 8152
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7850497961044312,
      "learning_rate": 1.7431357394139087e-05,
      "loss": 0.5488,
      "step": 8153
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7087773084640503,
      "learning_rate": 1.7430675895702932e-05,
      "loss": 0.5075,
      "step": 8154
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.747334361076355,
      "learning_rate": 1.7429994320198787e-05,
      "loss": 0.5818,
      "step": 8155
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6118460893630981,
      "learning_rate": 1.742931266763372e-05,
      "loss": 0.4763,
      "step": 8156
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7172759771347046,
      "learning_rate": 1.74286309380148e-05,
      "loss": 0.5462,
      "step": 8157
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6355253458023071,
      "learning_rate": 1.7427949131349098e-05,
      "loss": 0.5012,
      "step": 8158
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7213408946990967,
      "learning_rate": 1.7427267247643684e-05,
      "loss": 0.5474,
      "step": 8159
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5958961248397827,
      "learning_rate": 1.7426585286905634e-05,
      "loss": 0.4664,
      "step": 8160
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6842682361602783,
      "learning_rate": 1.7425903249142023e-05,
      "loss": 0.5258,
      "step": 8161
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.686570167541504,
      "learning_rate": 1.7425221134359915e-05,
      "loss": 0.4747,
      "step": 8162
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6871901750564575,
      "learning_rate": 1.742453894256639e-05,
      "loss": 0.5043,
      "step": 8163
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.524735927581787,
      "learning_rate": 1.7423856673768527e-05,
      "loss": 0.4463,
      "step": 8164
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5972143411636353,
      "learning_rate": 1.74231743279734e-05,
      "loss": 0.4976,
      "step": 8165
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7213938236236572,
      "learning_rate": 1.742249190518808e-05,
      "loss": 0.5146,
      "step": 8166
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7238361835479736,
      "learning_rate": 1.7421809405419653e-05,
      "loss": 0.5876,
      "step": 8167
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6935789585113525,
      "learning_rate": 1.742112682867519e-05,
      "loss": 0.5168,
      "step": 8168
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7208499908447266,
      "learning_rate": 1.742044417496178e-05,
      "loss": 0.6029,
      "step": 8169
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6366335153579712,
      "learning_rate": 1.74197614442865e-05,
      "loss": 0.5149,
      "step": 8170
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7037503719329834,
      "learning_rate": 1.7419078636656425e-05,
      "loss": 0.5642,
      "step": 8171
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.686016321182251,
      "learning_rate": 1.741839575207864e-05,
      "loss": 0.4782,
      "step": 8172
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.673075795173645,
      "learning_rate": 1.7417712790560228e-05,
      "loss": 0.479,
      "step": 8173
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6024487018585205,
      "learning_rate": 1.7417029752108278e-05,
      "loss": 0.485,
      "step": 8174
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.673791527748108,
      "learning_rate": 1.7416346636729863e-05,
      "loss": 0.4852,
      "step": 8175
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.754632592201233,
      "learning_rate": 1.7415663444432077e-05,
      "loss": 0.5673,
      "step": 8176
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7268871068954468,
      "learning_rate": 1.7414980175222003e-05,
      "loss": 0.5513,
      "step": 8177
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6366057395935059,
      "learning_rate": 1.7414296829106728e-05,
      "loss": 0.4967,
      "step": 8178
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6153206825256348,
      "learning_rate": 1.7413613406093336e-05,
      "loss": 0.4928,
      "step": 8179
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7326629161834717,
      "learning_rate": 1.741292990618892e-05,
      "loss": 0.5553,
      "step": 8180
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6907800436019897,
      "learning_rate": 1.7412246329400568e-05,
      "loss": 0.524,
      "step": 8181
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6651818752288818,
      "learning_rate": 1.7411562675735367e-05,
      "loss": 0.4521,
      "step": 8182
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6327029466629028,
      "learning_rate": 1.7410878945200413e-05,
      "loss": 0.5172,
      "step": 8183
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7504277229309082,
      "learning_rate": 1.741019513780279e-05,
      "loss": 0.5212,
      "step": 8184
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6265506744384766,
      "learning_rate": 1.7409511253549592e-05,
      "loss": 0.5534,
      "step": 8185
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.535555362701416,
      "learning_rate": 1.740882729244792e-05,
      "loss": 0.4383,
      "step": 8186
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5836360454559326,
      "learning_rate": 1.7408143254504857e-05,
      "loss": 0.5201,
      "step": 8187
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6453564167022705,
      "learning_rate": 1.74074591397275e-05,
      "loss": 0.5097,
      "step": 8188
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8375025987625122,
      "learning_rate": 1.740677494812295e-05,
      "loss": 0.5702,
      "step": 8189
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6182948350906372,
      "learning_rate": 1.74060906796983e-05,
      "loss": 0.4816,
      "step": 8190
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.734927773475647,
      "learning_rate": 1.7405406334460647e-05,
      "loss": 0.5699,
      "step": 8191
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.633570909500122,
      "learning_rate": 1.7404721912417088e-05,
      "loss": 0.4799,
      "step": 8192
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9134012460708618,
      "learning_rate": 1.7404037413574722e-05,
      "loss": 0.5996,
      "step": 8193
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.605228066444397,
      "learning_rate": 1.7403352837940647e-05,
      "loss": 0.4799,
      "step": 8194
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7012455463409424,
      "learning_rate": 1.740266818552197e-05,
      "loss": 0.5322,
      "step": 8195
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.747080683708191,
      "learning_rate": 1.740198345632578e-05,
      "loss": 0.5032,
      "step": 8196
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6436508893966675,
      "learning_rate": 1.7401298650359185e-05,
      "loss": 0.5428,
      "step": 8197
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.561108112335205,
      "learning_rate": 1.7400613767629293e-05,
      "loss": 0.4735,
      "step": 8198
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8128979206085205,
      "learning_rate": 1.7399928808143197e-05,
      "loss": 0.5443,
      "step": 8199
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7049504518508911,
      "learning_rate": 1.7399243771908008e-05,
      "loss": 0.496,
      "step": 8200
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8064823150634766,
      "learning_rate": 1.739855865893083e-05,
      "loss": 0.5667,
      "step": 8201
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5692988634109497,
      "learning_rate": 1.7397873469218768e-05,
      "loss": 0.4417,
      "step": 8202
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7345224618911743,
      "learning_rate": 1.7397188202778926e-05,
      "loss": 0.5809,
      "step": 8203
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6273274421691895,
      "learning_rate": 1.7396502859618418e-05,
      "loss": 0.4604,
      "step": 8204
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7225106954574585,
      "learning_rate": 1.7395817439744344e-05,
      "loss": 0.6042,
      "step": 8205
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5950969457626343,
      "learning_rate": 1.7395131943163818e-05,
      "loss": 0.4783,
      "step": 8206
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7984458208084106,
      "learning_rate": 1.739444636988395e-05,
      "loss": 0.5614,
      "step": 8207
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6217721700668335,
      "learning_rate": 1.7393760719911845e-05,
      "loss": 0.4842,
      "step": 8208
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6973764896392822,
      "learning_rate": 1.739307499325462e-05,
      "loss": 0.5453,
      "step": 8209
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5686026811599731,
      "learning_rate": 1.7392389189919386e-05,
      "loss": 0.4786,
      "step": 8210
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6921048164367676,
      "learning_rate": 1.739170330991325e-05,
      "loss": 0.5234,
      "step": 8211
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.682496428489685,
      "learning_rate": 1.7391017353243337e-05,
      "loss": 0.5035,
      "step": 8212
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6136598587036133,
      "learning_rate": 1.7390331319916755e-05,
      "loss": 0.5483,
      "step": 8213
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6338540315628052,
      "learning_rate": 1.7389645209940616e-05,
      "loss": 0.4903,
      "step": 8214
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.720808982849121,
      "learning_rate": 1.738895902332204e-05,
      "loss": 0.5716,
      "step": 8215
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.653243064880371,
      "learning_rate": 1.7388272760068147e-05,
      "loss": 0.4896,
      "step": 8216
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6961982250213623,
      "learning_rate": 1.738758642018605e-05,
      "loss": 0.5519,
      "step": 8217
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.676287055015564,
      "learning_rate": 1.7386900003682867e-05,
      "loss": 0.4622,
      "step": 8218
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9459198713302612,
      "learning_rate": 1.7386213510565716e-05,
      "loss": 0.5234,
      "step": 8219
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6583739519119263,
      "learning_rate": 1.7385526940841723e-05,
      "loss": 0.5111,
      "step": 8220
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7328156232833862,
      "learning_rate": 1.7384840294518004e-05,
      "loss": 0.535,
      "step": 8221
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5492092370986938,
      "learning_rate": 1.7384153571601685e-05,
      "loss": 0.4707,
      "step": 8222
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.726133108139038,
      "learning_rate": 1.738346677209988e-05,
      "loss": 0.6083,
      "step": 8223
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6475915908813477,
      "learning_rate": 1.7382779896019727e-05,
      "loss": 0.4689,
      "step": 8224
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.665625810623169,
      "learning_rate": 1.7382092943368334e-05,
      "loss": 0.518,
      "step": 8225
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6169285774230957,
      "learning_rate": 1.7381405914152834e-05,
      "loss": 0.4833,
      "step": 8226
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.904554009437561,
      "learning_rate": 1.738071880838035e-05,
      "loss": 0.4907,
      "step": 8227
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7463819980621338,
      "learning_rate": 1.738003162605801e-05,
      "loss": 0.5187,
      "step": 8228
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5945539474487305,
      "learning_rate": 1.7379344367192942e-05,
      "loss": 0.5616,
      "step": 8229
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.662097692489624,
      "learning_rate": 1.7378657031792273e-05,
      "loss": 0.5164,
      "step": 8230
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6192595958709717,
      "learning_rate": 1.737796961986313e-05,
      "loss": 0.5092,
      "step": 8231
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7339527606964111,
      "learning_rate": 1.7377282131412647e-05,
      "loss": 0.5314,
      "step": 8232
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7046056985855103,
      "learning_rate": 1.737659456644795e-05,
      "loss": 0.5744,
      "step": 8233
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6893991231918335,
      "learning_rate": 1.737590692497617e-05,
      "loss": 0.4968,
      "step": 8234
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7114481925964355,
      "learning_rate": 1.7375219207004442e-05,
      "loss": 0.5768,
      "step": 8235
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6831971406936646,
      "learning_rate": 1.7374531412539895e-05,
      "loss": 0.4825,
      "step": 8236
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7204440832138062,
      "learning_rate": 1.737384354158967e-05,
      "loss": 0.5558,
      "step": 8237
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6581450700759888,
      "learning_rate": 1.737315559416089e-05,
      "loss": 0.4724,
      "step": 8238
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7391563653945923,
      "learning_rate": 1.73724675702607e-05,
      "loss": 0.5438,
      "step": 8239
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6513739824295044,
      "learning_rate": 1.7371779469896228e-05,
      "loss": 0.481,
      "step": 8240
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9524437189102173,
      "learning_rate": 1.737109129307462e-05,
      "loss": 1.0689,
      "step": 8241
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6699824333190918,
      "learning_rate": 1.7370403039803007e-05,
      "loss": 1.0819,
      "step": 8242
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8513243198394775,
      "learning_rate": 1.7369714710088525e-05,
      "loss": 0.5581,
      "step": 8243
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.808366298675537,
      "learning_rate": 1.7369026303938322e-05,
      "loss": 0.4678,
      "step": 8244
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6549590826034546,
      "learning_rate": 1.736833782135953e-05,
      "loss": 0.4968,
      "step": 8245
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6856886148452759,
      "learning_rate": 1.736764926235929e-05,
      "loss": 0.5024,
      "step": 8246
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6225476264953613,
      "learning_rate": 1.736696062694475e-05,
      "loss": 0.5234,
      "step": 8247
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6665455102920532,
      "learning_rate": 1.7366271915123043e-05,
      "loss": 0.4988,
      "step": 8248
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7244914770126343,
      "learning_rate": 1.7365583126901323e-05,
      "loss": 0.5433,
      "step": 8249
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6558235883712769,
      "learning_rate": 1.736489426228672e-05,
      "loss": 0.5441,
      "step": 8250
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.744101881980896,
      "learning_rate": 1.7364205321286393e-05,
      "loss": 0.5042,
      "step": 8251
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6026886701583862,
      "learning_rate": 1.7363516303907477e-05,
      "loss": 0.4759,
      "step": 8252
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7670223712921143,
      "learning_rate": 1.7362827210157127e-05,
      "loss": 0.6108,
      "step": 8253
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5668690204620361,
      "learning_rate": 1.736213804004248e-05,
      "loss": 0.4743,
      "step": 8254
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6529252529144287,
      "learning_rate": 1.736144879357069e-05,
      "loss": 0.5579,
      "step": 8255
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6555143594741821,
      "learning_rate": 1.7360759470748907e-05,
      "loss": 0.4777,
      "step": 8256
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5924513339996338,
      "learning_rate": 1.7360070071584277e-05,
      "loss": 0.5132,
      "step": 8257
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6125580072402954,
      "learning_rate": 1.7359380596083947e-05,
      "loss": 0.489,
      "step": 8258
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.753892421722412,
      "learning_rate": 1.735869104425508e-05,
      "loss": 0.5808,
      "step": 8259
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6536837816238403,
      "learning_rate": 1.735800141610481e-05,
      "loss": 0.4764,
      "step": 8260
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.5061862468719482,
      "learning_rate": 1.7357311711640305e-05,
      "loss": 0.9717,
      "step": 8261
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.214672088623047,
      "learning_rate": 1.7356621930868714e-05,
      "loss": 1.0844,
      "step": 8262
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9816964864730835,
      "learning_rate": 1.7355932073797185e-05,
      "loss": 0.5566,
      "step": 8263
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.79886794090271,
      "learning_rate": 1.7355242140432883e-05,
      "loss": 0.4938,
      "step": 8264
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.837095022201538,
      "learning_rate": 1.7354552130782955e-05,
      "loss": 0.5732,
      "step": 8265
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.728576898574829,
      "learning_rate": 1.7353862044854558e-05,
      "loss": 0.5282,
      "step": 8266
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7526947259902954,
      "learning_rate": 1.735317188265486e-05,
      "loss": 0.6081,
      "step": 8267
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6289538145065308,
      "learning_rate": 1.7352481644191004e-05,
      "loss": 0.4596,
      "step": 8268
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.665848731994629,
      "learning_rate": 1.735179132947016e-05,
      "loss": 0.5308,
      "step": 8269
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6904557943344116,
      "learning_rate": 1.7351100938499482e-05,
      "loss": 0.5133,
      "step": 8270
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.647027850151062,
      "learning_rate": 1.7350410471286133e-05,
      "loss": 0.5179,
      "step": 8271
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6564137935638428,
      "learning_rate": 1.7349719927837274e-05,
      "loss": 0.5296,
      "step": 8272
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6017444133758545,
      "learning_rate": 1.7349029308160065e-05,
      "loss": 0.5086,
      "step": 8273
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6779470443725586,
      "learning_rate": 1.7348338612261674e-05,
      "loss": 0.4887,
      "step": 8274
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7490010261535645,
      "learning_rate": 1.7347647840149258e-05,
      "loss": 0.5811,
      "step": 8275
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6002577543258667,
      "learning_rate": 1.7346956991829985e-05,
      "loss": 0.4579,
      "step": 8276
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.678307294845581,
      "learning_rate": 1.734626606731102e-05,
      "loss": 0.5198,
      "step": 8277
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5187556743621826,
      "learning_rate": 1.7345575066599528e-05,
      "loss": 0.4337,
      "step": 8278
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7368714809417725,
      "learning_rate": 1.7344883989702675e-05,
      "loss": 0.5615,
      "step": 8279
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.743177890777588,
      "learning_rate": 1.7344192836627633e-05,
      "loss": 0.4744,
      "step": 8280
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.107320547103882,
      "learning_rate": 1.7343501607381567e-05,
      "loss": 1.0484,
      "step": 8281
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9270976781845093,
      "learning_rate": 1.7342810301971644e-05,
      "loss": 1.0951,
      "step": 8282
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.761896014213562,
      "learning_rate": 1.7342118920405035e-05,
      "loss": 0.5576,
      "step": 8283
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.854097843170166,
      "learning_rate": 1.7341427462688915e-05,
      "loss": 0.4542,
      "step": 8284
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7587075233459473,
      "learning_rate": 1.7340735928830455e-05,
      "loss": 0.6039,
      "step": 8285
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6278117895126343,
      "learning_rate": 1.7340044318836822e-05,
      "loss": 0.5171,
      "step": 8286
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5517576932907104,
      "learning_rate": 1.733935263271519e-05,
      "loss": 0.5066,
      "step": 8287
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.699330449104309,
      "learning_rate": 1.733866087047274e-05,
      "loss": 0.5228,
      "step": 8288
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8680733442306519,
      "learning_rate": 1.733796903211664e-05,
      "loss": 0.5643,
      "step": 8289
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6472491025924683,
      "learning_rate": 1.7337277117654067e-05,
      "loss": 0.4277,
      "step": 8290
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8026487827301025,
      "learning_rate": 1.7336585127092193e-05,
      "loss": 0.5639,
      "step": 8291
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.62098228931427,
      "learning_rate": 1.7335893060438205e-05,
      "loss": 0.4836,
      "step": 8292
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6003079414367676,
      "learning_rate": 1.7335200917699273e-05,
      "loss": 0.5305,
      "step": 8293
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7749279737472534,
      "learning_rate": 1.733450869888258e-05,
      "loss": 0.4953,
      "step": 8294
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7639282941818237,
      "learning_rate": 1.7333816403995302e-05,
      "loss": 0.5767,
      "step": 8295
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6373798847198486,
      "learning_rate": 1.733312403304462e-05,
      "loss": 0.4392,
      "step": 8296
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6527657508850098,
      "learning_rate": 1.7332431586037714e-05,
      "loss": 0.5001,
      "step": 8297
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7803305387496948,
      "learning_rate": 1.733173906298177e-05,
      "loss": 0.5546,
      "step": 8298
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.846482515335083,
      "learning_rate": 1.7331046463883966e-05,
      "loss": 0.5581,
      "step": 8299
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6951302289962769,
      "learning_rate": 1.7330353788751487e-05,
      "loss": 0.4761,
      "step": 8300
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.777646780014038,
      "learning_rate": 1.732966103759152e-05,
      "loss": 0.5375,
      "step": 8301
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6247189044952393,
      "learning_rate": 1.7328968210411246e-05,
      "loss": 0.4584,
      "step": 8302
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.776957631111145,
      "learning_rate": 1.7328275307217854e-05,
      "loss": 0.5717,
      "step": 8303
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5853902101516724,
      "learning_rate": 1.7327582328018524e-05,
      "loss": 0.475,
      "step": 8304
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7758041620254517,
      "learning_rate": 1.732688927282045e-05,
      "loss": 0.5799,
      "step": 8305
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6733068227767944,
      "learning_rate": 1.732619614163082e-05,
      "loss": 0.4708,
      "step": 8306
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6101959943771362,
      "learning_rate": 1.7325502934456818e-05,
      "loss": 0.5198,
      "step": 8307
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.626968264579773,
      "learning_rate": 1.7324809651305638e-05,
      "loss": 0.4835,
      "step": 8308
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6555793285369873,
      "learning_rate": 1.732411629218447e-05,
      "loss": 0.5227,
      "step": 8309
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6267917156219482,
      "learning_rate": 1.73234228571005e-05,
      "loss": 0.4647,
      "step": 8310
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6913076639175415,
      "learning_rate": 1.7322729346060927e-05,
      "loss": 0.5305,
      "step": 8311
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.626847267150879,
      "learning_rate": 1.7322035759072943e-05,
      "loss": 0.4693,
      "step": 8312
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6492503881454468,
      "learning_rate": 1.7321342096143736e-05,
      "loss": 0.5873,
      "step": 8313
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.669453501701355,
      "learning_rate": 1.7320648357280506e-05,
      "loss": 0.4987,
      "step": 8314
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6296374797821045,
      "learning_rate": 1.7319954542490448e-05,
      "loss": 0.5275,
      "step": 8315
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5687780380249023,
      "learning_rate": 1.7319260651780753e-05,
      "loss": 0.4757,
      "step": 8316
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8189160823822021,
      "learning_rate": 1.731856668515862e-05,
      "loss": 0.5824,
      "step": 8317
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6152496337890625,
      "learning_rate": 1.731787264263125e-05,
      "loss": 0.4727,
      "step": 8318
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7074466943740845,
      "learning_rate": 1.731717852420584e-05,
      "loss": 0.5569,
      "step": 8319
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6657207012176514,
      "learning_rate": 1.7316484329889584e-05,
      "loss": 0.5065,
      "step": 8320
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.694345474243164,
      "learning_rate": 1.731579005968969e-05,
      "loss": 0.5518,
      "step": 8321
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.69827401638031,
      "learning_rate": 1.731509571361335e-05,
      "loss": 0.5078,
      "step": 8322
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6792075634002686,
      "learning_rate": 1.7314401291667777e-05,
      "loss": 0.5197,
      "step": 8323
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6078048944473267,
      "learning_rate": 1.7313706793860162e-05,
      "loss": 0.4893,
      "step": 8324
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7817901372909546,
      "learning_rate": 1.731301222019771e-05,
      "loss": 0.5762,
      "step": 8325
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7470498085021973,
      "learning_rate": 1.731231757068763e-05,
      "loss": 0.4868,
      "step": 8326
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7697073221206665,
      "learning_rate": 1.7311622845337122e-05,
      "loss": 0.5423,
      "step": 8327
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.649631142616272,
      "learning_rate": 1.7310928044153395e-05,
      "loss": 0.4871,
      "step": 8328
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.628293514251709,
      "learning_rate": 1.7310233167143652e-05,
      "loss": 0.5996,
      "step": 8329
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7065789699554443,
      "learning_rate": 1.7309538214315102e-05,
      "loss": 0.5199,
      "step": 8330
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6471331119537354,
      "learning_rate": 1.7308843185674952e-05,
      "loss": 0.5193,
      "step": 8331
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6286588907241821,
      "learning_rate": 1.730814808123041e-05,
      "loss": 0.4859,
      "step": 8332
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0140810012817383,
      "learning_rate": 1.730745290098869e-05,
      "loss": 0.5874,
      "step": 8333
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7009189128875732,
      "learning_rate": 1.7306757644956996e-05,
      "loss": 0.4967,
      "step": 8334
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7121961116790771,
      "learning_rate": 1.730606231314254e-05,
      "loss": 0.5409,
      "step": 8335
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.620224118232727,
      "learning_rate": 1.7305366905552535e-05,
      "loss": 0.4693,
      "step": 8336
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.686437726020813,
      "learning_rate": 1.7304671422194195e-05,
      "loss": 0.5339,
      "step": 8337
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6387277841567993,
      "learning_rate": 1.730397586307473e-05,
      "loss": 0.5108,
      "step": 8338
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8189051151275635,
      "learning_rate": 1.7303280228201355e-05,
      "loss": 0.5751,
      "step": 8339
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6983556747436523,
      "learning_rate": 1.730258451758129e-05,
      "loss": 0.4724,
      "step": 8340
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6655919551849365,
      "learning_rate": 1.7301888731221742e-05,
      "loss": 0.5062,
      "step": 8341
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6441609859466553,
      "learning_rate": 1.7301192869129934e-05,
      "loss": 0.4916,
      "step": 8342
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8518351316452026,
      "learning_rate": 1.7300496931313075e-05,
      "loss": 0.5277,
      "step": 8343
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5985592603683472,
      "learning_rate": 1.7299800917778393e-05,
      "loss": 0.4664,
      "step": 8344
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7580649852752686,
      "learning_rate": 1.7299104828533103e-05,
      "loss": 0.625,
      "step": 8345
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6902921199798584,
      "learning_rate": 1.7298408663584424e-05,
      "loss": 0.5292,
      "step": 8346
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.753324270248413,
      "learning_rate": 1.7297712422939575e-05,
      "loss": 0.5404,
      "step": 8347
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5599464178085327,
      "learning_rate": 1.729701610660578e-05,
      "loss": 0.486,
      "step": 8348
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6523524522781372,
      "learning_rate": 1.7296319714590257e-05,
      "loss": 0.527,
      "step": 8349
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6545686721801758,
      "learning_rate": 1.7295623246900235e-05,
      "loss": 0.4698,
      "step": 8350
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6590392589569092,
      "learning_rate": 1.729492670354293e-05,
      "loss": 0.5684,
      "step": 8351
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6314486265182495,
      "learning_rate": 1.729423008452557e-05,
      "loss": 0.4848,
      "step": 8352
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.74275541305542,
      "learning_rate": 1.729353338985538e-05,
      "loss": 0.5568,
      "step": 8353
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6594469547271729,
      "learning_rate": 1.729283661953959e-05,
      "loss": 0.4645,
      "step": 8354
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7743828296661377,
      "learning_rate": 1.729213977358542e-05,
      "loss": 0.5588,
      "step": 8355
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.619725227355957,
      "learning_rate": 1.7291442852000098e-05,
      "loss": 0.4748,
      "step": 8356
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.70827317237854,
      "learning_rate": 1.7290745854790854e-05,
      "loss": 0.5277,
      "step": 8357
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7657454013824463,
      "learning_rate": 1.729004878196492e-05,
      "loss": 0.461,
      "step": 8358
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7847979068756104,
      "learning_rate": 1.728935163352952e-05,
      "loss": 0.5757,
      "step": 8359
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8058902025222778,
      "learning_rate": 1.7288654409491892e-05,
      "loss": 0.4595,
      "step": 8360
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.693352222442627,
      "learning_rate": 1.7287957109859255e-05,
      "loss": 0.5257,
      "step": 8361
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6200482845306396,
      "learning_rate": 1.7287259734638855e-05,
      "loss": 0.4599,
      "step": 8362
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.643895149230957,
      "learning_rate": 1.7286562283837916e-05,
      "loss": 0.5265,
      "step": 8363
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5996639728546143,
      "learning_rate": 1.7285864757463677e-05,
      "loss": 0.4755,
      "step": 8364
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.071352958679199,
      "learning_rate": 1.7285167155523365e-05,
      "loss": 1.0115,
      "step": 8365
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8468128442764282,
      "learning_rate": 1.7284469478024225e-05,
      "loss": 1.0749,
      "step": 8366
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8308274745941162,
      "learning_rate": 1.7283771724973483e-05,
      "loss": 0.6228,
      "step": 8367
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7515023946762085,
      "learning_rate": 1.7283073896378383e-05,
      "loss": 0.4856,
      "step": 8368
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.798211693763733,
      "learning_rate": 1.7282375992246162e-05,
      "loss": 0.5671,
      "step": 8369
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.686414122581482,
      "learning_rate": 1.7281678012584054e-05,
      "loss": 0.4604,
      "step": 8370
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5966993570327759,
      "learning_rate": 1.7280979957399303e-05,
      "loss": 0.5114,
      "step": 8371
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7405306100845337,
      "learning_rate": 1.7280281826699148e-05,
      "loss": 0.483,
      "step": 8372
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8314744234085083,
      "learning_rate": 1.727958362049083e-05,
      "loss": 0.5667,
      "step": 8373
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6927573680877686,
      "learning_rate": 1.7278885338781585e-05,
      "loss": 0.5074,
      "step": 8374
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8907572031021118,
      "learning_rate": 1.727818698157866e-05,
      "loss": 0.5663,
      "step": 8375
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.607118844985962,
      "learning_rate": 1.7277488548889302e-05,
      "loss": 0.4673,
      "step": 8376
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7835510969161987,
      "learning_rate": 1.7276790040720746e-05,
      "loss": 0.5305,
      "step": 8377
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6913968324661255,
      "learning_rate": 1.7276091457080243e-05,
      "loss": 0.4506,
      "step": 8378
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.2237985134124756,
      "learning_rate": 1.7275392797975034e-05,
      "loss": 1.0504,
      "step": 8379
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9408410787582397,
      "learning_rate": 1.727469406341237e-05,
      "loss": 1.1067,
      "step": 8380
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0394439697265625,
      "learning_rate": 1.7273995253399495e-05,
      "loss": 0.5937,
      "step": 8381
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6588714122772217,
      "learning_rate": 1.727329636794366e-05,
      "loss": 0.4673,
      "step": 8382
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.612861156463623,
      "learning_rate": 1.7272597407052108e-05,
      "loss": 0.4871,
      "step": 8383
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.771349310874939,
      "learning_rate": 1.7271898370732092e-05,
      "loss": 0.5222,
      "step": 8384
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7947335243225098,
      "learning_rate": 1.727119925899086e-05,
      "loss": 0.5528,
      "step": 8385
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7574177980422974,
      "learning_rate": 1.727050007183567e-05,
      "loss": 0.4918,
      "step": 8386
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.812210202217102,
      "learning_rate": 1.7269800809273762e-05,
      "loss": 0.6236,
      "step": 8387
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6313199996948242,
      "learning_rate": 1.7269101471312395e-05,
      "loss": 0.5002,
      "step": 8388
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.844090461730957,
      "learning_rate": 1.7268402057958823e-05,
      "loss": 0.5327,
      "step": 8389
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7128291130065918,
      "learning_rate": 1.7267702569220302e-05,
      "loss": 0.4937,
      "step": 8390
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7643555402755737,
      "learning_rate": 1.7267003005104082e-05,
      "loss": 0.5333,
      "step": 8391
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7772659063339233,
      "learning_rate": 1.7266303365617417e-05,
      "loss": 0.506,
      "step": 8392
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.616047739982605,
      "learning_rate": 1.7265603650767574e-05,
      "loss": 0.539,
      "step": 8393
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6848191022872925,
      "learning_rate": 1.7264903860561798e-05,
      "loss": 0.4582,
      "step": 8394
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5814718008041382,
      "learning_rate": 1.726420399500735e-05,
      "loss": 0.4856,
      "step": 8395
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7078666687011719,
      "learning_rate": 1.7263504054111493e-05,
      "loss": 0.4721,
      "step": 8396
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6838988065719604,
      "learning_rate": 1.7262804037881483e-05,
      "loss": 0.5291,
      "step": 8397
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6016197204589844,
      "learning_rate": 1.7262103946324586e-05,
      "loss": 0.4785,
      "step": 8398
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9014006853103638,
      "learning_rate": 1.7261403779448055e-05,
      "loss": 0.5661,
      "step": 8399
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7299976348876953,
      "learning_rate": 1.7260703537259155e-05,
      "loss": 0.4827,
      "step": 8400
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7630035877227783,
      "learning_rate": 1.7260003219765143e-05,
      "loss": 0.4705,
      "step": 8401
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6751289367675781,
      "learning_rate": 1.7259302826973298e-05,
      "loss": 0.5208,
      "step": 8402
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6396129131317139,
      "learning_rate": 1.7258602358890868e-05,
      "loss": 0.5229,
      "step": 8403
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6889704465866089,
      "learning_rate": 1.7257901815525126e-05,
      "loss": 0.5171,
      "step": 8404
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6635360717773438,
      "learning_rate": 1.7257201196883336e-05,
      "loss": 0.5054,
      "step": 8405
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7250800132751465,
      "learning_rate": 1.7256500502972765e-05,
      "loss": 0.5065,
      "step": 8406
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.697496771812439,
      "learning_rate": 1.7255799733800684e-05,
      "loss": 0.5145,
      "step": 8407
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5584383010864258,
      "learning_rate": 1.7255098889374354e-05,
      "loss": 0.4482,
      "step": 8408
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.214959144592285,
      "learning_rate": 1.7254397969701045e-05,
      "loss": 0.55,
      "step": 8409
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.682843804359436,
      "learning_rate": 1.725369697478803e-05,
      "loss": 0.482,
      "step": 8410
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6628544330596924,
      "learning_rate": 1.725299590464258e-05,
      "loss": 0.5714,
      "step": 8411
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5865429639816284,
      "learning_rate": 1.7252294759271963e-05,
      "loss": 0.4455,
      "step": 8412
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6605664491653442,
      "learning_rate": 1.7251593538683454e-05,
      "loss": 0.5389,
      "step": 8413
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5630016326904297,
      "learning_rate": 1.7250892242884322e-05,
      "loss": 0.4974,
      "step": 8414
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8218296766281128,
      "learning_rate": 1.7250190871881844e-05,
      "loss": 0.6596,
      "step": 8415
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5296293497085571,
      "learning_rate": 1.7249489425683294e-05,
      "loss": 0.4677,
      "step": 8416
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7849587202072144,
      "learning_rate": 1.7248787904295943e-05,
      "loss": 0.5502,
      "step": 8417
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.704140543937683,
      "learning_rate": 1.724808630772707e-05,
      "loss": 0.4831,
      "step": 8418
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7525354623794556,
      "learning_rate": 1.724738463598395e-05,
      "loss": 0.561,
      "step": 8419
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5639309883117676,
      "learning_rate": 1.724668288907387e-05,
      "loss": 0.4658,
      "step": 8420
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5963215827941895,
      "learning_rate": 1.7245981067004093e-05,
      "loss": 0.5359,
      "step": 8421
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5870485305786133,
      "learning_rate": 1.724527916978191e-05,
      "loss": 0.4858,
      "step": 8422
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.688790202140808,
      "learning_rate": 1.7244577197414592e-05,
      "loss": 0.5294,
      "step": 8423
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.745383381843567,
      "learning_rate": 1.7243875149909427e-05,
      "loss": 0.4647,
      "step": 8424
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6732640266418457,
      "learning_rate": 1.7243173027273692e-05,
      "loss": 0.5686,
      "step": 8425
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.562954306602478,
      "learning_rate": 1.7242470829514674e-05,
      "loss": 0.4541,
      "step": 8426
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.614028811454773,
      "learning_rate": 1.7241768556639645e-05,
      "loss": 0.5517,
      "step": 8427
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6602075099945068,
      "learning_rate": 1.72410662086559e-05,
      "loss": 0.4656,
      "step": 8428
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7355536222457886,
      "learning_rate": 1.724036378557072e-05,
      "loss": 0.5925,
      "step": 8429
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6385688781738281,
      "learning_rate": 1.7239661287391388e-05,
      "loss": 0.4907,
      "step": 8430
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.815545678138733,
      "learning_rate": 1.7238958714125194e-05,
      "loss": 0.6199,
      "step": 8431
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5820631980895996,
      "learning_rate": 1.723825606577942e-05,
      "loss": 0.4795,
      "step": 8432
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7805944681167603,
      "learning_rate": 1.7237553342361356e-05,
      "loss": 0.5654,
      "step": 8433
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7075214385986328,
      "learning_rate": 1.723685054387829e-05,
      "loss": 0.5071,
      "step": 8434
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6071109771728516,
      "learning_rate": 1.723614767033751e-05,
      "loss": 0.5412,
      "step": 8435
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6404320001602173,
      "learning_rate": 1.7235444721746317e-05,
      "loss": 0.4929,
      "step": 8436
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6537259817123413,
      "learning_rate": 1.723474169811198e-05,
      "loss": 0.5545,
      "step": 8437
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6765062808990479,
      "learning_rate": 1.723403859944181e-05,
      "loss": 0.524,
      "step": 8438
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.829217791557312,
      "learning_rate": 1.723333542574309e-05,
      "loss": 0.5159,
      "step": 8439
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6652584075927734,
      "learning_rate": 1.7232632177023116e-05,
      "loss": 0.5132,
      "step": 8440
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7615104913711548,
      "learning_rate": 1.723192885328918e-05,
      "loss": 0.5781,
      "step": 8441
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6788450479507446,
      "learning_rate": 1.7231225454548576e-05,
      "loss": 0.4781,
      "step": 8442
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7556637525558472,
      "learning_rate": 1.7230521980808602e-05,
      "loss": 0.5318,
      "step": 8443
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7131645679473877,
      "learning_rate": 1.7229818432076553e-05,
      "loss": 0.5729,
      "step": 8444
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1288070678710938,
      "learning_rate": 1.7229114808359722e-05,
      "loss": 1.0174,
      "step": 8445
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.756078839302063,
      "learning_rate": 1.7228411109665416e-05,
      "loss": 1.0662,
      "step": 8446
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9322842359542847,
      "learning_rate": 1.7227707336000925e-05,
      "loss": 0.6272,
      "step": 8447
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8433940410614014,
      "learning_rate": 1.7227003487373553e-05,
      "loss": 0.5286,
      "step": 8448
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8064796924591064,
      "learning_rate": 1.7226299563790597e-05,
      "loss": 0.5611,
      "step": 8449
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.617658019065857,
      "learning_rate": 1.722559556525936e-05,
      "loss": 0.4708,
      "step": 8450
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8120626211166382,
      "learning_rate": 1.722489149178714e-05,
      "loss": 0.5355,
      "step": 8451
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6966253519058228,
      "learning_rate": 1.7224187343381246e-05,
      "loss": 0.4775,
      "step": 8452
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6398212909698486,
      "learning_rate": 1.7223483120048972e-05,
      "loss": 0.5117,
      "step": 8453
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5247077941894531,
      "learning_rate": 1.7222778821797634e-05,
      "loss": 0.4629,
      "step": 8454
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6424468755722046,
      "learning_rate": 1.7222074448634525e-05,
      "loss": 0.4962,
      "step": 8455
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7291172742843628,
      "learning_rate": 1.722137000056696e-05,
      "loss": 0.5114,
      "step": 8456
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7322996854782104,
      "learning_rate": 1.7220665477602235e-05,
      "loss": 0.5314,
      "step": 8457
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6797164678573608,
      "learning_rate": 1.7219960879747662e-05,
      "loss": 0.5299,
      "step": 8458
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6836156845092773,
      "learning_rate": 1.7219256207010552e-05,
      "loss": 0.5694,
      "step": 8459
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.613638162612915,
      "learning_rate": 1.7218551459398213e-05,
      "loss": 0.4669,
      "step": 8460
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.562535285949707,
      "learning_rate": 1.721784663691795e-05,
      "loss": 0.5066,
      "step": 8461
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6908135414123535,
      "learning_rate": 1.7217141739577074e-05,
      "loss": 0.4974,
      "step": 8462
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7324230670928955,
      "learning_rate": 1.72164367673829e-05,
      "loss": 0.5554,
      "step": 8463
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6787185668945312,
      "learning_rate": 1.7215731720342738e-05,
      "loss": 0.4981,
      "step": 8464
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8157283067703247,
      "learning_rate": 1.7215026598463897e-05,
      "loss": 0.6131,
      "step": 8465
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5918066501617432,
      "learning_rate": 1.7214321401753693e-05,
      "loss": 0.4624,
      "step": 8466
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6754645109176636,
      "learning_rate": 1.7213616130219438e-05,
      "loss": 0.4987,
      "step": 8467
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7746412754058838,
      "learning_rate": 1.7212910783868453e-05,
      "loss": 0.4842,
      "step": 8468
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.665277361869812,
      "learning_rate": 1.7212205362708044e-05,
      "loss": 0.5583,
      "step": 8469
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5916749238967896,
      "learning_rate": 1.721149986674554e-05,
      "loss": 0.4684,
      "step": 8470
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.656312346458435,
      "learning_rate": 1.7210794295988245e-05,
      "loss": 0.5159,
      "step": 8471
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.615163803100586,
      "learning_rate": 1.7210088650443482e-05,
      "loss": 0.4651,
      "step": 8472
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7220896482467651,
      "learning_rate": 1.720938293011857e-05,
      "loss": 0.5488,
      "step": 8473
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6253560781478882,
      "learning_rate": 1.7208677135020833e-05,
      "loss": 0.4701,
      "step": 8474
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6413099765777588,
      "learning_rate": 1.7207971265157585e-05,
      "loss": 0.5448,
      "step": 8475
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.694638729095459,
      "learning_rate": 1.7207265320536147e-05,
      "loss": 0.5413,
      "step": 8476
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5785666704177856,
      "learning_rate": 1.7206559301163842e-05,
      "loss": 0.5188,
      "step": 8477
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.680977702140808,
      "learning_rate": 1.7205853207047997e-05,
      "loss": 0.4784,
      "step": 8478
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7301506996154785,
      "learning_rate": 1.7205147038195932e-05,
      "loss": 0.5517,
      "step": 8479
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.682363510131836,
      "learning_rate": 1.7204440794614964e-05,
      "loss": 0.5101,
      "step": 8480
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7072023153305054,
      "learning_rate": 1.720373447631243e-05,
      "loss": 0.5363,
      "step": 8481
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5723906755447388,
      "learning_rate": 1.720302808329565e-05,
      "loss": 0.4511,
      "step": 8482
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6830172538757324,
      "learning_rate": 1.7202321615571945e-05,
      "loss": 0.5994,
      "step": 8483
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6152673959732056,
      "learning_rate": 1.7201615073148653e-05,
      "loss": 0.4721,
      "step": 8484
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6437079906463623,
      "learning_rate": 1.7200908456033098e-05,
      "loss": 0.5113,
      "step": 8485
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6687783002853394,
      "learning_rate": 1.7200201764232605e-05,
      "loss": 0.4893,
      "step": 8486
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.653114676475525,
      "learning_rate": 1.7199494997754505e-05,
      "loss": 0.5638,
      "step": 8487
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.685308575630188,
      "learning_rate": 1.719878815660613e-05,
      "loss": 0.5093,
      "step": 8488
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6680939197540283,
      "learning_rate": 1.719808124079481e-05,
      "loss": 0.5828,
      "step": 8489
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.574530005455017,
      "learning_rate": 1.719737425032788e-05,
      "loss": 0.4688,
      "step": 8490
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.72073495388031,
      "learning_rate": 1.719666718521267e-05,
      "loss": 0.5424,
      "step": 8491
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6037659645080566,
      "learning_rate": 1.7195960045456508e-05,
      "loss": 0.4612,
      "step": 8492
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7137644290924072,
      "learning_rate": 1.7195252831066735e-05,
      "loss": 0.6205,
      "step": 8493
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5928016901016235,
      "learning_rate": 1.7194545542050687e-05,
      "loss": 0.4465,
      "step": 8494
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5781816244125366,
      "learning_rate": 1.7193838178415695e-05,
      "loss": 0.504,
      "step": 8495
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6118953227996826,
      "learning_rate": 1.7193130740169097e-05,
      "loss": 0.5136,
      "step": 8496
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7334117889404297,
      "learning_rate": 1.7192423227318237e-05,
      "loss": 0.5146,
      "step": 8497
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6841094493865967,
      "learning_rate": 1.7191715639870438e-05,
      "loss": 0.4729,
      "step": 8498
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.667651653289795,
      "learning_rate": 1.7191007977833053e-05,
      "loss": 0.506,
      "step": 8499
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6545253992080688,
      "learning_rate": 1.7190300241213415e-05,
      "loss": 0.4417,
      "step": 8500
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8302335739135742,
      "learning_rate": 1.7189592430018868e-05,
      "loss": 0.503,
      "step": 8501
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6206145286560059,
      "learning_rate": 1.7188884544256748e-05,
      "loss": 0.4831,
      "step": 8502
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7976793050765991,
      "learning_rate": 1.71881765839344e-05,
      "loss": 0.509,
      "step": 8503
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.619892954826355,
      "learning_rate": 1.7187468549059164e-05,
      "loss": 0.462,
      "step": 8504
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7891249656677246,
      "learning_rate": 1.7186760439638387e-05,
      "loss": 0.5662,
      "step": 8505
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6049882173538208,
      "learning_rate": 1.7186052255679416e-05,
      "loss": 0.4618,
      "step": 8506
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7669275999069214,
      "learning_rate": 1.718534399718959e-05,
      "loss": 0.5843,
      "step": 8507
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.611510157585144,
      "learning_rate": 1.718463566417625e-05,
      "loss": 0.4776,
      "step": 8508
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8112200498580933,
      "learning_rate": 1.718392725664676e-05,
      "loss": 0.5616,
      "step": 8509
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6171363592147827,
      "learning_rate": 1.7183218774608453e-05,
      "loss": 0.4651,
      "step": 8510
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7728073596954346,
      "learning_rate": 1.718251021806868e-05,
      "loss": 0.5758,
      "step": 8511
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5530112981796265,
      "learning_rate": 1.7181801587034786e-05,
      "loss": 0.4537,
      "step": 8512
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8167165517807007,
      "learning_rate": 1.7181092881514133e-05,
      "loss": 0.592,
      "step": 8513
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.687648892402649,
      "learning_rate": 1.718038410151406e-05,
      "loss": 0.4649,
      "step": 8514
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7288569211959839,
      "learning_rate": 1.7179675247041922e-05,
      "loss": 0.5806,
      "step": 8515
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6713966131210327,
      "learning_rate": 1.7178966318105073e-05,
      "loss": 0.4903,
      "step": 8516
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0515198707580566,
      "learning_rate": 1.717825731471086e-05,
      "loss": 1.0575,
      "step": 8517
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8015389442443848,
      "learning_rate": 1.7177548236866642e-05,
      "loss": 1.0077,
      "step": 8518
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8493329286575317,
      "learning_rate": 1.717683908457977e-05,
      "loss": 0.5241,
      "step": 8519
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7736265659332275,
      "learning_rate": 1.71761298578576e-05,
      "loss": 0.5583,
      "step": 8520
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6689289808273315,
      "learning_rate": 1.7175420556707488e-05,
      "loss": 0.5174,
      "step": 8521
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.692133903503418,
      "learning_rate": 1.7174711181136797e-05,
      "loss": 0.473,
      "step": 8522
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8007500171661377,
      "learning_rate": 1.717400173115287e-05,
      "loss": 0.5576,
      "step": 8523
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.62334406375885,
      "learning_rate": 1.717329220676308e-05,
      "loss": 0.4785,
      "step": 8524
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6701314449310303,
      "learning_rate": 1.7172582607974774e-05,
      "loss": 0.5827,
      "step": 8525
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6451494693756104,
      "learning_rate": 1.717187293479532e-05,
      "loss": 0.4964,
      "step": 8526
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7196924686431885,
      "learning_rate": 1.717116318723207e-05,
      "loss": 0.5747,
      "step": 8527
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7541630268096924,
      "learning_rate": 1.7170453365292397e-05,
      "loss": 0.4899,
      "step": 8528
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.811967134475708,
      "learning_rate": 1.7169743468983654e-05,
      "loss": 0.5722,
      "step": 8529
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6585171222686768,
      "learning_rate": 1.7169033498313204e-05,
      "loss": 0.4855,
      "step": 8530
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7490259408950806,
      "learning_rate": 1.7168323453288417e-05,
      "loss": 0.5342,
      "step": 8531
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7304965257644653,
      "learning_rate": 1.716761333391665e-05,
      "loss": 0.4891,
      "step": 8532
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7153620719909668,
      "learning_rate": 1.7166903140205272e-05,
      "loss": 0.549,
      "step": 8533
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6535210609436035,
      "learning_rate": 1.716619287216165e-05,
      "loss": 0.4695,
      "step": 8534
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8484690189361572,
      "learning_rate": 1.7165482529793146e-05,
      "loss": 0.599,
      "step": 8535
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7403359413146973,
      "learning_rate": 1.7164772113107127e-05,
      "loss": 0.4962,
      "step": 8536
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.765153408050537,
      "learning_rate": 1.716406162211097e-05,
      "loss": 0.5752,
      "step": 8537
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.658618688583374,
      "learning_rate": 1.7163351056812033e-05,
      "loss": 0.4992,
      "step": 8538
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.7876710891723633,
      "learning_rate": 1.7162640417217696e-05,
      "loss": 0.6127,
      "step": 8539
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6346274614334106,
      "learning_rate": 1.716192970333532e-05,
      "loss": 0.4788,
      "step": 8540
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.061148166656494,
      "learning_rate": 1.7161218915172282e-05,
      "loss": 0.5585,
      "step": 8541
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5911056995391846,
      "learning_rate": 1.716050805273595e-05,
      "loss": 0.4758,
      "step": 8542
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6132028102874756,
      "learning_rate": 1.7159797116033705e-05,
      "loss": 0.494,
      "step": 8543
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.722898244857788,
      "learning_rate": 1.715908610507291e-05,
      "loss": 0.4314,
      "step": 8544
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6867135763168335,
      "learning_rate": 1.7158375019860947e-05,
      "loss": 0.5728,
      "step": 8545
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5723649263381958,
      "learning_rate": 1.7157663860405187e-05,
      "loss": 0.4481,
      "step": 8546
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6966655254364014,
      "learning_rate": 1.7156952626713007e-05,
      "loss": 0.5582,
      "step": 8547
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6859980821609497,
      "learning_rate": 1.7156241318791786e-05,
      "loss": 0.4771,
      "step": 8548
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7269283533096313,
      "learning_rate": 1.7155529936648892e-05,
      "loss": 0.5336,
      "step": 8549
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6587594747543335,
      "learning_rate": 1.7154818480291716e-05,
      "loss": 0.45,
      "step": 8550
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.759791374206543,
      "learning_rate": 1.715410694972763e-05,
      "loss": 0.4936,
      "step": 8551
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6849762201309204,
      "learning_rate": 1.7153395344964018e-05,
      "loss": 0.5059,
      "step": 8552
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6668232679367065,
      "learning_rate": 1.7152683666008253e-05,
      "loss": 0.494,
      "step": 8553
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7630962133407593,
      "learning_rate": 1.715197191286772e-05,
      "loss": 0.5411,
      "step": 8554
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7347480058670044,
      "learning_rate": 1.7151260085549806e-05,
      "loss": 0.5793,
      "step": 8555
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7092117071151733,
      "learning_rate": 1.715054818406189e-05,
      "loss": 0.492,
      "step": 8556
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6469885110855103,
      "learning_rate": 1.7149836208411354e-05,
      "loss": 0.4995,
      "step": 8557
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6657969951629639,
      "learning_rate": 1.7149124158605578e-05,
      "loss": 0.5062,
      "step": 8558
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.750928282737732,
      "learning_rate": 1.7148412034651958e-05,
      "loss": 0.5328,
      "step": 8559
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6931391954421997,
      "learning_rate": 1.7147699836557873e-05,
      "loss": 0.4796,
      "step": 8560
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7620443105697632,
      "learning_rate": 1.7146987564330712e-05,
      "loss": 0.5264,
      "step": 8561
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6140528917312622,
      "learning_rate": 1.7146275217977858e-05,
      "loss": 0.444,
      "step": 8562
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8233132362365723,
      "learning_rate": 1.7145562797506704e-05,
      "loss": 0.545,
      "step": 8563
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.797268271446228,
      "learning_rate": 1.714485030292464e-05,
      "loss": 0.4818,
      "step": 8564
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7395292520523071,
      "learning_rate": 1.714413773423905e-05,
      "loss": 0.5497,
      "step": 8565
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7076270580291748,
      "learning_rate": 1.714342509145733e-05,
      "loss": 0.4903,
      "step": 8566
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7016041278839111,
      "learning_rate": 1.7142712374586867e-05,
      "loss": 0.5421,
      "step": 8567
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6895347833633423,
      "learning_rate": 1.7141999583635056e-05,
      "loss": 0.4898,
      "step": 8568
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5806618928909302,
      "learning_rate": 1.7141286718609288e-05,
      "loss": 0.5022,
      "step": 8569
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7032465934753418,
      "learning_rate": 1.7140573779516957e-05,
      "loss": 0.4775,
      "step": 8570
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7616865634918213,
      "learning_rate": 1.713986076636546e-05,
      "loss": 0.5799,
      "step": 8571
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6739091873168945,
      "learning_rate": 1.7139147679162186e-05,
      "loss": 0.4653,
      "step": 8572
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6922513246536255,
      "learning_rate": 1.7138434517914536e-05,
      "loss": 0.523,
      "step": 8573
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6945436000823975,
      "learning_rate": 1.7137721282629905e-05,
      "loss": 0.5035,
      "step": 8574
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.644856572151184,
      "learning_rate": 1.713700797331569e-05,
      "loss": 0.523,
      "step": 8575
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6448974609375,
      "learning_rate": 1.713629458997929e-05,
      "loss": 0.4742,
      "step": 8576
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6544523239135742,
      "learning_rate": 1.7135581132628104e-05,
      "loss": 0.5116,
      "step": 8577
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.589918613433838,
      "learning_rate": 1.713486760126953e-05,
      "loss": 0.4863,
      "step": 8578
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.755575180053711,
      "learning_rate": 1.7134153995910972e-05,
      "loss": 0.5499,
      "step": 8579
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5994986295700073,
      "learning_rate": 1.7133440316559828e-05,
      "loss": 0.4992,
      "step": 8580
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6840920448303223,
      "learning_rate": 1.7132726563223503e-05,
      "loss": 0.4932,
      "step": 8581
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7110867500305176,
      "learning_rate": 1.7132012735909395e-05,
      "loss": 0.4933,
      "step": 8582
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.767842173576355,
      "learning_rate": 1.713129883462491e-05,
      "loss": 0.5538,
      "step": 8583
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6909760236740112,
      "learning_rate": 1.7130584859377454e-05,
      "loss": 0.5235,
      "step": 8584
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6780136823654175,
      "learning_rate": 1.712987081017443e-05,
      "loss": 0.5469,
      "step": 8585
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7189034223556519,
      "learning_rate": 1.7129156687023245e-05,
      "loss": 0.5752,
      "step": 8586
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5792165994644165,
      "learning_rate": 1.7128442489931302e-05,
      "loss": 0.501,
      "step": 8587
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6507971286773682,
      "learning_rate": 1.7127728218906018e-05,
      "loss": 0.4711,
      "step": 8588
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7028250694274902,
      "learning_rate": 1.712701387395479e-05,
      "loss": 0.5373,
      "step": 8589
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5719457864761353,
      "learning_rate": 1.7126299455085032e-05,
      "loss": 0.4569,
      "step": 8590
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9238390922546387,
      "learning_rate": 1.712558496230415e-05,
      "loss": 0.6619,
      "step": 8591
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6704050302505493,
      "learning_rate": 1.7124870395619563e-05,
      "loss": 0.4673,
      "step": 8592
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8804199695587158,
      "learning_rate": 1.7124155755038676e-05,
      "loss": 0.5631,
      "step": 8593
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6029503345489502,
      "learning_rate": 1.71234410405689e-05,
      "loss": 0.4426,
      "step": 8594
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7437864542007446,
      "learning_rate": 1.712272625221765e-05,
      "loss": 0.5834,
      "step": 8595
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6257773637771606,
      "learning_rate": 1.712201138999234e-05,
      "loss": 0.4903,
      "step": 8596
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.943086624145508,
      "learning_rate": 1.712129645390038e-05,
      "loss": 0.6555,
      "step": 8597
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6697033643722534,
      "learning_rate": 1.712058144394919e-05,
      "loss": 0.4681,
      "step": 8598
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.791210412979126,
      "learning_rate": 1.7119866360146184e-05,
      "loss": 0.5811,
      "step": 8599
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7072811126708984,
      "learning_rate": 1.7119151202498777e-05,
      "loss": 0.5113,
      "step": 8600
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6015610694885254,
      "learning_rate": 1.711843597101439e-05,
      "loss": 0.5219,
      "step": 8601
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6332122087478638,
      "learning_rate": 1.711772066570044e-05,
      "loss": 0.4762,
      "step": 8602
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7038415670394897,
      "learning_rate": 1.7117005286564344e-05,
      "loss": 0.5528,
      "step": 8603
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0154895782470703,
      "learning_rate": 1.7116289833613524e-05,
      "loss": 0.5332,
      "step": 8604
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6514298915863037,
      "learning_rate": 1.7115574306855393e-05,
      "loss": 0.5647,
      "step": 8605
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.572265863418579,
      "learning_rate": 1.7114858706297382e-05,
      "loss": 0.4719,
      "step": 8606
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7111939191818237,
      "learning_rate": 1.7114143031946912e-05,
      "loss": 0.555,
      "step": 8607
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.606758713722229,
      "learning_rate": 1.7113427283811398e-05,
      "loss": 0.4741,
      "step": 8608
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7114481925964355,
      "learning_rate": 1.711271146189827e-05,
      "loss": 0.5351,
      "step": 8609
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7335323095321655,
      "learning_rate": 1.711199556621495e-05,
      "loss": 0.4623,
      "step": 8610
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7201261520385742,
      "learning_rate": 1.7111279596768863e-05,
      "loss": 0.547,
      "step": 8611
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6232614517211914,
      "learning_rate": 1.7110563553567433e-05,
      "loss": 0.4545,
      "step": 8612
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.794149398803711,
      "learning_rate": 1.7109847436618093e-05,
      "loss": 0.6254,
      "step": 8613
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7094742059707642,
      "learning_rate": 1.7109131245928264e-05,
      "loss": 0.5145,
      "step": 8614
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7357255220413208,
      "learning_rate": 1.7108414981505376e-05,
      "loss": 0.567,
      "step": 8615
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.564560055732727,
      "learning_rate": 1.710769864335686e-05,
      "loss": 0.4533,
      "step": 8616
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7741494178771973,
      "learning_rate": 1.710698223149014e-05,
      "loss": 0.5458,
      "step": 8617
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6199017763137817,
      "learning_rate": 1.710626574591265e-05,
      "loss": 0.4575,
      "step": 8618
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7834471464157104,
      "learning_rate": 1.7105549186631822e-05,
      "loss": 0.5393,
      "step": 8619
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5973587036132812,
      "learning_rate": 1.7104832553655083e-05,
      "loss": 0.4597,
      "step": 8620
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7229573726654053,
      "learning_rate": 1.7104115846989873e-05,
      "loss": 0.596,
      "step": 8621
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.670881748199463,
      "learning_rate": 1.710339906664362e-05,
      "loss": 0.4785,
      "step": 8622
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5970052480697632,
      "learning_rate": 1.710268221262376e-05,
      "loss": 0.5131,
      "step": 8623
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.572831153869629,
      "learning_rate": 1.710196528493773e-05,
      "loss": 0.4677,
      "step": 8624
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6607425212860107,
      "learning_rate": 1.7101248283592963e-05,
      "loss": 0.5238,
      "step": 8625
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.928007960319519,
      "learning_rate": 1.7100531208596895e-05,
      "loss": 0.5288,
      "step": 8626
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6252028942108154,
      "learning_rate": 1.709981405995696e-05,
      "loss": 0.5363,
      "step": 8627
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7056524753570557,
      "learning_rate": 1.7099096837680605e-05,
      "loss": 0.4816,
      "step": 8628
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6418812274932861,
      "learning_rate": 1.7098379541775265e-05,
      "loss": 0.4954,
      "step": 8629
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.67074716091156,
      "learning_rate": 1.709766217224837e-05,
      "loss": 0.536,
      "step": 8630
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7850561141967773,
      "learning_rate": 1.7096944729107376e-05,
      "loss": 0.5218,
      "step": 8631
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6553510427474976,
      "learning_rate": 1.7096227212359716e-05,
      "loss": 0.4891,
      "step": 8632
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7336450815200806,
      "learning_rate": 1.7095509622012828e-05,
      "loss": 0.5538,
      "step": 8633
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5945755243301392,
      "learning_rate": 1.7094791958074162e-05,
      "loss": 0.4833,
      "step": 8634
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7147685289382935,
      "learning_rate": 1.709407422055116e-05,
      "loss": 0.5832,
      "step": 8635
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5494468212127686,
      "learning_rate": 1.7093356409451262e-05,
      "loss": 0.4499,
      "step": 8636
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6543684005737305,
      "learning_rate": 1.7092638524781916e-05,
      "loss": 0.5443,
      "step": 8637
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5812753438949585,
      "learning_rate": 1.7091920566550566e-05,
      "loss": 0.4861,
      "step": 8638
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.62125825881958,
      "learning_rate": 1.7091202534764658e-05,
      "loss": 0.4813,
      "step": 8639
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6526209115982056,
      "learning_rate": 1.709048442943164e-05,
      "loss": 0.5415,
      "step": 8640
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6650300025939941,
      "learning_rate": 1.7089766250558963e-05,
      "loss": 0.5186,
      "step": 8641
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5553040504455566,
      "learning_rate": 1.7089047998154074e-05,
      "loss": 0.4765,
      "step": 8642
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7492669820785522,
      "learning_rate": 1.708832967222442e-05,
      "loss": 0.6084,
      "step": 8643
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6103569269180298,
      "learning_rate": 1.708761127277745e-05,
      "loss": 0.4596,
      "step": 8644
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7146904468536377,
      "learning_rate": 1.7086892799820617e-05,
      "loss": 0.5472,
      "step": 8645
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6349565982818604,
      "learning_rate": 1.708617425336138e-05,
      "loss": 0.4853,
      "step": 8646
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7395604848861694,
      "learning_rate": 1.7085455633407177e-05,
      "loss": 0.5751,
      "step": 8647
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6275147199630737,
      "learning_rate": 1.7084736939965473e-05,
      "loss": 0.453,
      "step": 8648
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6737418174743652,
      "learning_rate": 1.708401817304372e-05,
      "loss": 0.5478,
      "step": 8649
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.683871865272522,
      "learning_rate": 1.7083299332649364e-05,
      "loss": 0.5024,
      "step": 8650
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5697989463806152,
      "learning_rate": 1.708258041878987e-05,
      "loss": 0.5407,
      "step": 8651
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6714928150177002,
      "learning_rate": 1.708186143147269e-05,
      "loss": 0.4743,
      "step": 8652
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.636465311050415,
      "learning_rate": 1.7081142370705287e-05,
      "loss": 0.5328,
      "step": 8653
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.587062954902649,
      "learning_rate": 1.7080423236495112e-05,
      "loss": 0.4771,
      "step": 8654
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.698089599609375,
      "learning_rate": 1.7079704028849623e-05,
      "loss": 0.582,
      "step": 8655
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6960170269012451,
      "learning_rate": 1.7078984747776285e-05,
      "loss": 0.4919,
      "step": 8656
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6046934127807617,
      "learning_rate": 1.7078265393282557e-05,
      "loss": 0.5327,
      "step": 8657
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5849107503890991,
      "learning_rate": 1.7077545965375895e-05,
      "loss": 0.4944,
      "step": 8658
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7625459432601929,
      "learning_rate": 1.7076826464063765e-05,
      "loss": 0.5679,
      "step": 8659
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5701156854629517,
      "learning_rate": 1.7076106889353628e-05,
      "loss": 0.4544,
      "step": 8660
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6835898160934448,
      "learning_rate": 1.7075387241252944e-05,
      "loss": 0.5662,
      "step": 8661
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5428396463394165,
      "learning_rate": 1.707466751976919e-05,
      "loss": 0.4651,
      "step": 8662
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.712664246559143,
      "learning_rate": 1.7073947724909808e-05,
      "loss": 0.6093,
      "step": 8663
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.602544903755188,
      "learning_rate": 1.7073227856682287e-05,
      "loss": 0.4395,
      "step": 8664
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.732125163078308,
      "learning_rate": 1.7072507915094076e-05,
      "loss": 0.5956,
      "step": 8665
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7256969213485718,
      "learning_rate": 1.7071787900152653e-05,
      "loss": 0.498,
      "step": 8666
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6238616704940796,
      "learning_rate": 1.7071067811865477e-05,
      "loss": 0.5115,
      "step": 8667
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7179906368255615,
      "learning_rate": 1.707034765024002e-05,
      "loss": 0.5324,
      "step": 8668
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6436347961425781,
      "learning_rate": 1.7069627415283754e-05,
      "loss": 0.527,
      "step": 8669
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.626947045326233,
      "learning_rate": 1.706890710700415e-05,
      "loss": 0.4745,
      "step": 8670
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7102413177490234,
      "learning_rate": 1.706818672540867e-05,
      "loss": 0.6213,
      "step": 8671
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.633522868156433,
      "learning_rate": 1.7067466270504796e-05,
      "loss": 0.4709,
      "step": 8672
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6220940351486206,
      "learning_rate": 1.7066745742299993e-05,
      "loss": 0.503,
      "step": 8673
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6839553117752075,
      "learning_rate": 1.7066025140801735e-05,
      "loss": 0.5262,
      "step": 8674
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.81256103515625,
      "learning_rate": 1.70653044660175e-05,
      "loss": 0.5362,
      "step": 8675
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6450811624526978,
      "learning_rate": 1.7064583717954755e-05,
      "loss": 0.482,
      "step": 8676
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6737501621246338,
      "learning_rate": 1.7063862896620984e-05,
      "loss": 0.5485,
      "step": 8677
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6251864433288574,
      "learning_rate": 1.7063142002023657e-05,
      "loss": 0.4779,
      "step": 8678
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0406157970428467,
      "learning_rate": 1.7062421034170253e-05,
      "loss": 1.0928,
      "step": 8679
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6044474840164185,
      "learning_rate": 1.706169999306825e-05,
      "loss": 1.016,
      "step": 8680
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6135057210922241,
      "learning_rate": 1.7060978878725127e-05,
      "loss": 0.5352,
      "step": 8681
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6489694118499756,
      "learning_rate": 1.7060257691148364e-05,
      "loss": 0.5009,
      "step": 8682
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7107717990875244,
      "learning_rate": 1.7059536430345433e-05,
      "loss": 0.5576,
      "step": 8683
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.629634976387024,
      "learning_rate": 1.7058815096323826e-05,
      "loss": 0.5127,
      "step": 8684
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7287940979003906,
      "learning_rate": 1.7058093689091016e-05,
      "loss": 0.5501,
      "step": 8685
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5391433238983154,
      "learning_rate": 1.705737220865449e-05,
      "loss": 0.4368,
      "step": 8686
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.752541422843933,
      "learning_rate": 1.7056650655021724e-05,
      "loss": 0.5528,
      "step": 8687
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.688584566116333,
      "learning_rate": 1.705592902820021e-05,
      "loss": 0.4694,
      "step": 8688
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.602267861366272,
      "learning_rate": 1.7055207328197428e-05,
      "loss": 0.5509,
      "step": 8689
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7030835151672363,
      "learning_rate": 1.7054485555020866e-05,
      "loss": 0.4629,
      "step": 8690
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.675426721572876,
      "learning_rate": 1.7053763708678008e-05,
      "loss": 0.5075,
      "step": 8691
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6636481285095215,
      "learning_rate": 1.7053041789176338e-05,
      "loss": 0.4876,
      "step": 8692
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6993408203125,
      "learning_rate": 1.7052319796523347e-05,
      "loss": 0.558,
      "step": 8693
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6429864168167114,
      "learning_rate": 1.7051597730726524e-05,
      "loss": 0.4939,
      "step": 8694
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8633549213409424,
      "learning_rate": 1.7050875591793356e-05,
      "loss": 0.5994,
      "step": 8695
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7082020044326782,
      "learning_rate": 1.7050153379731334e-05,
      "loss": 0.4475,
      "step": 8696
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7115116119384766,
      "learning_rate": 1.7049431094547947e-05,
      "loss": 0.5961,
      "step": 8697
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6863837242126465,
      "learning_rate": 1.7048708736250684e-05,
      "loss": 0.4838,
      "step": 8698
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6823594570159912,
      "learning_rate": 1.7047986304847044e-05,
      "loss": 0.5439,
      "step": 8699
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.560344934463501,
      "learning_rate": 1.7047263800344516e-05,
      "loss": 0.4764,
      "step": 8700
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7565146684646606,
      "learning_rate": 1.7046541222750592e-05,
      "loss": 0.5733,
      "step": 8701
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7239680290222168,
      "learning_rate": 1.7045818572072766e-05,
      "loss": 0.479,
      "step": 8702
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6336086988449097,
      "learning_rate": 1.7045095848318537e-05,
      "loss": 0.5123,
      "step": 8703
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6748502254486084,
      "learning_rate": 1.7044373051495397e-05,
      "loss": 0.4967,
      "step": 8704
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6534950733184814,
      "learning_rate": 1.7043650181610844e-05,
      "loss": 0.5199,
      "step": 8705
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6561754941940308,
      "learning_rate": 1.7042927238672376e-05,
      "loss": 0.495,
      "step": 8706
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8134334087371826,
      "learning_rate": 1.7042204222687488e-05,
      "loss": 0.5746,
      "step": 8707
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6464667320251465,
      "learning_rate": 1.7041481133663684e-05,
      "loss": 0.4597,
      "step": 8708
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6339747905731201,
      "learning_rate": 1.7040757971608458e-05,
      "loss": 0.4783,
      "step": 8709
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.659520149230957,
      "learning_rate": 1.7040034736529313e-05,
      "loss": 0.4845,
      "step": 8710
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6336898803710938,
      "learning_rate": 1.7039311428433752e-05,
      "loss": 0.5414,
      "step": 8711
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7551109790802002,
      "learning_rate": 1.7038588047329277e-05,
      "loss": 0.4989,
      "step": 8712
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.916674256324768,
      "learning_rate": 1.7037864593223387e-05,
      "loss": 0.5468,
      "step": 8713
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6296859979629517,
      "learning_rate": 1.7037141066123585e-05,
      "loss": 0.441,
      "step": 8714
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7343021631240845,
      "learning_rate": 1.703641746603738e-05,
      "loss": 0.601,
      "step": 8715
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.580452799797058,
      "learning_rate": 1.7035693792972273e-05,
      "loss": 0.484,
      "step": 8716
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5907567739486694,
      "learning_rate": 1.7034970046935773e-05,
      "loss": 0.5392,
      "step": 8717
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.698801875114441,
      "learning_rate": 1.703424622793538e-05,
      "loss": 0.5251,
      "step": 8718
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5430353879928589,
      "learning_rate": 1.7033522335978606e-05,
      "loss": 0.5073,
      "step": 8719
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6084028482437134,
      "learning_rate": 1.703279837107296e-05,
      "loss": 0.481,
      "step": 8720
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6766175031661987,
      "learning_rate": 1.7032074333225948e-05,
      "loss": 0.5525,
      "step": 8721
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6398813724517822,
      "learning_rate": 1.703135022244508e-05,
      "loss": 0.499,
      "step": 8722
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5781084299087524,
      "learning_rate": 1.7030626038737866e-05,
      "loss": 0.5814,
      "step": 8723
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5747920274734497,
      "learning_rate": 1.7029901782111816e-05,
      "loss": 0.4775,
      "step": 8724
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6553233861923218,
      "learning_rate": 1.702917745257445e-05,
      "loss": 0.5419,
      "step": 8725
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6779128313064575,
      "learning_rate": 1.7028453050133267e-05,
      "loss": 0.5477,
      "step": 8726
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.683694839477539,
      "learning_rate": 1.702772857479579e-05,
      "loss": 0.5886,
      "step": 8727
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5718200206756592,
      "learning_rate": 1.702700402656953e-05,
      "loss": 0.4712,
      "step": 8728
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.611240029335022,
      "learning_rate": 1.7026279405462e-05,
      "loss": 0.4917,
      "step": 8729
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6695892810821533,
      "learning_rate": 1.7025554711480716e-05,
      "loss": 0.5329,
      "step": 8730
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.632810354232788,
      "learning_rate": 1.7024829944633195e-05,
      "loss": 0.521,
      "step": 8731
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6704362630844116,
      "learning_rate": 1.7024105104926957e-05,
      "loss": 0.4761,
      "step": 8732
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5986732244491577,
      "learning_rate": 1.7023380192369513e-05,
      "loss": 0.5445,
      "step": 8733
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.141209363937378,
      "learning_rate": 1.7022655206968387e-05,
      "loss": 0.465,
      "step": 8734
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8175934553146362,
      "learning_rate": 1.7021930148731104e-05,
      "loss": 0.6094,
      "step": 8735
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5783013105392456,
      "learning_rate": 1.7021205017665166e-05,
      "loss": 0.4521,
      "step": 8736
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7746193408966064,
      "learning_rate": 1.7020479813778112e-05,
      "loss": 0.5543,
      "step": 8737
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6655242443084717,
      "learning_rate": 1.7019754537077453e-05,
      "loss": 0.4554,
      "step": 8738
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.622936725616455,
      "learning_rate": 1.7019029187570713e-05,
      "loss": 0.5479,
      "step": 8739
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7532267570495605,
      "learning_rate": 1.7018303765265415e-05,
      "loss": 0.5108,
      "step": 8740
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6509591341018677,
      "learning_rate": 1.7017578270169088e-05,
      "loss": 0.5505,
      "step": 8741
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6597483158111572,
      "learning_rate": 1.7016852702289252e-05,
      "loss": 0.4949,
      "step": 8742
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7826554775238037,
      "learning_rate": 1.7016127061633433e-05,
      "loss": 0.5351,
      "step": 8743
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6555308103561401,
      "learning_rate": 1.7015401348209155e-05,
      "loss": 0.4658,
      "step": 8744
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6133276224136353,
      "learning_rate": 1.701467556202395e-05,
      "loss": 0.5905,
      "step": 8745
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6638243198394775,
      "learning_rate": 1.7013949703085336e-05,
      "loss": 0.542,
      "step": 8746
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.598075032234192,
      "learning_rate": 1.7013223771400855e-05,
      "loss": 0.5282,
      "step": 8747
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6505733728408813,
      "learning_rate": 1.7012497766978026e-05,
      "loss": 0.4413,
      "step": 8748
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6602054834365845,
      "learning_rate": 1.701177168982438e-05,
      "loss": 0.5176,
      "step": 8749
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7273683547973633,
      "learning_rate": 1.7011045539947454e-05,
      "loss": 0.5163,
      "step": 8750
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.860224723815918,
      "learning_rate": 1.701031931735477e-05,
      "loss": 0.5651,
      "step": 8751
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6526585817337036,
      "learning_rate": 1.7009593022053863e-05,
      "loss": 0.4898,
      "step": 8752
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7783374786376953,
      "learning_rate": 1.700886665405227e-05,
      "loss": 0.5371,
      "step": 8753
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.720097541809082,
      "learning_rate": 1.7008140213357523e-05,
      "loss": 0.462,
      "step": 8754
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6848872900009155,
      "learning_rate": 1.7007413699977157e-05,
      "loss": 0.5269,
      "step": 8755
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6465469598770142,
      "learning_rate": 1.7006687113918703e-05,
      "loss": 0.4708,
      "step": 8756
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8654404878616333,
      "learning_rate": 1.7005960455189702e-05,
      "loss": 0.5995,
      "step": 8757
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6324877738952637,
      "learning_rate": 1.700523372379768e-05,
      "loss": 0.4341,
      "step": 8758
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7617862224578857,
      "learning_rate": 1.7004506919750193e-05,
      "loss": 0.5905,
      "step": 8759
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.505319595336914,
      "learning_rate": 1.700378004305476e-05,
      "loss": 0.4477,
      "step": 8760
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6605877876281738,
      "learning_rate": 1.7003053093718932e-05,
      "loss": 0.5761,
      "step": 8761
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.667335033416748,
      "learning_rate": 1.7002326071750246e-05,
      "loss": 0.4846,
      "step": 8762
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.702286720275879,
      "learning_rate": 1.700159897715624e-05,
      "loss": 0.5733,
      "step": 8763
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7649625539779663,
      "learning_rate": 1.7000871809944457e-05,
      "loss": 0.5439,
      "step": 8764
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6442294120788574,
      "learning_rate": 1.7000144570122438e-05,
      "loss": 0.5188,
      "step": 8765
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6281484365463257,
      "learning_rate": 1.6999417257697724e-05,
      "loss": 0.4823,
      "step": 8766
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7287604808807373,
      "learning_rate": 1.6998689872677866e-05,
      "loss": 0.5538,
      "step": 8767
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6065517663955688,
      "learning_rate": 1.6997962415070398e-05,
      "loss": 0.4542,
      "step": 8768
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.583383560180664,
      "learning_rate": 1.699723488488287e-05,
      "loss": 0.4956,
      "step": 8769
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6136353015899658,
      "learning_rate": 1.6996507282122827e-05,
      "loss": 0.4425,
      "step": 8770
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.630570888519287,
      "learning_rate": 1.699577960679782e-05,
      "loss": 0.5451,
      "step": 8771
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5952049493789673,
      "learning_rate": 1.6995051858915385e-05,
      "loss": 0.4768,
      "step": 8772
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6822450160980225,
      "learning_rate": 1.6994324038483075e-05,
      "loss": 0.498,
      "step": 8773
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6222666501998901,
      "learning_rate": 1.6993596145508448e-05,
      "loss": 0.4828,
      "step": 8774
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6497551202774048,
      "learning_rate": 1.6992868179999042e-05,
      "loss": 0.4816,
      "step": 8775
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5783988237380981,
      "learning_rate": 1.699214014196241e-05,
      "loss": 0.455,
      "step": 8776
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6695986986160278,
      "learning_rate": 1.6991412031406105e-05,
      "loss": 0.4945,
      "step": 8777
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5974929332733154,
      "learning_rate": 1.6990683848337676e-05,
      "loss": 0.4843,
      "step": 8778
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7171015739440918,
      "learning_rate": 1.698995559276468e-05,
      "loss": 0.5384,
      "step": 8779
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6284719705581665,
      "learning_rate": 1.6989227264694662e-05,
      "loss": 0.4572,
      "step": 8780
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6562494039535522,
      "learning_rate": 1.6988498864135182e-05,
      "loss": 0.5181,
      "step": 8781
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7488560676574707,
      "learning_rate": 1.6987770391093795e-05,
      "loss": 0.5123,
      "step": 8782
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6908338069915771,
      "learning_rate": 1.6987041845578057e-05,
      "loss": 0.5354,
      "step": 8783
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6042845249176025,
      "learning_rate": 1.6986313227595523e-05,
      "loss": 0.4753,
      "step": 8784
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5347392559051514,
      "learning_rate": 1.6985584537153745e-05,
      "loss": 0.4889,
      "step": 8785
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6861495971679688,
      "learning_rate": 1.698485577426029e-05,
      "loss": 0.465,
      "step": 8786
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6094422340393066,
      "learning_rate": 1.6984126938922707e-05,
      "loss": 0.5463,
      "step": 8787
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7061716318130493,
      "learning_rate": 1.6983398031148558e-05,
      "loss": 0.5436,
      "step": 8788
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.67552649974823,
      "learning_rate": 1.6982669050945408e-05,
      "loss": 0.5409,
      "step": 8789
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6404258012771606,
      "learning_rate": 1.698193999832081e-05,
      "loss": 0.5134,
      "step": 8790
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.689302921295166,
      "learning_rate": 1.6981210873282336e-05,
      "loss": 0.552,
      "step": 8791
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7043052911758423,
      "learning_rate": 1.6980481675837535e-05,
      "loss": 0.479,
      "step": 8792
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.708617925643921,
      "learning_rate": 1.6979752405993983e-05,
      "loss": 0.533,
      "step": 8793
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5518178939819336,
      "learning_rate": 1.6979023063759234e-05,
      "loss": 0.433,
      "step": 8794
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6895850896835327,
      "learning_rate": 1.6978293649140854e-05,
      "loss": 0.4954,
      "step": 8795
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7588378190994263,
      "learning_rate": 1.6977564162146412e-05,
      "loss": 0.4731,
      "step": 8796
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7105720043182373,
      "learning_rate": 1.697683460278347e-05,
      "loss": 0.6178,
      "step": 8797
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6099125146865845,
      "learning_rate": 1.69761049710596e-05,
      "loss": 0.453,
      "step": 8798
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.624066948890686,
      "learning_rate": 1.6975375266982364e-05,
      "loss": 0.5079,
      "step": 8799
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7505764961242676,
      "learning_rate": 1.6974645490559333e-05,
      "loss": 0.5392,
      "step": 8800
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7741881608963013,
      "learning_rate": 1.6973915641798075e-05,
      "loss": 0.539,
      "step": 8801
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7039146423339844,
      "learning_rate": 1.697318572070616e-05,
      "loss": 0.463,
      "step": 8802
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6053016185760498,
      "learning_rate": 1.697245572729116e-05,
      "loss": 0.5635,
      "step": 8803
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7249683141708374,
      "learning_rate": 1.6971725661560644e-05,
      "loss": 0.4769,
      "step": 8804
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8405632972717285,
      "learning_rate": 1.697099552352218e-05,
      "loss": 0.6187,
      "step": 8805
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.620566725730896,
      "learning_rate": 1.697026531318335e-05,
      "loss": 0.4638,
      "step": 8806
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6328938007354736,
      "learning_rate": 1.6969535030551723e-05,
      "loss": 0.5087,
      "step": 8807
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5723278522491455,
      "learning_rate": 1.696880467563487e-05,
      "loss": 0.4617,
      "step": 8808
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.773619294166565,
      "learning_rate": 1.696807424844037e-05,
      "loss": 0.6282,
      "step": 8809
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6023517847061157,
      "learning_rate": 1.69673437489758e-05,
      "loss": 0.4632,
      "step": 8810
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6410887241363525,
      "learning_rate": 1.696661317724873e-05,
      "loss": 0.5316,
      "step": 8811
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.66350519657135,
      "learning_rate": 1.6965882533266743e-05,
      "loss": 0.4946,
      "step": 8812
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.597334384918213,
      "learning_rate": 1.6965151817037414e-05,
      "loss": 0.4939,
      "step": 8813
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6805764436721802,
      "learning_rate": 1.6964421028568325e-05,
      "loss": 0.5439,
      "step": 8814
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5777337551116943,
      "learning_rate": 1.6963690167867054e-05,
      "loss": 0.5028,
      "step": 8815
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7334554195404053,
      "learning_rate": 1.6962959234941177e-05,
      "loss": 0.4902,
      "step": 8816
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5665463209152222,
      "learning_rate": 1.696222822979828e-05,
      "loss": 0.5095,
      "step": 8817
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.748910665512085,
      "learning_rate": 1.696149715244594e-05,
      "loss": 0.5531,
      "step": 8818
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6587365865707397,
      "learning_rate": 1.696076600289175e-05,
      "loss": 0.583,
      "step": 8819
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5202062129974365,
      "learning_rate": 1.696003478114328e-05,
      "loss": 0.4487,
      "step": 8820
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6066876649856567,
      "learning_rate": 1.695930348720812e-05,
      "loss": 0.5271,
      "step": 8821
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6011959314346313,
      "learning_rate": 1.6958572121093853e-05,
      "loss": 0.5121,
      "step": 8822
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6519840955734253,
      "learning_rate": 1.695784068280807e-05,
      "loss": 0.5123,
      "step": 8823
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5502135753631592,
      "learning_rate": 1.695710917235835e-05,
      "loss": 0.4537,
      "step": 8824
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5843493938446045,
      "learning_rate": 1.6956377589752286e-05,
      "loss": 0.522,
      "step": 8825
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.533831000328064,
      "learning_rate": 1.695564593499746e-05,
      "loss": 0.4689,
      "step": 8826
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5251004695892334,
      "learning_rate": 1.6954914208101464e-05,
      "loss": 0.4987,
      "step": 8827
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6411646604537964,
      "learning_rate": 1.6954182409071884e-05,
      "loss": 0.4897,
      "step": 8828
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.696203351020813,
      "learning_rate": 1.6953450537916313e-05,
      "loss": 0.5373,
      "step": 8829
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.732131004333496,
      "learning_rate": 1.6952718594642344e-05,
      "loss": 0.5265,
      "step": 8830
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.713285207748413,
      "learning_rate": 1.6951986579257563e-05,
      "loss": 0.5236,
      "step": 8831
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7284592390060425,
      "learning_rate": 1.695125449176956e-05,
      "loss": 0.5503,
      "step": 8832
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.673668622970581,
      "learning_rate": 1.695052233218594e-05,
      "loss": 0.4744,
      "step": 8833
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6491156816482544,
      "learning_rate": 1.6949790100514284e-05,
      "loss": 0.4728,
      "step": 8834
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6973724365234375,
      "learning_rate": 1.6949057796762195e-05,
      "loss": 0.5389,
      "step": 8835
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6101540327072144,
      "learning_rate": 1.694832542093726e-05,
      "loss": 0.4936,
      "step": 8836
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6771254539489746,
      "learning_rate": 1.6947592973047084e-05,
      "loss": 0.5349,
      "step": 8837
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6638195514678955,
      "learning_rate": 1.6946860453099258e-05,
      "loss": 0.4975,
      "step": 8838
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7215080261230469,
      "learning_rate": 1.694612786110138e-05,
      "loss": 1.0333,
      "step": 8839
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5750178098678589,
      "learning_rate": 1.6945395197061045e-05,
      "loss": 1.0527,
      "step": 8840
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8143877983093262,
      "learning_rate": 1.694466246098586e-05,
      "loss": 0.5784,
      "step": 8841
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6516036987304688,
      "learning_rate": 1.694392965288342e-05,
      "loss": 0.4664,
      "step": 8842
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7104003429412842,
      "learning_rate": 1.6943196772761324e-05,
      "loss": 0.5201,
      "step": 8843
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5491771697998047,
      "learning_rate": 1.6942463820627174e-05,
      "loss": 0.4539,
      "step": 8844
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7949790954589844,
      "learning_rate": 1.6941730796488573e-05,
      "loss": 0.5337,
      "step": 8845
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.645601511001587,
      "learning_rate": 1.6940997700353127e-05,
      "loss": 0.4817,
      "step": 8846
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6411949396133423,
      "learning_rate": 1.694026453222843e-05,
      "loss": 0.56,
      "step": 8847
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5830538272857666,
      "learning_rate": 1.6939531292122094e-05,
      "loss": 0.4801,
      "step": 8848
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6803959608078003,
      "learning_rate": 1.6938797980041724e-05,
      "loss": 0.5315,
      "step": 8849
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.22733736038208,
      "learning_rate": 1.6938064595994917e-05,
      "loss": 0.4926,
      "step": 8850
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6964489221572876,
      "learning_rate": 1.6937331139989294e-05,
      "loss": 0.526,
      "step": 8851
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.63381826877594,
      "learning_rate": 1.693659761203245e-05,
      "loss": 0.4883,
      "step": 8852
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5850192308425903,
      "learning_rate": 1.6935864012131995e-05,
      "loss": 0.5219,
      "step": 8853
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6099849939346313,
      "learning_rate": 1.693513034029554e-05,
      "loss": 0.4613,
      "step": 8854
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7210711240768433,
      "learning_rate": 1.6934396596530692e-05,
      "loss": 0.5728,
      "step": 8855
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7302441596984863,
      "learning_rate": 1.6933662780845067e-05,
      "loss": 0.5173,
      "step": 8856
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6717149019241333,
      "learning_rate": 1.693292889324627e-05,
      "loss": 0.5467,
      "step": 8857
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6475269794464111,
      "learning_rate": 1.693219493374191e-05,
      "loss": 0.49,
      "step": 8858
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5676778554916382,
      "learning_rate": 1.693146090233961e-05,
      "loss": 0.5587,
      "step": 8859
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.748064637184143,
      "learning_rate": 1.693072679904697e-05,
      "loss": 0.5479,
      "step": 8860
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7120152711868286,
      "learning_rate": 1.6929992623871618e-05,
      "loss": 0.5583,
      "step": 8861
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6615996360778809,
      "learning_rate": 1.6929258376821152e-05,
      "loss": 0.5,
      "step": 8862
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7131037712097168,
      "learning_rate": 1.69285240579032e-05,
      "loss": 0.5623,
      "step": 8863
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6387100219726562,
      "learning_rate": 1.6927789667125376e-05,
      "loss": 0.4742,
      "step": 8864
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6086020469665527,
      "learning_rate": 1.6927055204495295e-05,
      "loss": 0.4944,
      "step": 8865
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7836604118347168,
      "learning_rate": 1.692632067002057e-05,
      "loss": 0.5174,
      "step": 8866
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6260449886322021,
      "learning_rate": 1.692558606370883e-05,
      "loss": 0.5152,
      "step": 8867
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7411423921585083,
      "learning_rate": 1.6924851385567684e-05,
      "loss": 0.4986,
      "step": 8868
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.666813611984253,
      "learning_rate": 1.692411663560476e-05,
      "loss": 0.5764,
      "step": 8869
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5498919486999512,
      "learning_rate": 1.692338181382767e-05,
      "loss": 0.4598,
      "step": 8870
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6339507102966309,
      "learning_rate": 1.692264692024404e-05,
      "loss": 0.5023,
      "step": 8871
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6115365028381348,
      "learning_rate": 1.692191195486149e-05,
      "loss": 0.4232,
      "step": 8872
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7984073162078857,
      "learning_rate": 1.6921176917687646e-05,
      "loss": 0.6201,
      "step": 8873
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6245570182800293,
      "learning_rate": 1.692044180873013e-05,
      "loss": 0.4873,
      "step": 8874
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6294821500778198,
      "learning_rate": 1.691970662799657e-05,
      "loss": 0.5049,
      "step": 8875
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7350306510925293,
      "learning_rate": 1.691897137549458e-05,
      "loss": 0.5599,
      "step": 8876
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.691654920578003,
      "learning_rate": 1.69182360512318e-05,
      "loss": 0.53,
      "step": 8877
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6821088790893555,
      "learning_rate": 1.6917500655215843e-05,
      "loss": 0.5345,
      "step": 8878
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6215356588363647,
      "learning_rate": 1.6916765187454346e-05,
      "loss": 0.5163,
      "step": 8879
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.682353138923645,
      "learning_rate": 1.6916029647954934e-05,
      "loss": 0.4796,
      "step": 8880
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7230541706085205,
      "learning_rate": 1.6915294036725235e-05,
      "loss": 0.6179,
      "step": 8881
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6134341955184937,
      "learning_rate": 1.6914558353772876e-05,
      "loss": 0.459,
      "step": 8882
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7159054279327393,
      "learning_rate": 1.6913822599105493e-05,
      "loss": 0.5339,
      "step": 8883
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.559846043586731,
      "learning_rate": 1.6913086772730713e-05,
      "loss": 0.4561,
      "step": 8884
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5892835855484009,
      "learning_rate": 1.6912350874656166e-05,
      "loss": 0.4851,
      "step": 8885
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6649377346038818,
      "learning_rate": 1.691161490488949e-05,
      "loss": 0.4952,
      "step": 8886
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5933470726013184,
      "learning_rate": 1.6910878863438316e-05,
      "loss": 0.5195,
      "step": 8887
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7552176713943481,
      "learning_rate": 1.6910142750310273e-05,
      "loss": 0.4918,
      "step": 8888
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8756574392318726,
      "learning_rate": 1.6909406565513003e-05,
      "loss": 0.512,
      "step": 8889
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5739524364471436,
      "learning_rate": 1.6908670309054137e-05,
      "loss": 0.4548,
      "step": 8890
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0765957832336426,
      "learning_rate": 1.6907933980941312e-05,
      "loss": 1.0661,
      "step": 8891
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8206344842910767,
      "learning_rate": 1.6907197581182168e-05,
      "loss": 1.0933,
      "step": 8892
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.969501256942749,
      "learning_rate": 1.6906461109784338e-05,
      "loss": 0.5627,
      "step": 8893
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8811100721359253,
      "learning_rate": 1.6905724566755463e-05,
      "loss": 0.4678,
      "step": 8894
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6358507871627808,
      "learning_rate": 1.690498795210318e-05,
      "loss": 0.4973,
      "step": 8895
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.676559567451477,
      "learning_rate": 1.6904251265835134e-05,
      "loss": 0.5077,
      "step": 8896
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7341541051864624,
      "learning_rate": 1.6903514507958957e-05,
      "loss": 0.549,
      "step": 8897
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7012934684753418,
      "learning_rate": 1.6902777678482295e-05,
      "loss": 0.4867,
      "step": 8898
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7036296129226685,
      "learning_rate": 1.6902040777412797e-05,
      "loss": 0.565,
      "step": 8899
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6287657022476196,
      "learning_rate": 1.6901303804758094e-05,
      "loss": 0.4651,
      "step": 8900
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6746479272842407,
      "learning_rate": 1.6900566760525837e-05,
      "loss": 0.5571,
      "step": 8901
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6918140649795532,
      "learning_rate": 1.6899829644723668e-05,
      "loss": 0.5051,
      "step": 8902
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7629631757736206,
      "learning_rate": 1.6899092457359228e-05,
      "loss": 0.5517,
      "step": 8903
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5861637592315674,
      "learning_rate": 1.689835519844017e-05,
      "loss": 0.4816,
      "step": 8904
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.367642402648926,
      "learning_rate": 1.6897617867974138e-05,
      "loss": 1.065,
      "step": 8905
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8383338451385498,
      "learning_rate": 1.689688046596878e-05,
      "loss": 1.0734,
      "step": 8906
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9249639511108398,
      "learning_rate": 1.689614299243174e-05,
      "loss": 0.5811,
      "step": 8907
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8224939107894897,
      "learning_rate": 1.689540544737067e-05,
      "loss": 0.4643,
      "step": 8908
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8755584955215454,
      "learning_rate": 1.6894667830793217e-05,
      "loss": 0.5932,
      "step": 8909
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6671050786972046,
      "learning_rate": 1.6893930142707043e-05,
      "loss": 0.4592,
      "step": 8910
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7559757232666016,
      "learning_rate": 1.689319238311978e-05,
      "loss": 0.5111,
      "step": 8911
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8512166738510132,
      "learning_rate": 1.6892454552039096e-05,
      "loss": 0.5213,
      "step": 8912
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7614319324493408,
      "learning_rate": 1.6891716649472633e-05,
      "loss": 0.5022,
      "step": 8913
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9336533546447754,
      "learning_rate": 1.6890978675428048e-05,
      "loss": 0.5893,
      "step": 8914
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5515074729919434,
      "learning_rate": 1.6890240629912994e-05,
      "loss": 0.5096,
      "step": 8915
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6123625040054321,
      "learning_rate": 1.6889502512935127e-05,
      "loss": 0.4637,
      "step": 8916
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9717659950256348,
      "learning_rate": 1.68887643245021e-05,
      "loss": 0.5252,
      "step": 8917
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7614011764526367,
      "learning_rate": 1.6888026064621573e-05,
      "loss": 0.4995,
      "step": 8918
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.696585774421692,
      "learning_rate": 1.6887287733301204e-05,
      "loss": 0.5066,
      "step": 8919
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.556459665298462,
      "learning_rate": 1.6886549330548647e-05,
      "loss": 0.4451,
      "step": 8920
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7834367752075195,
      "learning_rate": 1.688581085637156e-05,
      "loss": 0.5826,
      "step": 8921
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6756144762039185,
      "learning_rate": 1.6885072310777603e-05,
      "loss": 0.4666,
      "step": 8922
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8298300504684448,
      "learning_rate": 1.688433369377444e-05,
      "loss": 0.5338,
      "step": 8923
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6892657279968262,
      "learning_rate": 1.6883595005369723e-05,
      "loss": 0.4854,
      "step": 8924
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6837871074676514,
      "learning_rate": 1.688285624557112e-05,
      "loss": 0.5871,
      "step": 8925
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6389015913009644,
      "learning_rate": 1.688211741438629e-05,
      "loss": 0.469,
      "step": 8926
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7782713174819946,
      "learning_rate": 1.68813785118229e-05,
      "loss": 0.6272,
      "step": 8927
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8867686986923218,
      "learning_rate": 1.688063953788861e-05,
      "loss": 0.4919,
      "step": 8928
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.644863247871399,
      "learning_rate": 1.6879900492591086e-05,
      "loss": 0.5841,
      "step": 8929
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7049204111099243,
      "learning_rate": 1.6879161375937992e-05,
      "loss": 0.4932,
      "step": 8930
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6170408725738525,
      "learning_rate": 1.6878422187936995e-05,
      "loss": 0.5661,
      "step": 8931
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6519182920455933,
      "learning_rate": 1.687768292859576e-05,
      "loss": 0.4984,
      "step": 8932
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6685329675674438,
      "learning_rate": 1.6876943597921955e-05,
      "loss": 0.5998,
      "step": 8933
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6681476831436157,
      "learning_rate": 1.6876204195923248e-05,
      "loss": 0.4731,
      "step": 8934
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6008065938949585,
      "learning_rate": 1.687546472260731e-05,
      "loss": 0.5348,
      "step": 8935
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6176866292953491,
      "learning_rate": 1.6874725177981807e-05,
      "loss": 0.4827,
      "step": 8936
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.736162543296814,
      "learning_rate": 1.6873985562054408e-05,
      "loss": 0.5465,
      "step": 8937
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.755217432975769,
      "learning_rate": 1.6873245874832788e-05,
      "loss": 0.5246,
      "step": 8938
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7162904739379883,
      "learning_rate": 1.6872506116324618e-05,
      "loss": 0.5261,
      "step": 8939
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6390364170074463,
      "learning_rate": 1.687176628653757e-05,
      "loss": 0.4725,
      "step": 8940
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7365920543670654,
      "learning_rate": 1.6871026385479317e-05,
      "loss": 0.5741,
      "step": 8941
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.552756428718567,
      "learning_rate": 1.6870286413157534e-05,
      "loss": 0.4498,
      "step": 8942
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7118465900421143,
      "learning_rate": 1.6869546369579893e-05,
      "loss": 0.5555,
      "step": 8943
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5869661569595337,
      "learning_rate": 1.686880625475407e-05,
      "loss": 0.4765,
      "step": 8944
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7533361911773682,
      "learning_rate": 1.6868066068687744e-05,
      "loss": 0.5519,
      "step": 8945
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6922607421875,
      "learning_rate": 1.6867325811388593e-05,
      "loss": 0.4813,
      "step": 8946
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7546367645263672,
      "learning_rate": 1.686658548286429e-05,
      "loss": 0.5764,
      "step": 8947
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6584751605987549,
      "learning_rate": 1.6865845083122515e-05,
      "loss": 0.488,
      "step": 8948
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6698977947235107,
      "learning_rate": 1.6865104612170945e-05,
      "loss": 0.5196,
      "step": 8949
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.741740107536316,
      "learning_rate": 1.6864364070017267e-05,
      "loss": 0.519,
      "step": 8950
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5670483112335205,
      "learning_rate": 1.6863623456669152e-05,
      "loss": 0.5217,
      "step": 8951
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.597656488418579,
      "learning_rate": 1.686288277213429e-05,
      "loss": 0.4477,
      "step": 8952
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.708709478378296,
      "learning_rate": 1.6862142016420358e-05,
      "loss": 0.5221,
      "step": 8953
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5961823463439941,
      "learning_rate": 1.686140118953504e-05,
      "loss": 0.479,
      "step": 8954
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6688833236694336,
      "learning_rate": 1.6860660291486023e-05,
      "loss": 0.5849,
      "step": 8955
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6955033540725708,
      "learning_rate": 1.6859919322280986e-05,
      "loss": 0.4637,
      "step": 8956
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6287477016448975,
      "learning_rate": 1.685917828192761e-05,
      "loss": 0.513,
      "step": 8957
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6904836893081665,
      "learning_rate": 1.6858437170433592e-05,
      "loss": 0.4936,
      "step": 8958
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6788053512573242,
      "learning_rate": 1.6857695987806614e-05,
      "loss": 0.5716,
      "step": 8959
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6335233449935913,
      "learning_rate": 1.6856954734054363e-05,
      "loss": 0.4974,
      "step": 8960
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8559683561325073,
      "learning_rate": 1.6856213409184526e-05,
      "loss": 0.5285,
      "step": 8961
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.669294834136963,
      "learning_rate": 1.685547201320479e-05,
      "loss": 0.4428,
      "step": 8962
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8009474277496338,
      "learning_rate": 1.685473054612285e-05,
      "loss": 0.5708,
      "step": 8963
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5686997175216675,
      "learning_rate": 1.685398900794639e-05,
      "loss": 0.4897,
      "step": 8964
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.830374836921692,
      "learning_rate": 1.685324739868311e-05,
      "loss": 0.5424,
      "step": 8965
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6457979679107666,
      "learning_rate": 1.685250571834069e-05,
      "loss": 0.4698,
      "step": 8966
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6432199478149414,
      "learning_rate": 1.6851763966926828e-05,
      "loss": 0.5893,
      "step": 8967
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7282708883285522,
      "learning_rate": 1.6851022144449216e-05,
      "loss": 0.5104,
      "step": 8968
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7294813394546509,
      "learning_rate": 1.6850280250915552e-05,
      "loss": 0.5924,
      "step": 8969
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6128036975860596,
      "learning_rate": 1.6849538286333527e-05,
      "loss": 0.4848,
      "step": 8970
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.71047043800354,
      "learning_rate": 1.6848796250710835e-05,
      "loss": 0.5532,
      "step": 8971
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6377699375152588,
      "learning_rate": 1.6848054144055178e-05,
      "loss": 0.4761,
      "step": 8972
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5835705995559692,
      "learning_rate": 1.6847311966374245e-05,
      "loss": 0.5207,
      "step": 8973
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.700022578239441,
      "learning_rate": 1.684656971767574e-05,
      "loss": 0.5026,
      "step": 8974
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.212273597717285,
      "learning_rate": 1.684582739796736e-05,
      "loss": 1.0114,
      "step": 8975
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7817679643630981,
      "learning_rate": 1.68450850072568e-05,
      "loss": 1.0083,
      "step": 8976
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8749072551727295,
      "learning_rate": 1.6844342545551762e-05,
      "loss": 0.5177,
      "step": 8977
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9604195356369019,
      "learning_rate": 1.684360001285995e-05,
      "loss": 0.4874,
      "step": 8978
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7834298610687256,
      "learning_rate": 1.684285740918906e-05,
      "loss": 0.5291,
      "step": 8979
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6551711559295654,
      "learning_rate": 1.68421147345468e-05,
      "loss": 0.4855,
      "step": 8980
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.582902193069458,
      "learning_rate": 1.6841371988940865e-05,
      "loss": 0.5101,
      "step": 8981
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7699544429779053,
      "learning_rate": 1.6840629172378962e-05,
      "loss": 0.481,
      "step": 8982
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7491912841796875,
      "learning_rate": 1.68398862848688e-05,
      "loss": 0.525,
      "step": 8983
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.618108868598938,
      "learning_rate": 1.683914332641808e-05,
      "loss": 0.5032,
      "step": 8984
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.753187656402588,
      "learning_rate": 1.6838400297034503e-05,
      "loss": 0.5576,
      "step": 8985
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6465948820114136,
      "learning_rate": 1.6837657196725783e-05,
      "loss": 0.5141,
      "step": 8986
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.652029275894165,
      "learning_rate": 1.683691402549962e-05,
      "loss": 0.5007,
      "step": 8987
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5818253755569458,
      "learning_rate": 1.683617078336373e-05,
      "loss": 0.4719,
      "step": 8988
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8350293636322021,
      "learning_rate": 1.6835427470325817e-05,
      "loss": 0.5668,
      "step": 8989
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6462515592575073,
      "learning_rate": 1.683468408639359e-05,
      "loss": 0.4752,
      "step": 8990
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.271688461303711,
      "learning_rate": 1.6833940631574756e-05,
      "loss": 1.0369,
      "step": 8991
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9853790998458862,
      "learning_rate": 1.6833197105877037e-05,
      "loss": 1.1175,
      "step": 8992
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8422733545303345,
      "learning_rate": 1.683245350930813e-05,
      "loss": 0.5934,
      "step": 8993
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8037192821502686,
      "learning_rate": 1.683170984187576e-05,
      "loss": 0.5306,
      "step": 8994
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7978283166885376,
      "learning_rate": 1.6830966103587632e-05,
      "loss": 0.6009,
      "step": 8995
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5955783128738403,
      "learning_rate": 1.6830222294451465e-05,
      "loss": 0.464,
      "step": 8996
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6424033641815186,
      "learning_rate": 1.6829478414474965e-05,
      "loss": 0.5087,
      "step": 8997
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.700451374053955,
      "learning_rate": 1.6828734463665856e-05,
      "loss": 0.5012,
      "step": 8998
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6059128046035767,
      "learning_rate": 1.682799044203185e-05,
      "loss": 0.5236,
      "step": 8999
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.592237949371338,
      "learning_rate": 1.6827246349580665e-05,
      "loss": 0.474,
      "step": 9000
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.734879732131958,
      "learning_rate": 1.6826502186320022e-05,
      "loss": 0.5652,
      "step": 9001
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6102744340896606,
      "learning_rate": 1.682575795225763e-05,
      "loss": 0.4553,
      "step": 9002
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9968098402023315,
      "learning_rate": 1.6825013647401215e-05,
      "loss": 1.1032,
      "step": 9003
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6457605361938477,
      "learning_rate": 1.6824269271758496e-05,
      "loss": 1.0339,
      "step": 9004
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.672714114189148,
      "learning_rate": 1.682352482533719e-05,
      "loss": 0.5134,
      "step": 9005
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6581209897994995,
      "learning_rate": 1.682278030814502e-05,
      "loss": 0.4685,
      "step": 9006
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7139685153961182,
      "learning_rate": 1.6822035720189707e-05,
      "loss": 0.5662,
      "step": 9007
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6498665809631348,
      "learning_rate": 1.682129106147898e-05,
      "loss": 0.4925,
      "step": 9008
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6774768829345703,
      "learning_rate": 1.6820546332020552e-05,
      "loss": 0.5131,
      "step": 9009
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.617845892906189,
      "learning_rate": 1.6819801531822152e-05,
      "loss": 0.4756,
      "step": 9010
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6163547039031982,
      "learning_rate": 1.6819056660891507e-05,
      "loss": 0.5478,
      "step": 9011
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7259060144424438,
      "learning_rate": 1.681831171923634e-05,
      "loss": 0.532,
      "step": 9012
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.605991005897522,
      "learning_rate": 1.6817566706864377e-05,
      "loss": 0.5076,
      "step": 9013
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.588334083557129,
      "learning_rate": 1.6816821623783345e-05,
      "loss": 0.4658,
      "step": 9014
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5750138759613037,
      "learning_rate": 1.6816076470000972e-05,
      "loss": 0.4978,
      "step": 9015
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7012239694595337,
      "learning_rate": 1.6815331245524985e-05,
      "loss": 0.4771,
      "step": 9016
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5817644596099854,
      "learning_rate": 1.681458595036312e-05,
      "loss": 0.5388,
      "step": 9017
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5838207006454468,
      "learning_rate": 1.68138405845231e-05,
      "loss": 0.4365,
      "step": 9018
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5828428268432617,
      "learning_rate": 1.681309514801265e-05,
      "loss": 0.481,
      "step": 9019
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6736401319503784,
      "learning_rate": 1.6812349640839516e-05,
      "loss": 0.5214,
      "step": 9020
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6156301498413086,
      "learning_rate": 1.681160406301142e-05,
      "loss": 0.4995,
      "step": 9021
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.628563642501831,
      "learning_rate": 1.68108584145361e-05,
      "loss": 0.5001,
      "step": 9022
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7267777919769287,
      "learning_rate": 1.6810112695421286e-05,
      "loss": 0.6118,
      "step": 9023
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5742353200912476,
      "learning_rate": 1.680936690567471e-05,
      "loss": 0.4586,
      "step": 9024
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6983119249343872,
      "learning_rate": 1.6808621045304114e-05,
      "loss": 0.538,
      "step": 9025
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5505489110946655,
      "learning_rate": 1.680787511431723e-05,
      "loss": 0.4597,
      "step": 9026
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.648183822631836,
      "learning_rate": 1.6807129112721792e-05,
      "loss": 0.5563,
      "step": 9027
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5944887399673462,
      "learning_rate": 1.6806383040525545e-05,
      "loss": 0.4638,
      "step": 9028
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7765086889266968,
      "learning_rate": 1.6805636897736216e-05,
      "loss": 0.567,
      "step": 9029
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.581311583518982,
      "learning_rate": 1.6804890684361555e-05,
      "loss": 0.4574,
      "step": 9030
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7316170930862427,
      "learning_rate": 1.6804144400409294e-05,
      "loss": 0.6443,
      "step": 9031
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6313652992248535,
      "learning_rate": 1.6803398045887177e-05,
      "loss": 0.462,
      "step": 9032
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6883502006530762,
      "learning_rate": 1.6802651620802936e-05,
      "loss": 0.5399,
      "step": 9033
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.683804988861084,
      "learning_rate": 1.6801905125164327e-05,
      "loss": 0.4631,
      "step": 9034
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6554831266403198,
      "learning_rate": 1.680115855897908e-05,
      "loss": 0.5395,
      "step": 9035
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.687589168548584,
      "learning_rate": 1.6800411922254948e-05,
      "loss": 0.5137,
      "step": 9036
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5963830947875977,
      "learning_rate": 1.6799665214999668e-05,
      "loss": 0.5726,
      "step": 9037
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6539485454559326,
      "learning_rate": 1.6798918437220988e-05,
      "loss": 0.4346,
      "step": 9038
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.205367088317871,
      "learning_rate": 1.679817158892665e-05,
      "loss": 0.6001,
      "step": 9039
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6427783966064453,
      "learning_rate": 1.67974246701244e-05,
      "loss": 0.4931,
      "step": 9040
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7405824661254883,
      "learning_rate": 1.679667768082199e-05,
      "loss": 0.5436,
      "step": 9041
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5362123250961304,
      "learning_rate": 1.679593062102716e-05,
      "loss": 0.4507,
      "step": 9042
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8911362886428833,
      "learning_rate": 1.6795183490747666e-05,
      "loss": 0.5312,
      "step": 9043
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5792053937911987,
      "learning_rate": 1.6794436289991255e-05,
      "loss": 0.4644,
      "step": 9044
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6042449474334717,
      "learning_rate": 1.679368901876567e-05,
      "loss": 0.51,
      "step": 9045
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7283998727798462,
      "learning_rate": 1.6792941677078668e-05,
      "loss": 0.497,
      "step": 9046
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.642228603363037,
      "learning_rate": 1.6792194264937996e-05,
      "loss": 0.5515,
      "step": 9047
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6241852045059204,
      "learning_rate": 1.6791446782351413e-05,
      "loss": 0.4909,
      "step": 9048
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8254122734069824,
      "learning_rate": 1.6790699229326665e-05,
      "loss": 0.5436,
      "step": 9049
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5711123943328857,
      "learning_rate": 1.6789951605871503e-05,
      "loss": 0.4595,
      "step": 9050
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6440963745117188,
      "learning_rate": 1.678920391199369e-05,
      "loss": 0.4905,
      "step": 9051
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8277679681777954,
      "learning_rate": 1.6788456147700977e-05,
      "loss": 0.5148,
      "step": 9052
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6497362852096558,
      "learning_rate": 1.678770831300111e-05,
      "loss": 0.5297,
      "step": 9053
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6600991487503052,
      "learning_rate": 1.6786960407901866e-05,
      "loss": 0.5093,
      "step": 9054
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.107795238494873,
      "learning_rate": 1.6786212432410982e-05,
      "loss": 0.9602,
      "step": 9055
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8395949602127075,
      "learning_rate": 1.6785464386536223e-05,
      "loss": 1.089,
      "step": 9056
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8311270475387573,
      "learning_rate": 1.678471627028535e-05,
      "loss": 0.507,
      "step": 9057
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8129295110702515,
      "learning_rate": 1.678396808366612e-05,
      "loss": 0.5508,
      "step": 9058
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6893900632858276,
      "learning_rate": 1.6783219826686292e-05,
      "loss": 0.5001,
      "step": 9059
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.836509108543396,
      "learning_rate": 1.678247149935363e-05,
      "loss": 0.5883,
      "step": 9060
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6591689586639404,
      "learning_rate": 1.6781723101675885e-05,
      "loss": 0.5724,
      "step": 9061
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6750938892364502,
      "learning_rate": 1.6780974633660834e-05,
      "loss": 0.486,
      "step": 9062
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6932989358901978,
      "learning_rate": 1.678022609531623e-05,
      "loss": 0.541,
      "step": 9063
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6691904067993164,
      "learning_rate": 1.6779477486649845e-05,
      "loss": 0.4718,
      "step": 9064
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.658645749092102,
      "learning_rate": 1.6778728807669428e-05,
      "loss": 0.4733,
      "step": 9065
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.704160451889038,
      "learning_rate": 1.6777980058382762e-05,
      "loss": 0.5276,
      "step": 9066
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8011521100997925,
      "learning_rate": 1.67772312387976e-05,
      "loss": 0.5436,
      "step": 9067
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.642177700996399,
      "learning_rate": 1.6776482348921713e-05,
      "loss": 0.4969,
      "step": 9068
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6357665061950684,
      "learning_rate": 1.677573338876287e-05,
      "loss": 0.5908,
      "step": 9069
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6320465803146362,
      "learning_rate": 1.6774984358328836e-05,
      "loss": 0.4703,
      "step": 9070
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6980476379394531,
      "learning_rate": 1.6774235257627377e-05,
      "loss": 0.5212,
      "step": 9071
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6679131984710693,
      "learning_rate": 1.677348608666627e-05,
      "loss": 0.4519,
      "step": 9072
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6431658267974854,
      "learning_rate": 1.677273684545328e-05,
      "loss": 0.5641,
      "step": 9073
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6016513109207153,
      "learning_rate": 1.6771987533996176e-05,
      "loss": 0.4959,
      "step": 9074
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5928338766098022,
      "learning_rate": 1.6771238152302735e-05,
      "loss": 0.5314,
      "step": 9075
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.617832899093628,
      "learning_rate": 1.6770488700380728e-05,
      "loss": 0.4739,
      "step": 9076
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6849427223205566,
      "learning_rate": 1.6769739178237924e-05,
      "loss": 0.5215,
      "step": 9077
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.709747552871704,
      "learning_rate": 1.67689895858821e-05,
      "loss": 0.4891,
      "step": 9078
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7769325971603394,
      "learning_rate": 1.676823992332103e-05,
      "loss": 0.5738,
      "step": 9079
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6828782558441162,
      "learning_rate": 1.676749019056249e-05,
      "loss": 0.48,
      "step": 9080
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6319079399108887,
      "learning_rate": 1.676674038761425e-05,
      "loss": 0.5412,
      "step": 9081
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6607999801635742,
      "learning_rate": 1.67659905144841e-05,
      "loss": 0.5042,
      "step": 9082
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6805020570755005,
      "learning_rate": 1.67652405711798e-05,
      "loss": 0.5284,
      "step": 9083
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.633088231086731,
      "learning_rate": 1.6764490557709147e-05,
      "loss": 0.491,
      "step": 9084
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8532429933547974,
      "learning_rate": 1.67637404740799e-05,
      "loss": 0.5124,
      "step": 9085
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5337127447128296,
      "learning_rate": 1.6762990320299855e-05,
      "loss": 0.4307,
      "step": 9086
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5943520069122314,
      "learning_rate": 1.6762240096376784e-05,
      "loss": 0.5431,
      "step": 9087
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6348302364349365,
      "learning_rate": 1.676148980231847e-05,
      "loss": 0.4949,
      "step": 9088
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8745863437652588,
      "learning_rate": 1.6760739438132695e-05,
      "loss": 0.5482,
      "step": 9089
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7159284353256226,
      "learning_rate": 1.675998900382724e-05,
      "loss": 0.4315,
      "step": 9090
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6385239362716675,
      "learning_rate": 1.6759238499409888e-05,
      "loss": 0.5737,
      "step": 9091
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5754748582839966,
      "learning_rate": 1.6758487924888428e-05,
      "loss": 0.4588,
      "step": 9092
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6665287017822266,
      "learning_rate": 1.6757737280270637e-05,
      "loss": 0.513,
      "step": 9093
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.658071756362915,
      "learning_rate": 1.6756986565564306e-05,
      "loss": 0.4708,
      "step": 9094
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7504819631576538,
      "learning_rate": 1.675623578077722e-05,
      "loss": 0.5477,
      "step": 9095
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5506166219711304,
      "learning_rate": 1.6755484925917165e-05,
      "loss": 0.4603,
      "step": 9096
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7248057126998901,
      "learning_rate": 1.6754734000991927e-05,
      "loss": 0.5123,
      "step": 9097
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6130563020706177,
      "learning_rate": 1.6753983006009298e-05,
      "loss": 0.4801,
      "step": 9098
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6518210172653198,
      "learning_rate": 1.6753231940977062e-05,
      "loss": 0.5042,
      "step": 9099
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6957987546920776,
      "learning_rate": 1.6752480805903014e-05,
      "loss": 0.5042,
      "step": 9100
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.665348768234253,
      "learning_rate": 1.6751729600794944e-05,
      "loss": 0.544,
      "step": 9101
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6111102104187012,
      "learning_rate": 1.6750978325660637e-05,
      "loss": 0.4884,
      "step": 9102
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6048234701156616,
      "learning_rate": 1.6750226980507896e-05,
      "loss": 0.5474,
      "step": 9103
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7091164588928223,
      "learning_rate": 1.67494755653445e-05,
      "loss": 0.4832,
      "step": 9104
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6238619089126587,
      "learning_rate": 1.6748724080178255e-05,
      "loss": 0.5269,
      "step": 9105
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7142112255096436,
      "learning_rate": 1.6747972525016946e-05,
      "loss": 0.5392,
      "step": 9106
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7624961137771606,
      "learning_rate": 1.674722089986837e-05,
      "loss": 0.5175,
      "step": 9107
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.676020622253418,
      "learning_rate": 1.674646920474033e-05,
      "loss": 0.5149,
      "step": 9108
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8350850343704224,
      "learning_rate": 1.674571743964061e-05,
      "loss": 0.557,
      "step": 9109
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.614533543586731,
      "learning_rate": 1.674496560457702e-05,
      "loss": 0.4658,
      "step": 9110
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7084726095199585,
      "learning_rate": 1.6744213699557346e-05,
      "loss": 0.5031,
      "step": 9111
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6192938089370728,
      "learning_rate": 1.674346172458939e-05,
      "loss": 0.4829,
      "step": 9112
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6897374391555786,
      "learning_rate": 1.6742709679680958e-05,
      "loss": 0.5514,
      "step": 9113
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6345762014389038,
      "learning_rate": 1.674195756483984e-05,
      "loss": 0.4757,
      "step": 9114
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6250227689743042,
      "learning_rate": 1.6741205380073843e-05,
      "loss": 0.5011,
      "step": 9115
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5497040748596191,
      "learning_rate": 1.6740453125390764e-05,
      "loss": 0.4636,
      "step": 9116
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.971028208732605,
      "learning_rate": 1.673970080079841e-05,
      "loss": 0.5635,
      "step": 9117
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5865199565887451,
      "learning_rate": 1.673894840630458e-05,
      "loss": 0.4732,
      "step": 9118
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6446967124938965,
      "learning_rate": 1.673819594191708e-05,
      "loss": 0.5478,
      "step": 9119
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6653587818145752,
      "learning_rate": 1.6737443407643718e-05,
      "loss": 0.4864,
      "step": 9120
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.557937502861023,
      "learning_rate": 1.6736690803492287e-05,
      "loss": 0.4752,
      "step": 9121
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0018703937530518,
      "learning_rate": 1.67359381294706e-05,
      "loss": 0.4839,
      "step": 9122
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6405209302902222,
      "learning_rate": 1.6735185385586465e-05,
      "loss": 0.5663,
      "step": 9123
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6527924537658691,
      "learning_rate": 1.673443257184769e-05,
      "loss": 0.4655,
      "step": 9124
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5754176378250122,
      "learning_rate": 1.6733679688262077e-05,
      "loss": 0.4816,
      "step": 9125
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5676695108413696,
      "learning_rate": 1.6732926734837438e-05,
      "loss": 0.4499,
      "step": 9126
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.647802710533142,
      "learning_rate": 1.6732173711581583e-05,
      "loss": 0.4999,
      "step": 9127
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6550781726837158,
      "learning_rate": 1.673142061850232e-05,
      "loss": 0.4852,
      "step": 9128
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.66445791721344,
      "learning_rate": 1.6730667455607464e-05,
      "loss": 0.5228,
      "step": 9129
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6767915487289429,
      "learning_rate": 1.6729914222904823e-05,
      "loss": 0.4525,
      "step": 9130
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6890408992767334,
      "learning_rate": 1.672916092040221e-05,
      "loss": 0.5501,
      "step": 9131
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6663391590118408,
      "learning_rate": 1.6728407548107434e-05,
      "loss": 0.4595,
      "step": 9132
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6744492053985596,
      "learning_rate": 1.6727654106028315e-05,
      "loss": 0.5685,
      "step": 9133
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6004372835159302,
      "learning_rate": 1.6726900594172664e-05,
      "loss": 0.463,
      "step": 9134
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6393682956695557,
      "learning_rate": 1.6726147012548303e-05,
      "loss": 0.5504,
      "step": 9135
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5979753732681274,
      "learning_rate": 1.6725393361163033e-05,
      "loss": 0.457,
      "step": 9136
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8542627096176147,
      "learning_rate": 1.6724639640024684e-05,
      "loss": 0.5975,
      "step": 9137
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6595265865325928,
      "learning_rate": 1.672388584914107e-05,
      "loss": 0.509,
      "step": 9138
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7542378902435303,
      "learning_rate": 1.6723131988520003e-05,
      "loss": 0.5979,
      "step": 9139
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5530608892440796,
      "learning_rate": 1.672237805816931e-05,
      "loss": 0.4499,
      "step": 9140
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7255665063858032,
      "learning_rate": 1.6721624058096806e-05,
      "loss": 0.5831,
      "step": 9141
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6117594242095947,
      "learning_rate": 1.6720869988310314e-05,
      "loss": 0.442,
      "step": 9142
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8794196844100952,
      "learning_rate": 1.672011584881765e-05,
      "loss": 0.6052,
      "step": 9143
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6175181865692139,
      "learning_rate": 1.671936163962664e-05,
      "loss": 0.4556,
      "step": 9144
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5573503971099854,
      "learning_rate": 1.6718607360745103e-05,
      "loss": 0.4983,
      "step": 9145
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6719616651535034,
      "learning_rate": 1.671785301218087e-05,
      "loss": 0.524,
      "step": 9146
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6689637899398804,
      "learning_rate": 1.6717098593941753e-05,
      "loss": 0.518,
      "step": 9147
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7739704847335815,
      "learning_rate": 1.6716344106035584e-05,
      "loss": 0.5165,
      "step": 9148
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7181925773620605,
      "learning_rate": 1.6715589548470187e-05,
      "loss": 0.5403,
      "step": 9149
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5894368886947632,
      "learning_rate": 1.6714834921253384e-05,
      "loss": 0.4455,
      "step": 9150
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.768636703491211,
      "learning_rate": 1.6714080224393012e-05,
      "loss": 0.53,
      "step": 9151
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6251940727233887,
      "learning_rate": 1.6713325457896887e-05,
      "loss": 0.4881,
      "step": 9152
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.861736536026001,
      "learning_rate": 1.6712570621772844e-05,
      "loss": 0.5731,
      "step": 9153
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6171178817749023,
      "learning_rate": 1.6711815716028706e-05,
      "loss": 0.4527,
      "step": 9154
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6376385688781738,
      "learning_rate": 1.671106074067231e-05,
      "loss": 0.5434,
      "step": 9155
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.582701563835144,
      "learning_rate": 1.6710305695711482e-05,
      "loss": 0.4899,
      "step": 9156
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7137237787246704,
      "learning_rate": 1.6709550581154055e-05,
      "loss": 0.5333,
      "step": 9157
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5947911739349365,
      "learning_rate": 1.6708795397007857e-05,
      "loss": 0.4654,
      "step": 9158
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.712204933166504,
      "learning_rate": 1.6708040143280725e-05,
      "loss": 0.4958,
      "step": 9159
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5757477283477783,
      "learning_rate": 1.6707284819980488e-05,
      "loss": 0.454,
      "step": 9160
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7158061265945435,
      "learning_rate": 1.6706529427114984e-05,
      "loss": 0.5238,
      "step": 9161
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6286050081253052,
      "learning_rate": 1.6705773964692042e-05,
      "loss": 0.4713,
      "step": 9162
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6258584260940552,
      "learning_rate": 1.6705018432719503e-05,
      "loss": 0.5307,
      "step": 9163
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5951613187789917,
      "learning_rate": 1.67042628312052e-05,
      "loss": 0.4733,
      "step": 9164
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6049187183380127,
      "learning_rate": 1.6703507160156972e-05,
      "loss": 0.5356,
      "step": 9165
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6277953386306763,
      "learning_rate": 1.6702751419582654e-05,
      "loss": 0.4553,
      "step": 9166
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.125657081604004,
      "learning_rate": 1.670199560949009e-05,
      "loss": 1.0405,
      "step": 9167
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7734049558639526,
      "learning_rate": 1.670123972988711e-05,
      "loss": 1.0309,
      "step": 9168
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6902492046356201,
      "learning_rate": 1.670048378078156e-05,
      "loss": 1.0218,
      "step": 9169
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5257837772369385,
      "learning_rate": 1.669972776218128e-05,
      "loss": 1.0896,
      "step": 9170
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0050971508026123,
      "learning_rate": 1.6698971674094106e-05,
      "loss": 0.5341,
      "step": 9171
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9486521482467651,
      "learning_rate": 1.6698215516527884e-05,
      "loss": 0.5228,
      "step": 9172
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8986855745315552,
      "learning_rate": 1.6697459289490456e-05,
      "loss": 0.9453,
      "step": 9173
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7445148229599,
      "learning_rate": 1.6696702992989668e-05,
      "loss": 1.013,
      "step": 9174
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7273445129394531,
      "learning_rate": 1.669594662703336e-05,
      "loss": 0.5339,
      "step": 9175
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5645921230316162,
      "learning_rate": 1.6695190191629378e-05,
      "loss": 0.4493,
      "step": 9176
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8314499855041504,
      "learning_rate": 1.6694433686785567e-05,
      "loss": 0.5485,
      "step": 9177
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6732407808303833,
      "learning_rate": 1.6693677112509773e-05,
      "loss": 0.4911,
      "step": 9178
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8038896322250366,
      "learning_rate": 1.6692920468809845e-05,
      "loss": 0.6098,
      "step": 9179
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.608020305633545,
      "learning_rate": 1.6692163755693632e-05,
      "loss": 0.4572,
      "step": 9180
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6435935497283936,
      "learning_rate": 1.6691406973168974e-05,
      "loss": 0.534,
      "step": 9181
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7076225280761719,
      "learning_rate": 1.669065012124373e-05,
      "loss": 0.4644,
      "step": 9182
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.916355848312378,
      "learning_rate": 1.6689893199925744e-05,
      "loss": 1.0458,
      "step": 9183
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6239056587219238,
      "learning_rate": 1.668913620922287e-05,
      "loss": 1.0668,
      "step": 9184
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7931396961212158,
      "learning_rate": 1.668837914914295e-05,
      "loss": 0.5544,
      "step": 9185
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.679776906967163,
      "learning_rate": 1.668762201969385e-05,
      "loss": 0.4516,
      "step": 9186
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.626629114151001,
      "learning_rate": 1.6686864820883413e-05,
      "loss": 0.5178,
      "step": 9187
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6280560493469238,
      "learning_rate": 1.6686107552719497e-05,
      "loss": 0.4751,
      "step": 9188
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5871490240097046,
      "learning_rate": 1.668535021520995e-05,
      "loss": 0.5458,
      "step": 9189
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6564019918441772,
      "learning_rate": 1.668459280836263e-05,
      "loss": 0.525,
      "step": 9190
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.705898642539978,
      "learning_rate": 1.66838353321854e-05,
      "loss": 0.5684,
      "step": 9191
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5788146257400513,
      "learning_rate": 1.6683077786686105e-05,
      "loss": 0.4637,
      "step": 9192
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6434950828552246,
      "learning_rate": 1.6682320171872606e-05,
      "loss": 0.5607,
      "step": 9193
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6095024347305298,
      "learning_rate": 1.6681562487752765e-05,
      "loss": 0.4612,
      "step": 9194
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6471176147460938,
      "learning_rate": 1.6680804734334433e-05,
      "loss": 0.5258,
      "step": 9195
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6343002319335938,
      "learning_rate": 1.6680046911625476e-05,
      "loss": 0.4767,
      "step": 9196
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6934926509857178,
      "learning_rate": 1.6679289019633747e-05,
      "loss": 0.5715,
      "step": 9197
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6166123151779175,
      "learning_rate": 1.6678531058367114e-05,
      "loss": 0.4453,
      "step": 9198
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6118475198745728,
      "learning_rate": 1.667777302783343e-05,
      "loss": 0.5175,
      "step": 9199
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.584499716758728,
      "learning_rate": 1.6677014928040564e-05,
      "loss": 0.4908,
      "step": 9200
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6186156272888184,
      "learning_rate": 1.667625675899638e-05,
      "loss": 0.5054,
      "step": 9201
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.680251121520996,
      "learning_rate": 1.667549852070873e-05,
      "loss": 0.4743,
      "step": 9202
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6942604780197144,
      "learning_rate": 1.6674740213185488e-05,
      "loss": 0.5607,
      "step": 9203
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6807503700256348,
      "learning_rate": 1.667398183643452e-05,
      "loss": 0.4925,
      "step": 9204
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.778597116470337,
      "learning_rate": 1.6673223390463684e-05,
      "loss": 0.5677,
      "step": 9205
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6015669107437134,
      "learning_rate": 1.667246487528085e-05,
      "loss": 0.4653,
      "step": 9206
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6262078285217285,
      "learning_rate": 1.667170629089389e-05,
      "loss": 0.5147,
      "step": 9207
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6944596767425537,
      "learning_rate": 1.6670947637310664e-05,
      "loss": 0.4711,
      "step": 9208
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.679169774055481,
      "learning_rate": 1.6670188914539046e-05,
      "loss": 0.5269,
      "step": 9209
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5202136039733887,
      "learning_rate": 1.66694301225869e-05,
      "loss": 0.4374,
      "step": 9210
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6952869892120361,
      "learning_rate": 1.6668671261462102e-05,
      "loss": 0.5116,
      "step": 9211
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.724410057067871,
      "learning_rate": 1.6667912331172514e-05,
      "loss": 0.4754,
      "step": 9212
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.736348271369934,
      "learning_rate": 1.6667153331726017e-05,
      "loss": 0.5785,
      "step": 9213
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7436959743499756,
      "learning_rate": 1.6666394263130477e-05,
      "loss": 0.4655,
      "step": 9214
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8312110900878906,
      "learning_rate": 1.666563512539377e-05,
      "loss": 0.5588,
      "step": 9215
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6252411603927612,
      "learning_rate": 1.6664875918523767e-05,
      "loss": 0.4512,
      "step": 9216
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0818066596984863,
      "learning_rate": 1.6664116642528344e-05,
      "loss": 1.0351,
      "step": 9217
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7892348766326904,
      "learning_rate": 1.666335729741537e-05,
      "loss": 1.0897,
      "step": 9218
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7290409803390503,
      "learning_rate": 1.6662597883192733e-05,
      "loss": 0.4865,
      "step": 9219
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.772020936012268,
      "learning_rate": 1.6661838399868294e-05,
      "loss": 0.5075,
      "step": 9220
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.747696042060852,
      "learning_rate": 1.6661078847449944e-05,
      "loss": 0.5597,
      "step": 9221
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.669683814048767,
      "learning_rate": 1.6660319225945553e-05,
      "loss": 0.4351,
      "step": 9222
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.768661379814148,
      "learning_rate": 1.6659559535363e-05,
      "loss": 0.5027,
      "step": 9223
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.17505407333374,
      "learning_rate": 1.6658799775710165e-05,
      "loss": 0.5062,
      "step": 9224
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6323232650756836,
      "learning_rate": 1.665803994699493e-05,
      "loss": 0.49,
      "step": 9225
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8842233419418335,
      "learning_rate": 1.6657280049225177e-05,
      "loss": 0.5226,
      "step": 9226
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6134706735610962,
      "learning_rate": 1.665652008240878e-05,
      "loss": 0.5442,
      "step": 9227
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5195931196212769,
      "learning_rate": 1.6655760046553627e-05,
      "loss": 0.4685,
      "step": 9228
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.704007625579834,
      "learning_rate": 1.6654999941667598e-05,
      "loss": 0.5316,
      "step": 9229
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6241483688354492,
      "learning_rate": 1.665423976775858e-05,
      "loss": 0.4601,
      "step": 9230
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6887030601501465,
      "learning_rate": 1.6653479524834453e-05,
      "loss": 0.5317,
      "step": 9231
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6524384021759033,
      "learning_rate": 1.6652719212903103e-05,
      "loss": 0.49,
      "step": 9232
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.634545922279358,
      "learning_rate": 1.6651958831972418e-05,
      "loss": 0.5571,
      "step": 9233
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6702960729599,
      "learning_rate": 1.6651198382050283e-05,
      "loss": 0.4525,
      "step": 9234
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7052876949310303,
      "learning_rate": 1.6650437863144585e-05,
      "loss": 0.5342,
      "step": 9235
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7634670734405518,
      "learning_rate": 1.664967727526321e-05,
      "loss": 0.5037,
      "step": 9236
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.71726393699646,
      "learning_rate": 1.664891661841405e-05,
      "loss": 0.5307,
      "step": 9237
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.842914342880249,
      "learning_rate": 1.6648155892604995e-05,
      "loss": 0.477,
      "step": 9238
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.611464023590088,
      "learning_rate": 1.664739509784393e-05,
      "loss": 0.5068,
      "step": 9239
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6520452499389648,
      "learning_rate": 1.6646634234138745e-05,
      "loss": 0.5002,
      "step": 9240
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6619094610214233,
      "learning_rate": 1.664587330149734e-05,
      "loss": 0.5589,
      "step": 9241
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6762138605117798,
      "learning_rate": 1.6645112299927598e-05,
      "loss": 0.5138,
      "step": 9242
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5445548295974731,
      "learning_rate": 1.6644351229437416e-05,
      "loss": 0.4804,
      "step": 9243
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6482653617858887,
      "learning_rate": 1.6643590090034686e-05,
      "loss": 0.4693,
      "step": 9244
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7148524522781372,
      "learning_rate": 1.6642828881727308e-05,
      "loss": 0.5286,
      "step": 9245
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.610556721687317,
      "learning_rate": 1.6642067604523163e-05,
      "loss": 0.4744,
      "step": 9246
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6325359344482422,
      "learning_rate": 1.6641306258430165e-05,
      "loss": 0.5207,
      "step": 9247
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5229308605194092,
      "learning_rate": 1.6640544843456196e-05,
      "loss": 0.4755,
      "step": 9248
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6028398275375366,
      "learning_rate": 1.6639783359609162e-05,
      "loss": 0.4946,
      "step": 9249
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7694611549377441,
      "learning_rate": 1.663902180689695e-05,
      "loss": 0.4939,
      "step": 9250
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.184929132461548,
      "learning_rate": 1.6638260185327472e-05,
      "loss": 1.0531,
      "step": 9251
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8606648445129395,
      "learning_rate": 1.663749849490862e-05,
      "loss": 1.059,
      "step": 9252
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8808549642562866,
      "learning_rate": 1.6636736735648292e-05,
      "loss": 0.543,
      "step": 9253
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7970020771026611,
      "learning_rate": 1.6635974907554393e-05,
      "loss": 0.4731,
      "step": 9254
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.761526107788086,
      "learning_rate": 1.6635213010634826e-05,
      "loss": 0.5245,
      "step": 9255
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.647426962852478,
      "learning_rate": 1.6634451044897484e-05,
      "loss": 0.4963,
      "step": 9256
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5990008115768433,
      "learning_rate": 1.6633689010350278e-05,
      "loss": 0.524,
      "step": 9257
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6516499519348145,
      "learning_rate": 1.6632926907001107e-05,
      "loss": 0.4599,
      "step": 9258
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.673192024230957,
      "learning_rate": 1.663216473485788e-05,
      "loss": 0.5325,
      "step": 9259
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6450080871582031,
      "learning_rate": 1.66314024939285e-05,
      "loss": 0.5069,
      "step": 9260
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7533601522445679,
      "learning_rate": 1.6630640184220873e-05,
      "loss": 0.5168,
      "step": 9261
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5431839227676392,
      "learning_rate": 1.66298778057429e-05,
      "loss": 0.4525,
      "step": 9262
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.63210129737854,
      "learning_rate": 1.6629115358502494e-05,
      "loss": 0.5113,
      "step": 9263
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6132690906524658,
      "learning_rate": 1.6628352842507562e-05,
      "loss": 0.5151,
      "step": 9264
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7060247659683228,
      "learning_rate": 1.6627590257766013e-05,
      "loss": 0.5259,
      "step": 9265
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5974774360656738,
      "learning_rate": 1.6626827604285755e-05,
      "loss": 0.4845,
      "step": 9266
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.228729486465454,
      "learning_rate": 1.6626064882074695e-05,
      "loss": 1.0676,
      "step": 9267
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8521630764007568,
      "learning_rate": 1.662530209114075e-05,
      "loss": 1.0486,
      "step": 9268
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.12595272064209,
      "learning_rate": 1.6624539231491824e-05,
      "loss": 0.6116,
      "step": 9269
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7463115453720093,
      "learning_rate": 1.6623776303135835e-05,
      "loss": 0.4689,
      "step": 9270
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7127437591552734,
      "learning_rate": 1.6623013306080694e-05,
      "loss": 0.4961,
      "step": 9271
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6510744094848633,
      "learning_rate": 1.6622250240334315e-05,
      "loss": 0.4949,
      "step": 9272
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5907517671585083,
      "learning_rate": 1.662148710590461e-05,
      "loss": 0.5306,
      "step": 9273
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6953154802322388,
      "learning_rate": 1.6620723902799497e-05,
      "loss": 0.4869,
      "step": 9274
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6353085041046143,
      "learning_rate": 1.661996063102689e-05,
      "loss": 0.4909,
      "step": 9275
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7423371076583862,
      "learning_rate": 1.6619197290594704e-05,
      "loss": 0.4855,
      "step": 9276
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6934572458267212,
      "learning_rate": 1.6618433881510858e-05,
      "loss": 0.5218,
      "step": 9277
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.550036907196045,
      "learning_rate": 1.661767040378327e-05,
      "loss": 0.4585,
      "step": 9278
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6530719995498657,
      "learning_rate": 1.6616906857419858e-05,
      "loss": 0.5423,
      "step": 9279
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6925042867660522,
      "learning_rate": 1.6616143242428537e-05,
      "loss": 0.4956,
      "step": 9280
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.793178915977478,
      "learning_rate": 1.6615379558817233e-05,
      "loss": 0.5637,
      "step": 9281
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5582154989242554,
      "learning_rate": 1.6614615806593866e-05,
      "loss": 0.4352,
      "step": 9282
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6141798496246338,
      "learning_rate": 1.6613851985766352e-05,
      "loss": 0.5378,
      "step": 9283
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5621321201324463,
      "learning_rate": 1.661308809634262e-05,
      "loss": 0.444,
      "step": 9284
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7027900218963623,
      "learning_rate": 1.661232413833059e-05,
      "loss": 0.5509,
      "step": 9285
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6262948513031006,
      "learning_rate": 1.6611560111738187e-05,
      "loss": 0.4701,
      "step": 9286
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7087293863296509,
      "learning_rate": 1.661079601657333e-05,
      "loss": 0.5204,
      "step": 9287
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6196181774139404,
      "learning_rate": 1.6610031852843946e-05,
      "loss": 0.4783,
      "step": 9288
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5955837965011597,
      "learning_rate": 1.6609267620557964e-05,
      "loss": 0.501,
      "step": 9289
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.68942391872406,
      "learning_rate": 1.660850331972331e-05,
      "loss": 0.4771,
      "step": 9290
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5851436853408813,
      "learning_rate": 1.6607738950347908e-05,
      "loss": 0.5227,
      "step": 9291
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.600723147392273,
      "learning_rate": 1.660697451243968e-05,
      "loss": 0.4691,
      "step": 9292
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5984331369400024,
      "learning_rate": 1.660621000600657e-05,
      "loss": 0.4712,
      "step": 9293
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5752211809158325,
      "learning_rate": 1.660544543105649e-05,
      "loss": 0.49,
      "step": 9294
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7252558469772339,
      "learning_rate": 1.6604680787597386e-05,
      "loss": 0.5856,
      "step": 9295
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6681582927703857,
      "learning_rate": 1.660391607563718e-05,
      "loss": 0.4659,
      "step": 9296
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.54177725315094,
      "learning_rate": 1.66031512951838e-05,
      "loss": 0.4994,
      "step": 9297
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5445659160614014,
      "learning_rate": 1.6602386446245184e-05,
      "loss": 0.4847,
      "step": 9298
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5448976755142212,
      "learning_rate": 1.6601621528829264e-05,
      "loss": 0.4932,
      "step": 9299
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.626384973526001,
      "learning_rate": 1.660085654294397e-05,
      "loss": 0.4521,
      "step": 9300
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7057114839553833,
      "learning_rate": 1.660009148859724e-05,
      "loss": 0.5965,
      "step": 9301
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.591545820236206,
      "learning_rate": 1.6599326365797006e-05,
      "loss": 0.4898,
      "step": 9302
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7209796905517578,
      "learning_rate": 1.6598561174551207e-05,
      "loss": 0.5564,
      "step": 9303
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6009732484817505,
      "learning_rate": 1.6597795914867772e-05,
      "loss": 0.4629,
      "step": 9304
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6545372009277344,
      "learning_rate": 1.659703058675465e-05,
      "loss": 0.5011,
      "step": 9305
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.547104835510254,
      "learning_rate": 1.6596265190219768e-05,
      "loss": 0.4601,
      "step": 9306
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6350103616714478,
      "learning_rate": 1.6595499725271068e-05,
      "loss": 0.5334,
      "step": 9307
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6924740076065063,
      "learning_rate": 1.659473419191649e-05,
      "loss": 0.5608,
      "step": 9308
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.642388105392456,
      "learning_rate": 1.6593968590163968e-05,
      "loss": 0.504,
      "step": 9309
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.675340175628662,
      "learning_rate": 1.6593202920021456e-05,
      "loss": 0.5332,
      "step": 9310
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5894845724105835,
      "learning_rate": 1.659243718149688e-05,
      "loss": 0.5386,
      "step": 9311
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5695525407791138,
      "learning_rate": 1.659167137459819e-05,
      "loss": 0.4694,
      "step": 9312
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6486022472381592,
      "learning_rate": 1.6590905499333327e-05,
      "loss": 0.518,
      "step": 9313
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.701745867729187,
      "learning_rate": 1.6590139555710235e-05,
      "loss": 0.4688,
      "step": 9314
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7092920541763306,
      "learning_rate": 1.658937354373686e-05,
      "loss": 0.6195,
      "step": 9315
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6326637268066406,
      "learning_rate": 1.6588607463421143e-05,
      "loss": 0.4576,
      "step": 9316
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6213195323944092,
      "learning_rate": 1.658784131477103e-05,
      "loss": 0.5376,
      "step": 9317
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6971673965454102,
      "learning_rate": 1.6587075097794468e-05,
      "loss": 0.4722,
      "step": 9318
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8449078798294067,
      "learning_rate": 1.6586308812499403e-05,
      "loss": 0.5841,
      "step": 9319
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.568735122680664,
      "learning_rate": 1.6585542458893784e-05,
      "loss": 0.4421,
      "step": 9320
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6043776273727417,
      "learning_rate": 1.658477603698556e-05,
      "loss": 0.5203,
      "step": 9321
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6066138744354248,
      "learning_rate": 1.658400954678268e-05,
      "loss": 0.4465,
      "step": 9322
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7179757356643677,
      "learning_rate": 1.6583242988293095e-05,
      "loss": 0.5769,
      "step": 9323
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5908207893371582,
      "learning_rate": 1.6582476361524745e-05,
      "loss": 0.481,
      "step": 9324
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5700972080230713,
      "learning_rate": 1.6581709666485593e-05,
      "loss": 0.5011,
      "step": 9325
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.61664617061615,
      "learning_rate": 1.658094290318359e-05,
      "loss": 0.4662,
      "step": 9326
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.623643398284912,
      "learning_rate": 1.6580176071626683e-05,
      "loss": 0.5287,
      "step": 9327
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6068307161331177,
      "learning_rate": 1.657940917182283e-05,
      "loss": 0.4697,
      "step": 9328
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6836626529693604,
      "learning_rate": 1.6578642203779978e-05,
      "loss": 0.5577,
      "step": 9329
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6394295692443848,
      "learning_rate": 1.657787516750609e-05,
      "loss": 0.4619,
      "step": 9330
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6717212200164795,
      "learning_rate": 1.657710806300912e-05,
      "loss": 0.5263,
      "step": 9331
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.726442575454712,
      "learning_rate": 1.6576340890297017e-05,
      "loss": 0.5502,
      "step": 9332
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6387503147125244,
      "learning_rate": 1.657557364937775e-05,
      "loss": 0.535,
      "step": 9333
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6849044561386108,
      "learning_rate": 1.6574806340259263e-05,
      "loss": 0.4822,
      "step": 9334
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.679418921470642,
      "learning_rate": 1.6574038962949523e-05,
      "loss": 0.5856,
      "step": 9335
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7763941287994385,
      "learning_rate": 1.6573271517456487e-05,
      "loss": 0.4641,
      "step": 9336
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6835001707077026,
      "learning_rate": 1.657250400378811e-05,
      "loss": 0.5012,
      "step": 9337
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5799905061721802,
      "learning_rate": 1.657173642195236e-05,
      "loss": 0.4292,
      "step": 9338
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5917166471481323,
      "learning_rate": 1.6570968771957195e-05,
      "loss": 0.5165,
      "step": 9339
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6762866973876953,
      "learning_rate": 1.6570201053810575e-05,
      "loss": 0.5458,
      "step": 9340
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6868747472763062,
      "learning_rate": 1.6569433267520467e-05,
      "loss": 0.5504,
      "step": 9341
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5980346202850342,
      "learning_rate": 1.6568665413094828e-05,
      "loss": 0.4745,
      "step": 9342
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6590760946273804,
      "learning_rate": 1.6567897490541628e-05,
      "loss": 0.5902,
      "step": 9343
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6023741960525513,
      "learning_rate": 1.6567129499868825e-05,
      "loss": 0.4635,
      "step": 9344
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7048346996307373,
      "learning_rate": 1.6566361441084385e-05,
      "loss": 0.5723,
      "step": 9345
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6333187818527222,
      "learning_rate": 1.656559331419628e-05,
      "loss": 0.4661,
      "step": 9346
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5980128049850464,
      "learning_rate": 1.6564825119212477e-05,
      "loss": 0.5282,
      "step": 9347
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5679892301559448,
      "learning_rate": 1.6564056856140932e-05,
      "loss": 0.4341,
      "step": 9348
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6432297229766846,
      "learning_rate": 1.656328852498963e-05,
      "loss": 0.5439,
      "step": 9349
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7168158292770386,
      "learning_rate": 1.6562520125766524e-05,
      "loss": 0.4969,
      "step": 9350
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7340137958526611,
      "learning_rate": 1.6561751658479592e-05,
      "loss": 0.5224,
      "step": 9351
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6598670482635498,
      "learning_rate": 1.65609831231368e-05,
      "loss": 0.4794,
      "step": 9352
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7717212438583374,
      "learning_rate": 1.656021451974613e-05,
      "loss": 0.5683,
      "step": 9353
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8233191967010498,
      "learning_rate": 1.6559445848315537e-05,
      "loss": 0.4803,
      "step": 9354
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6494276523590088,
      "learning_rate": 1.6558677108853002e-05,
      "loss": 0.5211,
      "step": 9355
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.718879222869873,
      "learning_rate": 1.65579083013665e-05,
      "loss": 0.5741,
      "step": 9356
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.712484359741211,
      "learning_rate": 1.6557139425864003e-05,
      "loss": 0.5562,
      "step": 9357
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7021480798721313,
      "learning_rate": 1.6556370482353484e-05,
      "loss": 0.4844,
      "step": 9358
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.650793194770813,
      "learning_rate": 1.655560147084292e-05,
      "loss": 0.5405,
      "step": 9359
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6209828853607178,
      "learning_rate": 1.655483239134028e-05,
      "loss": 0.4615,
      "step": 9360
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6954922676086426,
      "learning_rate": 1.6554063243853552e-05,
      "loss": 0.5466,
      "step": 9361
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6447478532791138,
      "learning_rate": 1.6553294028390708e-05,
      "loss": 0.4422,
      "step": 9362
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.697481393814087,
      "learning_rate": 1.6552524744959724e-05,
      "loss": 0.5479,
      "step": 9363
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.567551612854004,
      "learning_rate": 1.655175539356858e-05,
      "loss": 0.4647,
      "step": 9364
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6707442998886108,
      "learning_rate": 1.6550985974225255e-05,
      "loss": 0.5463,
      "step": 9365
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5390788316726685,
      "learning_rate": 1.655021648693773e-05,
      "loss": 0.4462,
      "step": 9366
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5190351009368896,
      "learning_rate": 1.654944693171399e-05,
      "loss": 0.52,
      "step": 9367
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5545281171798706,
      "learning_rate": 1.654867730856201e-05,
      "loss": 0.4604,
      "step": 9368
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9896924495697021,
      "learning_rate": 1.654790761748977e-05,
      "loss": 1.0275,
      "step": 9369
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6704634428024292,
      "learning_rate": 1.6547137858505264e-05,
      "loss": 1.0255,
      "step": 9370
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7385773658752441,
      "learning_rate": 1.6546368031616464e-05,
      "loss": 0.5748,
      "step": 9371
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.667651653289795,
      "learning_rate": 1.654559813683136e-05,
      "loss": 0.4499,
      "step": 9372
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6639350652694702,
      "learning_rate": 1.6544828174157937e-05,
      "loss": 0.5374,
      "step": 9373
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7151304483413696,
      "learning_rate": 1.6544058143604182e-05,
      "loss": 0.4949,
      "step": 9374
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6277427673339844,
      "learning_rate": 1.654328804517808e-05,
      "loss": 0.4968,
      "step": 9375
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5247639417648315,
      "learning_rate": 1.6542517878887613e-05,
      "loss": 0.464,
      "step": 9376
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8372750282287598,
      "learning_rate": 1.6541747644740778e-05,
      "loss": 0.5532,
      "step": 9377
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5995714664459229,
      "learning_rate": 1.6540977342745557e-05,
      "loss": 0.4565,
      "step": 9378
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.998659610748291,
      "learning_rate": 1.6540206972909945e-05,
      "loss": 0.9805,
      "step": 9379
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6137605905532837,
      "learning_rate": 1.6539436535241923e-05,
      "loss": 1.0597,
      "step": 9380
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7896511554718018,
      "learning_rate": 1.653866602974949e-05,
      "loss": 0.585,
      "step": 9381
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7134430408477783,
      "learning_rate": 1.6537895456440637e-05,
      "loss": 0.474,
      "step": 9382
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7162202596664429,
      "learning_rate": 1.6537124815323348e-05,
      "loss": 0.5433,
      "step": 9383
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6047048568725586,
      "learning_rate": 1.6536354106405625e-05,
      "loss": 0.4839,
      "step": 9384
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5747320652008057,
      "learning_rate": 1.6535583329695457e-05,
      "loss": 0.5189,
      "step": 9385
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6311535835266113,
      "learning_rate": 1.6534812485200834e-05,
      "loss": 0.4361,
      "step": 9386
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7081116437911987,
      "learning_rate": 1.6534041572929762e-05,
      "loss": 0.5543,
      "step": 9387
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6473362445831299,
      "learning_rate": 1.6533270592890226e-05,
      "loss": 0.4595,
      "step": 9388
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7364892959594727,
      "learning_rate": 1.6532499545090234e-05,
      "loss": 0.5116,
      "step": 9389
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5650829076766968,
      "learning_rate": 1.6531728429537766e-05,
      "loss": 0.4576,
      "step": 9390
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6396379470825195,
      "learning_rate": 1.6530957246240837e-05,
      "loss": 0.5414,
      "step": 9391
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6355441808700562,
      "learning_rate": 1.6530185995207438e-05,
      "loss": 0.4909,
      "step": 9392
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6201841831207275,
      "learning_rate": 1.6529414676445563e-05,
      "loss": 0.5046,
      "step": 9393
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6466305255889893,
      "learning_rate": 1.652864328996322e-05,
      "loss": 0.4798,
      "step": 9394
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.646087884902954,
      "learning_rate": 1.6527871835768404e-05,
      "loss": 0.5039,
      "step": 9395
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.516500473022461,
      "learning_rate": 1.652710031386912e-05,
      "loss": 0.4579,
      "step": 9396
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7552194595336914,
      "learning_rate": 1.6526328724273368e-05,
      "loss": 0.5235,
      "step": 9397
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.694946527481079,
      "learning_rate": 1.652555706698915e-05,
      "loss": 0.4488,
      "step": 9398
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.675594687461853,
      "learning_rate": 1.652478534202447e-05,
      "loss": 0.5372,
      "step": 9399
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6269875764846802,
      "learning_rate": 1.652401354938734e-05,
      "loss": 0.4372,
      "step": 9400
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6880736351013184,
      "learning_rate": 1.6523241689085748e-05,
      "loss": 0.5344,
      "step": 9401
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6650944948196411,
      "learning_rate": 1.652246976112771e-05,
      "loss": 0.4565,
      "step": 9402
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5685054063796997,
      "learning_rate": 1.6521697765521232e-05,
      "loss": 0.5895,
      "step": 9403
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6301906108856201,
      "learning_rate": 1.6520925702274323e-05,
      "loss": 0.469,
      "step": 9404
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8414968252182007,
      "learning_rate": 1.652015357139498e-05,
      "loss": 0.5541,
      "step": 9405
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7316710948944092,
      "learning_rate": 1.6519381372891223e-05,
      "loss": 0.4475,
      "step": 9406
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8904937505722046,
      "learning_rate": 1.6518609106771054e-05,
      "loss": 1.0303,
      "step": 9407
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5964056253433228,
      "learning_rate": 1.6517836773042486e-05,
      "loss": 0.9683,
      "step": 9408
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.759153127670288,
      "learning_rate": 1.6517064371713527e-05,
      "loss": 0.5437,
      "step": 9409
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6531291007995605,
      "learning_rate": 1.651629190279219e-05,
      "loss": 0.466,
      "step": 9410
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7018442153930664,
      "learning_rate": 1.6515519366286486e-05,
      "loss": 0.489,
      "step": 9411
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7341341972351074,
      "learning_rate": 1.651474676220443e-05,
      "loss": 0.47,
      "step": 9412
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6657428741455078,
      "learning_rate": 1.6513974090554026e-05,
      "loss": 0.4976,
      "step": 9413
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.70192551612854,
      "learning_rate": 1.65132013513433e-05,
      "loss": 0.494,
      "step": 9414
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6107451915740967,
      "learning_rate": 1.6512428544580258e-05,
      "loss": 0.5074,
      "step": 9415
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.705358862876892,
      "learning_rate": 1.6511655670272916e-05,
      "loss": 0.5449,
      "step": 9416
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6202837228775024,
      "learning_rate": 1.6510882728429296e-05,
      "loss": 0.5078,
      "step": 9417
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6680243015289307,
      "learning_rate": 1.6510109719057407e-05,
      "loss": 0.4861,
      "step": 9418
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.784709095954895,
      "learning_rate": 1.6509336642165276e-05,
      "loss": 0.6208,
      "step": 9419
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6483696699142456,
      "learning_rate": 1.650856349776091e-05,
      "loss": 0.4672,
      "step": 9420
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1365737915039062,
      "learning_rate": 1.6507790285852333e-05,
      "loss": 1.0218,
      "step": 9421
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6336958408355713,
      "learning_rate": 1.6507017006447567e-05,
      "loss": 1.0135,
      "step": 9422
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9579068422317505,
      "learning_rate": 1.6506243659554628e-05,
      "loss": 1.0251,
      "step": 9423
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5828789472579956,
      "learning_rate": 1.6505470245181538e-05,
      "loss": 1.047,
      "step": 9424
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.852946162223816,
      "learning_rate": 1.6504696763336318e-05,
      "loss": 0.5404,
      "step": 9425
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8439773321151733,
      "learning_rate": 1.6503923214026993e-05,
      "loss": 0.4816,
      "step": 9426
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7373239994049072,
      "learning_rate": 1.650314959726158e-05,
      "loss": 0.6107,
      "step": 9427
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.644690752029419,
      "learning_rate": 1.650237591304811e-05,
      "loss": 0.483,
      "step": 9428
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6430151462554932,
      "learning_rate": 1.6501602161394607e-05,
      "loss": 0.4956,
      "step": 9429
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7063590288162231,
      "learning_rate": 1.650082834230909e-05,
      "loss": 0.4983,
      "step": 9430
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6621710062026978,
      "learning_rate": 1.6500054455799585e-05,
      "loss": 0.5197,
      "step": 9431
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.730332374572754,
      "learning_rate": 1.6499280501874128e-05,
      "loss": 0.4452,
      "step": 9432
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7045814990997314,
      "learning_rate": 1.6498506480540732e-05,
      "loss": 0.6734,
      "step": 9433
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.707870602607727,
      "learning_rate": 1.6497732391807437e-05,
      "loss": 0.5204,
      "step": 9434
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.834444522857666,
      "learning_rate": 1.6496958235682262e-05,
      "loss": 0.5035,
      "step": 9435
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6462464332580566,
      "learning_rate": 1.6496184012173245e-05,
      "loss": 0.4667,
      "step": 9436
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6454981565475464,
      "learning_rate": 1.649540972128841e-05,
      "loss": 0.5261,
      "step": 9437
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.647355318069458,
      "learning_rate": 1.649463536303579e-05,
      "loss": 0.4565,
      "step": 9438
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6729888916015625,
      "learning_rate": 1.6493860937423413e-05,
      "loss": 0.4897,
      "step": 9439
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6698325872421265,
      "learning_rate": 1.6493086444459317e-05,
      "loss": 0.464,
      "step": 9440
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7322971820831299,
      "learning_rate": 1.649231188415153e-05,
      "loss": 0.5099,
      "step": 9441
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6563129425048828,
      "learning_rate": 1.6491537256508093e-05,
      "loss": 0.4483,
      "step": 9442
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6336801052093506,
      "learning_rate": 1.6490762561537027e-05,
      "loss": 0.5952,
      "step": 9443
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5629688501358032,
      "learning_rate": 1.6489987799246375e-05,
      "loss": 0.4593,
      "step": 9444
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6999289989471436,
      "learning_rate": 1.6489212969644176e-05,
      "loss": 0.5632,
      "step": 9445
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6106253862380981,
      "learning_rate": 1.648843807273846e-05,
      "loss": 0.4668,
      "step": 9446
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5889490842819214,
      "learning_rate": 1.648766310853726e-05,
      "loss": 0.498,
      "step": 9447
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6760447025299072,
      "learning_rate": 1.6486888077048628e-05,
      "loss": 0.4972,
      "step": 9448
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7284337282180786,
      "learning_rate": 1.6486112978280585e-05,
      "loss": 0.5513,
      "step": 9449
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5887707471847534,
      "learning_rate": 1.6485337812241185e-05,
      "loss": 0.4917,
      "step": 9450
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8960773944854736,
      "learning_rate": 1.648456257893846e-05,
      "loss": 0.535,
      "step": 9451
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6453534364700317,
      "learning_rate": 1.648378727838045e-05,
      "loss": 0.4642,
      "step": 9452
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8132396936416626,
      "learning_rate": 1.64830119105752e-05,
      "loss": 0.5389,
      "step": 9453
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5677047967910767,
      "learning_rate": 1.6482236475530746e-05,
      "loss": 0.4586,
      "step": 9454
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7131626605987549,
      "learning_rate": 1.6481460973255136e-05,
      "loss": 0.6224,
      "step": 9455
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6065846681594849,
      "learning_rate": 1.6480685403756413e-05,
      "loss": 0.4727,
      "step": 9456
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6889928579330444,
      "learning_rate": 1.6479909767042618e-05,
      "loss": 0.5516,
      "step": 9457
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6370961666107178,
      "learning_rate": 1.6479134063121796e-05,
      "loss": 0.4852,
      "step": 9458
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.3446948528289795,
      "learning_rate": 1.6478358292001994e-05,
      "loss": 1.0514,
      "step": 9459
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1569325923919678,
      "learning_rate": 1.6477582453691257e-05,
      "loss": 1.1124,
      "step": 9460
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.894230604171753,
      "learning_rate": 1.6476806548197634e-05,
      "loss": 0.5974,
      "step": 9461
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8955481052398682,
      "learning_rate": 1.6476030575529165e-05,
      "loss": 0.4648,
      "step": 9462
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6298408508300781,
      "learning_rate": 1.6475254535693904e-05,
      "loss": 0.5373,
      "step": 9463
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.737594723701477,
      "learning_rate": 1.6474478428699905e-05,
      "loss": 0.5193,
      "step": 9464
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5906736850738525,
      "learning_rate": 1.6473702254555207e-05,
      "loss": 0.526,
      "step": 9465
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.763759970664978,
      "learning_rate": 1.6472926013267864e-05,
      "loss": 0.5439,
      "step": 9466
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7252229452133179,
      "learning_rate": 1.647214970484593e-05,
      "loss": 0.5481,
      "step": 9467
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7474123239517212,
      "learning_rate": 1.647137332929745e-05,
      "loss": 0.5135,
      "step": 9468
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6366685628890991,
      "learning_rate": 1.6470596886630484e-05,
      "loss": 0.5117,
      "step": 9469
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5026315450668335,
      "learning_rate": 1.6469820376853084e-05,
      "loss": 0.4467,
      "step": 9470
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6952955722808838,
      "learning_rate": 1.6469043799973293e-05,
      "loss": 0.5297,
      "step": 9471
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.652364730834961,
      "learning_rate": 1.646826715599918e-05,
      "loss": 0.4851,
      "step": 9472
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8751353025436401,
      "learning_rate": 1.646749044493879e-05,
      "loss": 0.6094,
      "step": 9473
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5763260126113892,
      "learning_rate": 1.6466713666800184e-05,
      "loss": 0.4314,
      "step": 9474
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6045689582824707,
      "learning_rate": 1.6465936821591414e-05,
      "loss": 0.5548,
      "step": 9475
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6269351243972778,
      "learning_rate": 1.6465159909320542e-05,
      "loss": 0.48,
      "step": 9476
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5853315591812134,
      "learning_rate": 1.646438292999562e-05,
      "loss": 0.5995,
      "step": 9477
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.590741515159607,
      "learning_rate": 1.6463605883624713e-05,
      "loss": 0.4799,
      "step": 9478
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6334863901138306,
      "learning_rate": 1.6462828770215876e-05,
      "loss": 0.4966,
      "step": 9479
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9099938869476318,
      "learning_rate": 1.6462051589777175e-05,
      "loss": 0.4648,
      "step": 9480
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7645152807235718,
      "learning_rate": 1.6461274342316656e-05,
      "loss": 0.5453,
      "step": 9481
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5558996200561523,
      "learning_rate": 1.6460497027842398e-05,
      "loss": 0.4959,
      "step": 9482
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5830398797988892,
      "learning_rate": 1.645971964636245e-05,
      "loss": 0.5564,
      "step": 9483
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.672990322113037,
      "learning_rate": 1.6458942197884885e-05,
      "loss": 0.5092,
      "step": 9484
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7083687782287598,
      "learning_rate": 1.6458164682417758e-05,
      "loss": 0.5979,
      "step": 9485
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6291956901550293,
      "learning_rate": 1.6457387099969136e-05,
      "loss": 0.4585,
      "step": 9486
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6314010620117188,
      "learning_rate": 1.6456609450547083e-05,
      "loss": 0.575,
      "step": 9487
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5366549491882324,
      "learning_rate": 1.6455831734159665e-05,
      "loss": 0.4683,
      "step": 9488
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9977222681045532,
      "learning_rate": 1.6455053950814954e-05,
      "loss": 0.4885,
      "step": 9489
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5975358486175537,
      "learning_rate": 1.6454276100521002e-05,
      "loss": 0.4746,
      "step": 9490
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6931077241897583,
      "learning_rate": 1.6453498183285894e-05,
      "loss": 0.5392,
      "step": 9491
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.580763339996338,
      "learning_rate": 1.6452720199117687e-05,
      "loss": 0.4522,
      "step": 9492
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6979140043258667,
      "learning_rate": 1.6451942148024455e-05,
      "loss": 0.499,
      "step": 9493
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5870146751403809,
      "learning_rate": 1.645116403001426e-05,
      "loss": 0.4698,
      "step": 9494
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6732258796691895,
      "learning_rate": 1.6450385845095183e-05,
      "loss": 0.6021,
      "step": 9495
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6361042261123657,
      "learning_rate": 1.644960759327529e-05,
      "loss": 0.4622,
      "step": 9496
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9482762813568115,
      "learning_rate": 1.6448829274562654e-05,
      "loss": 0.5862,
      "step": 9497
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6100993156433105,
      "learning_rate": 1.6448050888965343e-05,
      "loss": 0.4421,
      "step": 9498
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6038018465042114,
      "learning_rate": 1.6447272436491434e-05,
      "loss": 0.5641,
      "step": 9499
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6801438331604004,
      "learning_rate": 1.6446493917149e-05,
      "loss": 0.4708,
      "step": 9500
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6624794006347656,
      "learning_rate": 1.6445715330946118e-05,
      "loss": 0.5261,
      "step": 9501
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7214607000350952,
      "learning_rate": 1.644493667789086e-05,
      "loss": 0.5072,
      "step": 9502
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6288340091705322,
      "learning_rate": 1.6444157957991302e-05,
      "loss": 0.5029,
      "step": 9503
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6693041324615479,
      "learning_rate": 1.6443379171255522e-05,
      "loss": 0.5066,
      "step": 9504
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.604386568069458,
      "learning_rate": 1.6442600317691593e-05,
      "loss": 0.5434,
      "step": 9505
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.568540096282959,
      "learning_rate": 1.64418213973076e-05,
      "loss": 0.4616,
      "step": 9506
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.738578200340271,
      "learning_rate": 1.6441042410111617e-05,
      "loss": 0.4998,
      "step": 9507
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6790040731430054,
      "learning_rate": 1.6440263356111724e-05,
      "loss": 0.5009,
      "step": 9508
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5712133646011353,
      "learning_rate": 1.6439484235316002e-05,
      "loss": 0.521,
      "step": 9509
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.60373055934906,
      "learning_rate": 1.6438705047732532e-05,
      "loss": 0.4866,
      "step": 9510
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.66226327419281,
      "learning_rate": 1.6437925793369394e-05,
      "loss": 0.5135,
      "step": 9511
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6393494606018066,
      "learning_rate": 1.643714647223467e-05,
      "loss": 0.4987,
      "step": 9512
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6598212718963623,
      "learning_rate": 1.6436367084336445e-05,
      "loss": 0.5536,
      "step": 9513
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.558632254600525,
      "learning_rate": 1.6435587629682802e-05,
      "loss": 0.4405,
      "step": 9514
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.673513412475586,
      "learning_rate": 1.643480810828182e-05,
      "loss": 0.5552,
      "step": 9515
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5771496295928955,
      "learning_rate": 1.6434028520141593e-05,
      "loss": 0.4391,
      "step": 9516
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6273466348648071,
      "learning_rate": 1.64332488652702e-05,
      "loss": 0.5332,
      "step": 9517
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5998069047927856,
      "learning_rate": 1.6432469143675725e-05,
      "loss": 0.4826,
      "step": 9518
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5306036472320557,
      "learning_rate": 1.6431689355366263e-05,
      "loss": 0.4906,
      "step": 9519
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.683928370475769,
      "learning_rate": 1.6430909500349897e-05,
      "loss": 0.5116,
      "step": 9520
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.560678482055664,
      "learning_rate": 1.6430129578634718e-05,
      "loss": 0.5237,
      "step": 9521
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7389006614685059,
      "learning_rate": 1.642934959022881e-05,
      "loss": 0.4901,
      "step": 9522
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.070769786834717,
      "learning_rate": 1.6428569535140267e-05,
      "loss": 1.0007,
      "step": 9523
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7115975618362427,
      "learning_rate": 1.6427789413377176e-05,
      "loss": 1.063,
      "step": 9524
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6785634756088257,
      "learning_rate": 1.6427009224947636e-05,
      "loss": 0.5776,
      "step": 9525
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7387491464614868,
      "learning_rate": 1.6426228969859726e-05,
      "loss": 0.4549,
      "step": 9526
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.655333399772644,
      "learning_rate": 1.642544864812155e-05,
      "loss": 0.5652,
      "step": 9527
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.699262261390686,
      "learning_rate": 1.6424668259741194e-05,
      "loss": 0.4663,
      "step": 9528
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.654621958732605,
      "learning_rate": 1.6423887804726753e-05,
      "loss": 0.624,
      "step": 9529
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7338935136795044,
      "learning_rate": 1.6423107283086327e-05,
      "loss": 0.4705,
      "step": 9530
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.500483512878418,
      "learning_rate": 1.642232669482801e-05,
      "loss": 0.514,
      "step": 9531
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5469859838485718,
      "learning_rate": 1.642154603995989e-05,
      "loss": 0.4499,
      "step": 9532
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.632947325706482,
      "learning_rate": 1.642076531849007e-05,
      "loss": 0.5281,
      "step": 9533
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7930001020431519,
      "learning_rate": 1.641998453042665e-05,
      "loss": 0.5015,
      "step": 9534
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7477294206619263,
      "learning_rate": 1.6419203675777717e-05,
      "loss": 0.579,
      "step": 9535
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.531384825706482,
      "learning_rate": 1.6418422754551385e-05,
      "loss": 0.4416,
      "step": 9536
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6673436164855957,
      "learning_rate": 1.641764176675574e-05,
      "loss": 0.5516,
      "step": 9537
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5667248964309692,
      "learning_rate": 1.6416860712398892e-05,
      "loss": 0.4914,
      "step": 9538
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6239503622055054,
      "learning_rate": 1.6416079591488934e-05,
      "loss": 0.4961,
      "step": 9539
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6508572101593018,
      "learning_rate": 1.641529840403397e-05,
      "loss": 0.4881,
      "step": 9540
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6301546096801758,
      "learning_rate": 1.6414517150042105e-05,
      "loss": 0.5166,
      "step": 9541
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.648948073387146,
      "learning_rate": 1.641373582952144e-05,
      "loss": 0.4899,
      "step": 9542
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7075468301773071,
      "learning_rate": 1.641295444248008e-05,
      "loss": 0.5633,
      "step": 9543
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6450861692428589,
      "learning_rate": 1.6412172988926123e-05,
      "loss": 0.4474,
      "step": 9544
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6449865102767944,
      "learning_rate": 1.641139146886768e-05,
      "loss": 0.5421,
      "step": 9545
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6391223669052124,
      "learning_rate": 1.641060988231286e-05,
      "loss": 0.4687,
      "step": 9546
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5187804698944092,
      "learning_rate": 1.640982822926976e-05,
      "loss": 0.5019,
      "step": 9547
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6164976358413696,
      "learning_rate": 1.640904650974649e-05,
      "loss": 0.4954,
      "step": 9548
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6707756519317627,
      "learning_rate": 1.6408264723751156e-05,
      "loss": 0.5806,
      "step": 9549
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6649869680404663,
      "learning_rate": 1.640748287129188e-05,
      "loss": 0.465,
      "step": 9550
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6006412506103516,
      "learning_rate": 1.6406700952376754e-05,
      "loss": 0.5198,
      "step": 9551
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5614895820617676,
      "learning_rate": 1.6405918967013895e-05,
      "loss": 0.4687,
      "step": 9552
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5863672494888306,
      "learning_rate": 1.640513691521141e-05,
      "loss": 0.541,
      "step": 9553
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7288131713867188,
      "learning_rate": 1.6404354796977416e-05,
      "loss": 0.5047,
      "step": 9554
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6567116975784302,
      "learning_rate": 1.6403572612320023e-05,
      "loss": 0.5557,
      "step": 9555
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5814764499664307,
      "learning_rate": 1.6402790361247342e-05,
      "loss": 0.4691,
      "step": 9556
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6915415525436401,
      "learning_rate": 1.6402008043767485e-05,
      "loss": 0.5532,
      "step": 9557
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6350642442703247,
      "learning_rate": 1.6401225659888566e-05,
      "loss": 0.4789,
      "step": 9558
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6481152772903442,
      "learning_rate": 1.64004432096187e-05,
      "loss": 0.5535,
      "step": 9559
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5944321155548096,
      "learning_rate": 1.6399660692966008e-05,
      "loss": 0.4723,
      "step": 9560
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7122740745544434,
      "learning_rate": 1.6398878109938598e-05,
      "loss": 0.5613,
      "step": 9561
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6143856048583984,
      "learning_rate": 1.639809546054459e-05,
      "loss": 0.4512,
      "step": 9562
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.963739275932312,
      "learning_rate": 1.63973127447921e-05,
      "loss": 0.576,
      "step": 9563
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6722731590270996,
      "learning_rate": 1.639652996268925e-05,
      "loss": 0.4823,
      "step": 9564
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6146528720855713,
      "learning_rate": 1.6395747114244153e-05,
      "loss": 0.5168,
      "step": 9565
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5552911758422852,
      "learning_rate": 1.6394964199464932e-05,
      "loss": 0.44,
      "step": 9566
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.802268147468567,
      "learning_rate": 1.6394181218359704e-05,
      "loss": 0.6002,
      "step": 9567
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6596405506134033,
      "learning_rate": 1.6393398170936598e-05,
      "loss": 0.4726,
      "step": 9568
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.77066969871521,
      "learning_rate": 1.6392615057203724e-05,
      "loss": 0.5415,
      "step": 9569
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6327033042907715,
      "learning_rate": 1.639183187716921e-05,
      "loss": 0.4498,
      "step": 9570
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8943617343902588,
      "learning_rate": 1.6391048630841178e-05,
      "loss": 0.6121,
      "step": 9571
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5305160284042358,
      "learning_rate": 1.6390265318227753e-05,
      "loss": 0.4416,
      "step": 9572
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7688910961151123,
      "learning_rate": 1.6389481939337056e-05,
      "loss": 0.569,
      "step": 9573
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5481144189834595,
      "learning_rate": 1.638869849417722e-05,
      "loss": 0.4773,
      "step": 9574
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7188515663146973,
      "learning_rate": 1.6387914982756358e-05,
      "loss": 0.615,
      "step": 9575
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5183676481246948,
      "learning_rate": 1.6387131405082607e-05,
      "loss": 0.4574,
      "step": 9576
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6358468532562256,
      "learning_rate": 1.638634776116409e-05,
      "loss": 0.5132,
      "step": 9577
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5949794054031372,
      "learning_rate": 1.6385564051008933e-05,
      "loss": 0.4499,
      "step": 9578
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.531327247619629,
      "learning_rate": 1.6384780274625264e-05,
      "loss": 0.4736,
      "step": 9579
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5975873470306396,
      "learning_rate": 1.6383996432021216e-05,
      "loss": 0.4601,
      "step": 9580
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8589547872543335,
      "learning_rate": 1.6383212523204915e-05,
      "loss": 0.5645,
      "step": 9581
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6316401958465576,
      "learning_rate": 1.6382428548184495e-05,
      "loss": 0.4689,
      "step": 9582
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0356712341308594,
      "learning_rate": 1.638164450696808e-05,
      "loss": 1.0076,
      "step": 9583
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.728845477104187,
      "learning_rate": 1.6380860399563813e-05,
      "loss": 1.1407,
      "step": 9584
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7056559324264526,
      "learning_rate": 1.638007622597982e-05,
      "loss": 0.5861,
      "step": 9585
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.637511968612671,
      "learning_rate": 1.637929198622423e-05,
      "loss": 0.4621,
      "step": 9586
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7340267896652222,
      "learning_rate": 1.6378507680305182e-05,
      "loss": 0.5788,
      "step": 9587
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.709920048713684,
      "learning_rate": 1.637772330823081e-05,
      "loss": 0.4674,
      "step": 9588
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7318936586380005,
      "learning_rate": 1.6376938870009248e-05,
      "loss": 0.529,
      "step": 9589
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6388198137283325,
      "learning_rate": 1.6376154365648638e-05,
      "loss": 0.4489,
      "step": 9590
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7847795486450195,
      "learning_rate": 1.6375369795157107e-05,
      "loss": 0.5235,
      "step": 9591
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5933959484100342,
      "learning_rate": 1.6374585158542795e-05,
      "loss": 0.4586,
      "step": 9592
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.605748176574707,
      "learning_rate": 1.6373800455813846e-05,
      "loss": 0.4949,
      "step": 9593
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6786373853683472,
      "learning_rate": 1.637301568697839e-05,
      "loss": 0.4461,
      "step": 9594
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7200205326080322,
      "learning_rate": 1.637223085204457e-05,
      "loss": 0.4889,
      "step": 9595
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.654871940612793,
      "learning_rate": 1.637144595102053e-05,
      "loss": 0.5276,
      "step": 9596
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6058622598648071,
      "learning_rate": 1.6370660983914403e-05,
      "loss": 0.5057,
      "step": 9597
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8000248670578003,
      "learning_rate": 1.6369875950734336e-05,
      "loss": 0.5443,
      "step": 9598
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.560150384902954,
      "learning_rate": 1.636909085148847e-05,
      "loss": 0.4813,
      "step": 9599
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.697799801826477,
      "learning_rate": 1.636830568618495e-05,
      "loss": 0.4864,
      "step": 9600
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6391451358795166,
      "learning_rate": 1.6367520454831912e-05,
      "loss": 0.5076,
      "step": 9601
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.524610161781311,
      "learning_rate": 1.6366735157437505e-05,
      "loss": 0.4419,
      "step": 9602
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6595946550369263,
      "learning_rate": 1.6365949794009873e-05,
      "loss": 0.5066,
      "step": 9603
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5785293579101562,
      "learning_rate": 1.6365164364557165e-05,
      "loss": 0.4662,
      "step": 9604
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6811692714691162,
      "learning_rate": 1.6364378869087522e-05,
      "loss": 0.5501,
      "step": 9605
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6723134517669678,
      "learning_rate": 1.6363593307609096e-05,
      "loss": 0.4902,
      "step": 9606
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6665592193603516,
      "learning_rate": 1.636280768013003e-05,
      "loss": 0.5186,
      "step": 9607
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5324705839157104,
      "learning_rate": 1.636202198665847e-05,
      "loss": 0.4672,
      "step": 9608
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5687228441238403,
      "learning_rate": 1.6361236227202572e-05,
      "loss": 0.4719,
      "step": 9609
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.571007251739502,
      "learning_rate": 1.6360450401770482e-05,
      "loss": 0.4584,
      "step": 9610
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5814175605773926,
      "learning_rate": 1.635966451037035e-05,
      "loss": 0.4905,
      "step": 9611
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7587858438491821,
      "learning_rate": 1.6358878553010325e-05,
      "loss": 0.4699,
      "step": 9612
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8539966344833374,
      "learning_rate": 1.6358092529698565e-05,
      "loss": 0.5456,
      "step": 9613
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5824247598648071,
      "learning_rate": 1.635730644044322e-05,
      "loss": 0.4445,
      "step": 9614
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.604615569114685,
      "learning_rate": 1.6356520285252436e-05,
      "loss": 0.552,
      "step": 9615
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6842561960220337,
      "learning_rate": 1.6355734064134377e-05,
      "loss": 0.5608,
      "step": 9616
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.698653221130371,
      "learning_rate": 1.6354947777097188e-05,
      "loss": 0.4873,
      "step": 9617
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5016419887542725,
      "learning_rate": 1.635416142414903e-05,
      "loss": 0.4499,
      "step": 9618
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7572041749954224,
      "learning_rate": 1.635337500529806e-05,
      "loss": 0.5908,
      "step": 9619
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6001368761062622,
      "learning_rate": 1.635258852055243e-05,
      "loss": 0.4627,
      "step": 9620
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.602315902709961,
      "learning_rate": 1.6351801969920295e-05,
      "loss": 0.489,
      "step": 9621
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5895525217056274,
      "learning_rate": 1.635101535340982e-05,
      "loss": 0.5145,
      "step": 9622
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6050394773483276,
      "learning_rate": 1.6350228671029162e-05,
      "loss": 0.5667,
      "step": 9623
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6072025299072266,
      "learning_rate": 1.6349441922786473e-05,
      "loss": 0.4927,
      "step": 9624
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7064054012298584,
      "learning_rate": 1.634865510868992e-05,
      "loss": 0.5844,
      "step": 9625
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5448514223098755,
      "learning_rate": 1.6347868228747663e-05,
      "loss": 0.4701,
      "step": 9626
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6155600547790527,
      "learning_rate": 1.634708128296786e-05,
      "loss": 0.5861,
      "step": 9627
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8779792785644531,
      "learning_rate": 1.6346294271358677e-05,
      "loss": 0.5014,
      "step": 9628
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6435019969940186,
      "learning_rate": 1.634550719392827e-05,
      "loss": 0.5173,
      "step": 9629
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5655933618545532,
      "learning_rate": 1.634472005068481e-05,
      "loss": 0.4824,
      "step": 9630
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6988445520401,
      "learning_rate": 1.6343932841636455e-05,
      "loss": 0.5641,
      "step": 9631
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6168909072875977,
      "learning_rate": 1.6343145566791374e-05,
      "loss": 0.4839,
      "step": 9632
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6146055459976196,
      "learning_rate": 1.6342358226157732e-05,
      "loss": 0.5092,
      "step": 9633
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6448256969451904,
      "learning_rate": 1.6341570819743688e-05,
      "loss": 0.4682,
      "step": 9634
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.572927474975586,
      "learning_rate": 1.634078334755742e-05,
      "loss": 0.5614,
      "step": 9635
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5893925428390503,
      "learning_rate": 1.6339995809607084e-05,
      "loss": 0.4677,
      "step": 9636
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7039854526519775,
      "learning_rate": 1.6339208205900854e-05,
      "loss": 0.5037,
      "step": 9637
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7349209785461426,
      "learning_rate": 1.63384205364469e-05,
      "loss": 0.5016,
      "step": 9638
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.652197003364563,
      "learning_rate": 1.633763280125339e-05,
      "loss": 0.4997,
      "step": 9639
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.742577314376831,
      "learning_rate": 1.6336845000328492e-05,
      "loss": 0.4756,
      "step": 9640
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5680909156799316,
      "learning_rate": 1.633605713368038e-05,
      "loss": 0.4913,
      "step": 9641
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.59381103515625,
      "learning_rate": 1.633526920131722e-05,
      "loss": 0.466,
      "step": 9642
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6273709535598755,
      "learning_rate": 1.633448120324719e-05,
      "loss": 0.5489,
      "step": 9643
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6247167587280273,
      "learning_rate": 1.633369313947846e-05,
      "loss": 0.49,
      "step": 9644
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.627426266670227,
      "learning_rate": 1.6332905010019205e-05,
      "loss": 0.5269,
      "step": 9645
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6224066019058228,
      "learning_rate": 1.6332116814877597e-05,
      "loss": 0.454,
      "step": 9646
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6913176774978638,
      "learning_rate": 1.6331328554061815e-05,
      "loss": 0.5146,
      "step": 9647
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7158631086349487,
      "learning_rate": 1.6330540227580028e-05,
      "loss": 0.5236,
      "step": 9648
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6795434951782227,
      "learning_rate": 1.6329751835440414e-05,
      "loss": 0.5157,
      "step": 9649
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6600971221923828,
      "learning_rate": 1.632896337765116e-05,
      "loss": 0.4559,
      "step": 9650
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6695045232772827,
      "learning_rate": 1.6328174854220428e-05,
      "loss": 0.5258,
      "step": 9651
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6252275705337524,
      "learning_rate": 1.6327386265156404e-05,
      "loss": 0.4659,
      "step": 9652
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6923974752426147,
      "learning_rate": 1.632659761046727e-05,
      "loss": 0.5135,
      "step": 9653
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6069461107254028,
      "learning_rate": 1.63258088901612e-05,
      "loss": 0.4559,
      "step": 9654
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6892238855361938,
      "learning_rate": 1.6325020104246373e-05,
      "loss": 0.585,
      "step": 9655
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5375518798828125,
      "learning_rate": 1.632423125273098e-05,
      "loss": 0.4589,
      "step": 9656
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7420296669006348,
      "learning_rate": 1.632344233562319e-05,
      "loss": 0.5438,
      "step": 9657
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6707696914672852,
      "learning_rate": 1.6322653352931196e-05,
      "loss": 0.4762,
      "step": 9658
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6763590574264526,
      "learning_rate": 1.6321864304663174e-05,
      "loss": 0.5318,
      "step": 9659
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5932397842407227,
      "learning_rate": 1.6321075190827308e-05,
      "loss": 0.4991,
      "step": 9660
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5896774530410767,
      "learning_rate": 1.632028601143179e-05,
      "loss": 0.5164,
      "step": 9661
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7206884622573853,
      "learning_rate": 1.6319496766484794e-05,
      "loss": 0.4624,
      "step": 9662
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7132315635681152,
      "learning_rate": 1.6318707455994515e-05,
      "loss": 0.5108,
      "step": 9663
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6165682077407837,
      "learning_rate": 1.631791807996913e-05,
      "loss": 0.4477,
      "step": 9664
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6847243309020996,
      "learning_rate": 1.6317128638416833e-05,
      "loss": 0.5743,
      "step": 9665
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6998635530471802,
      "learning_rate": 1.6316339131345816e-05,
      "loss": 0.498,
      "step": 9666
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8889398574829102,
      "learning_rate": 1.6315549558764258e-05,
      "loss": 0.9782,
      "step": 9667
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5925862789154053,
      "learning_rate": 1.6314759920680354e-05,
      "loss": 1.0326,
      "step": 9668
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.709612250328064,
      "learning_rate": 1.631397021710229e-05,
      "loss": 0.5099,
      "step": 9669
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7422798871994019,
      "learning_rate": 1.6313180448038257e-05,
      "loss": 0.4826,
      "step": 9670
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6653447151184082,
      "learning_rate": 1.631239061349645e-05,
      "loss": 0.5114,
      "step": 9671
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.592558741569519,
      "learning_rate": 1.631160071348506e-05,
      "loss": 0.4797,
      "step": 9672
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5397930145263672,
      "learning_rate": 1.6310810748012272e-05,
      "loss": 0.501,
      "step": 9673
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.646958827972412,
      "learning_rate": 1.631002071708629e-05,
      "loss": 0.4716,
      "step": 9674
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6146116256713867,
      "learning_rate": 1.63092306207153e-05,
      "loss": 0.505,
      "step": 9675
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7458932399749756,
      "learning_rate": 1.6308440458907503e-05,
      "loss": 0.4787,
      "step": 9676
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6747297048568726,
      "learning_rate": 1.6307650231671087e-05,
      "loss": 0.5385,
      "step": 9677
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5911945104599,
      "learning_rate": 1.6306859939014253e-05,
      "loss": 0.471,
      "step": 9678
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6823228597640991,
      "learning_rate": 1.6306069580945195e-05,
      "loss": 0.5585,
      "step": 9679
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6475272178649902,
      "learning_rate": 1.6305279157472113e-05,
      "loss": 0.4938,
      "step": 9680
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7071927785873413,
      "learning_rate": 1.6304488668603204e-05,
      "loss": 0.5058,
      "step": 9681
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5629726648330688,
      "learning_rate": 1.6303698114346665e-05,
      "loss": 0.4607,
      "step": 9682
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7042895555496216,
      "learning_rate": 1.63029074947107e-05,
      "loss": 0.5263,
      "step": 9683
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6795070171356201,
      "learning_rate": 1.6302116809703502e-05,
      "loss": 0.4909,
      "step": 9684
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6663190126419067,
      "learning_rate": 1.6301326059333273e-05,
      "loss": 0.4865,
      "step": 9685
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4993337392807007,
      "learning_rate": 1.630053524360822e-05,
      "loss": 0.4432,
      "step": 9686
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6063058376312256,
      "learning_rate": 1.6299744362536542e-05,
      "loss": 0.5325,
      "step": 9687
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5934547185897827,
      "learning_rate": 1.6298953416126435e-05,
      "loss": 0.4938,
      "step": 9688
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6615060567855835,
      "learning_rate": 1.6298162404386115e-05,
      "loss": 0.523,
      "step": 9689
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5787906646728516,
      "learning_rate": 1.6297371327323778e-05,
      "loss": 0.4988,
      "step": 9690
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7069851160049438,
      "learning_rate": 1.6296580184947633e-05,
      "loss": 0.6218,
      "step": 9691
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5649539232254028,
      "learning_rate": 1.6295788977265877e-05,
      "loss": 0.4736,
      "step": 9692
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6347371339797974,
      "learning_rate": 1.6294997704286727e-05,
      "loss": 0.5061,
      "step": 9693
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5645438432693481,
      "learning_rate": 1.629420636601838e-05,
      "loss": 0.4513,
      "step": 9694
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7550990581512451,
      "learning_rate": 1.629341496246905e-05,
      "loss": 0.5433,
      "step": 9695
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5661643743515015,
      "learning_rate": 1.6292623493646943e-05,
      "loss": 0.4604,
      "step": 9696
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.668052077293396,
      "learning_rate": 1.6291831959560268e-05,
      "loss": 0.5892,
      "step": 9697
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5949184894561768,
      "learning_rate": 1.6291040360217234e-05,
      "loss": 0.4558,
      "step": 9698
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6620874404907227,
      "learning_rate": 1.6290248695626055e-05,
      "loss": 0.5309,
      "step": 9699
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5807515382766724,
      "learning_rate": 1.6289456965794935e-05,
      "loss": 0.4589,
      "step": 9700
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7456262111663818,
      "learning_rate": 1.6288665170732087e-05,
      "loss": 0.6158,
      "step": 9701
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6437736749649048,
      "learning_rate": 1.6287873310445727e-05,
      "loss": 0.4764,
      "step": 9702
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7102991342544556,
      "learning_rate": 1.6287081384944067e-05,
      "loss": 0.5765,
      "step": 9703
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6537009477615356,
      "learning_rate": 1.628628939423532e-05,
      "loss": 0.5003,
      "step": 9704
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.937308430671692,
      "learning_rate": 1.6285497338327697e-05,
      "loss": 0.5786,
      "step": 9705
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5823767185211182,
      "learning_rate": 1.628470521722942e-05,
      "loss": 0.4532,
      "step": 9706
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6594105958938599,
      "learning_rate": 1.62839130309487e-05,
      "loss": 0.5399,
      "step": 9707
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6621627807617188,
      "learning_rate": 1.628312077949375e-05,
      "loss": 0.4754,
      "step": 9708
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5682677030563354,
      "learning_rate": 1.628232846287279e-05,
      "loss": 0.5486,
      "step": 9709
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.559158444404602,
      "learning_rate": 1.628153608109404e-05,
      "loss": 0.484,
      "step": 9710
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.603009581565857,
      "learning_rate": 1.6280743634165713e-05,
      "loss": 0.5366,
      "step": 9711
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7131484746932983,
      "learning_rate": 1.6279951122096034e-05,
      "loss": 0.5304,
      "step": 9712
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7188937664031982,
      "learning_rate": 1.6279158544893222e-05,
      "loss": 0.4876,
      "step": 9713
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5869898796081543,
      "learning_rate": 1.627836590256549e-05,
      "loss": 0.495,
      "step": 9714
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5595297813415527,
      "learning_rate": 1.6277573195121067e-05,
      "loss": 0.5198,
      "step": 9715
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.815398931503296,
      "learning_rate": 1.6276780422568168e-05,
      "loss": 0.5148,
      "step": 9716
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5591963529586792,
      "learning_rate": 1.6275987584915026e-05,
      "loss": 0.5042,
      "step": 9717
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.678736925125122,
      "learning_rate": 1.627519468216985e-05,
      "loss": 0.4796,
      "step": 9718
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7518612146377563,
      "learning_rate": 1.6274401714340873e-05,
      "loss": 0.5764,
      "step": 9719
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6484360694885254,
      "learning_rate": 1.6273608681436317e-05,
      "loss": 0.4764,
      "step": 9720
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6911401748657227,
      "learning_rate": 1.6272815583464408e-05,
      "loss": 0.5666,
      "step": 9721
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6618329286575317,
      "learning_rate": 1.6272022420433368e-05,
      "loss": 0.4578,
      "step": 9722
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6452826261520386,
      "learning_rate": 1.6271229192351428e-05,
      "loss": 0.5334,
      "step": 9723
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6282317638397217,
      "learning_rate": 1.6270435899226814e-05,
      "loss": 0.4929,
      "step": 9724
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7896455526351929,
      "learning_rate": 1.6269642541067753e-05,
      "loss": 0.5816,
      "step": 9725
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7841894626617432,
      "learning_rate": 1.626884911788247e-05,
      "loss": 0.4771,
      "step": 9726
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6007049083709717,
      "learning_rate": 1.6268055629679202e-05,
      "loss": 0.4776,
      "step": 9727
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.571635127067566,
      "learning_rate": 1.6267262076466166e-05,
      "loss": 0.4576,
      "step": 9728
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8090044260025024,
      "learning_rate": 1.6266468458251607e-05,
      "loss": 0.5476,
      "step": 9729
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6144088506698608,
      "learning_rate": 1.626567477504375e-05,
      "loss": 0.4809,
      "step": 9730
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6578224897384644,
      "learning_rate": 1.626488102685082e-05,
      "loss": 0.5184,
      "step": 9731
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5628594160079956,
      "learning_rate": 1.6264087213681063e-05,
      "loss": 0.4511,
      "step": 9732
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6012073755264282,
      "learning_rate": 1.62632933355427e-05,
      "loss": 0.4981,
      "step": 9733
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7502052783966064,
      "learning_rate": 1.626249939244397e-05,
      "loss": 0.5165,
      "step": 9734
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6786093711853027,
      "learning_rate": 1.626170538439311e-05,
      "loss": 0.5449,
      "step": 9735
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6421043872833252,
      "learning_rate": 1.6260911311398353e-05,
      "loss": 0.4555,
      "step": 9736
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.553410530090332,
      "learning_rate": 1.626011717346793e-05,
      "loss": 0.4769,
      "step": 9737
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8046314716339111,
      "learning_rate": 1.6259322970610083e-05,
      "loss": 0.5177,
      "step": 9738
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.806990146636963,
      "learning_rate": 1.625852870283305e-05,
      "loss": 0.5554,
      "step": 9739
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6845144033432007,
      "learning_rate": 1.625773437014506e-05,
      "loss": 0.502,
      "step": 9740
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6330351829528809,
      "learning_rate": 1.6256939972554363e-05,
      "loss": 0.5417,
      "step": 9741
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.681787371635437,
      "learning_rate": 1.6256145510069192e-05,
      "loss": 0.4461,
      "step": 9742
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6185557842254639,
      "learning_rate": 1.625535098269779e-05,
      "loss": 0.5252,
      "step": 9743
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.582208275794983,
      "learning_rate": 1.6254556390448394e-05,
      "loss": 0.4677,
      "step": 9744
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5955603122711182,
      "learning_rate": 1.6253761733329247e-05,
      "loss": 0.5312,
      "step": 9745
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5954943895339966,
      "learning_rate": 1.625296701134859e-05,
      "loss": 0.4612,
      "step": 9746
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6234811544418335,
      "learning_rate": 1.6252172224514665e-05,
      "loss": 0.5114,
      "step": 9747
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.701286792755127,
      "learning_rate": 1.6251377372835716e-05,
      "loss": 0.4963,
      "step": 9748
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6819908618927002,
      "learning_rate": 1.625058245631999e-05,
      "loss": 0.5531,
      "step": 9749
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5588756799697876,
      "learning_rate": 1.6249787474975727e-05,
      "loss": 0.4584,
      "step": 9750
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.707962155342102,
      "learning_rate": 1.6248992428811173e-05,
      "loss": 0.5775,
      "step": 9751
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5560179948806763,
      "learning_rate": 1.6248197317834575e-05,
      "loss": 0.4332,
      "step": 9752
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6727477312088013,
      "learning_rate": 1.6247402142054183e-05,
      "loss": 0.5484,
      "step": 9753
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5936576128005981,
      "learning_rate": 1.624660690147824e-05,
      "loss": 0.4319,
      "step": 9754
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.678472638130188,
      "learning_rate": 1.624581159611499e-05,
      "loss": 0.5186,
      "step": 9755
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5747822523117065,
      "learning_rate": 1.6245016225972688e-05,
      "loss": 0.4514,
      "step": 9756
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.618344783782959,
      "learning_rate": 1.6244220791059585e-05,
      "loss": 0.4938,
      "step": 9757
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.626061201095581,
      "learning_rate": 1.6243425291383924e-05,
      "loss": 0.4859,
      "step": 9758
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6026580333709717,
      "learning_rate": 1.6242629726953957e-05,
      "loss": 0.5408,
      "step": 9759
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5351753234863281,
      "learning_rate": 1.6241834097777943e-05,
      "loss": 0.4622,
      "step": 9760
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0634028911590576,
      "learning_rate": 1.6241038403864123e-05,
      "loss": 0.5167,
      "step": 9761
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5267225503921509,
      "learning_rate": 1.624024264522076e-05,
      "loss": 0.4492,
      "step": 9762
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7274824380874634,
      "learning_rate": 1.6239446821856097e-05,
      "loss": 0.5306,
      "step": 9763
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.59360671043396,
      "learning_rate": 1.62386509337784e-05,
      "loss": 0.5005,
      "step": 9764
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6691560745239258,
      "learning_rate": 1.6237854980995907e-05,
      "loss": 0.4972,
      "step": 9765
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.616862177848816,
      "learning_rate": 1.623705896351689e-05,
      "loss": 0.4972,
      "step": 9766
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.639092206954956,
      "learning_rate": 1.6236262881349594e-05,
      "loss": 0.5376,
      "step": 9767
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6684402227401733,
      "learning_rate": 1.6235466734502283e-05,
      "loss": 0.4524,
      "step": 9768
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7027571201324463,
      "learning_rate": 1.623467052298321e-05,
      "loss": 0.5422,
      "step": 9769
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5458669662475586,
      "learning_rate": 1.6233874246800633e-05,
      "loss": 0.4633,
      "step": 9770
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6414705514907837,
      "learning_rate": 1.6233077905962814e-05,
      "loss": 0.5009,
      "step": 9771
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6885870695114136,
      "learning_rate": 1.6232281500478008e-05,
      "loss": 0.4725,
      "step": 9772
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6787376403808594,
      "learning_rate": 1.6231485030354477e-05,
      "loss": 0.5622,
      "step": 9773
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5665751695632935,
      "learning_rate": 1.623068849560048e-05,
      "loss": 0.4763,
      "step": 9774
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.821465015411377,
      "learning_rate": 1.622989189622428e-05,
      "loss": 0.578,
      "step": 9775
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.692433476448059,
      "learning_rate": 1.622909523223414e-05,
      "loss": 0.4827,
      "step": 9776
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6720253229141235,
      "learning_rate": 1.6228298503638323e-05,
      "loss": 0.5267,
      "step": 9777
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7232619524002075,
      "learning_rate": 1.622750171044509e-05,
      "loss": 0.5,
      "step": 9778
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6657850742340088,
      "learning_rate": 1.6226704852662707e-05,
      "loss": 0.4866,
      "step": 9779
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.779679775238037,
      "learning_rate": 1.6225907930299436e-05,
      "loss": 0.4466,
      "step": 9780
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6290338039398193,
      "learning_rate": 1.6225110943363543e-05,
      "loss": 0.5177,
      "step": 9781
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6800199747085571,
      "learning_rate": 1.6224313891863296e-05,
      "loss": 0.4801,
      "step": 9782
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6931869983673096,
      "learning_rate": 1.6223516775806965e-05,
      "loss": 0.5425,
      "step": 9783
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6659680604934692,
      "learning_rate": 1.6222719595202807e-05,
      "loss": 0.4994,
      "step": 9784
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7082208395004272,
      "learning_rate": 1.62219223500591e-05,
      "loss": 0.533,
      "step": 9785
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6096853017807007,
      "learning_rate": 1.6221125040384106e-05,
      "loss": 0.4422,
      "step": 9786
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6604821681976318,
      "learning_rate": 1.62203276661861e-05,
      "loss": 0.5055,
      "step": 9787
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5589622259140015,
      "learning_rate": 1.621953022747335e-05,
      "loss": 0.4796,
      "step": 9788
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6263446807861328,
      "learning_rate": 1.6218732724254124e-05,
      "loss": 0.5111,
      "step": 9789
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9804441928863525,
      "learning_rate": 1.6217935156536695e-05,
      "loss": 0.4321,
      "step": 9790
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.722230076789856,
      "learning_rate": 1.6217137524329338e-05,
      "loss": 0.5714,
      "step": 9791
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.609865665435791,
      "learning_rate": 1.6216339827640317e-05,
      "loss": 0.4837,
      "step": 9792
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6476763486862183,
      "learning_rate": 1.621554206647792e-05,
      "loss": 0.6078,
      "step": 9793
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4989396333694458,
      "learning_rate": 1.6214744240850406e-05,
      "loss": 0.4702,
      "step": 9794
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7795952558517456,
      "learning_rate": 1.621394635076606e-05,
      "loss": 0.5424,
      "step": 9795
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5993194580078125,
      "learning_rate": 1.621314839623315e-05,
      "loss": 0.4569,
      "step": 9796
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6516796350479126,
      "learning_rate": 1.621235037725996e-05,
      "loss": 0.5014,
      "step": 9797
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6667777299880981,
      "learning_rate": 1.6211552293854758e-05,
      "loss": 0.4524,
      "step": 9798
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7466543912887573,
      "learning_rate": 1.6210754146025832e-05,
      "loss": 0.549,
      "step": 9799
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5231207609176636,
      "learning_rate": 1.620995593378145e-05,
      "loss": 0.4415,
      "step": 9800
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6191718578338623,
      "learning_rate": 1.6209157657129898e-05,
      "loss": 0.5153,
      "step": 9801
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6328012943267822,
      "learning_rate": 1.6208359316079448e-05,
      "loss": 0.4925,
      "step": 9802
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6006780862808228,
      "learning_rate": 1.6207560910638384e-05,
      "loss": 0.5212,
      "step": 9803
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6188613176345825,
      "learning_rate": 1.620676244081499e-05,
      "loss": 0.4386,
      "step": 9804
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6431968212127686,
      "learning_rate": 1.620596390661754e-05,
      "loss": 0.5515,
      "step": 9805
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6243809461593628,
      "learning_rate": 1.6205165308054326e-05,
      "loss": 0.4799,
      "step": 9806
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7306920289993286,
      "learning_rate": 1.6204366645133626e-05,
      "loss": 0.521,
      "step": 9807
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5605287551879883,
      "learning_rate": 1.6203567917863718e-05,
      "loss": 0.4536,
      "step": 9808
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7353568077087402,
      "learning_rate": 1.6202769126252892e-05,
      "loss": 0.53,
      "step": 9809
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7544089555740356,
      "learning_rate": 1.620197027030943e-05,
      "loss": 0.4646,
      "step": 9810
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8780802488327026,
      "learning_rate": 1.620117135004162e-05,
      "loss": 0.58,
      "step": 9811
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6445300579071045,
      "learning_rate": 1.6200372365457747e-05,
      "loss": 0.461,
      "step": 9812
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6698477268218994,
      "learning_rate": 1.61995733165661e-05,
      "loss": 0.4866,
      "step": 9813
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.556257963180542,
      "learning_rate": 1.619877420337496e-05,
      "loss": 0.4866,
      "step": 9814
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6445897817611694,
      "learning_rate": 1.6197975025892623e-05,
      "loss": 0.5181,
      "step": 9815
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.661373257637024,
      "learning_rate": 1.619717578412737e-05,
      "loss": 0.4868,
      "step": 9816
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6688584089279175,
      "learning_rate": 1.61963764780875e-05,
      "loss": 0.5024,
      "step": 9817
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.571146845817566,
      "learning_rate": 1.6195577107781292e-05,
      "loss": 0.4922,
      "step": 9818
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5797709226608276,
      "learning_rate": 1.6194777673217043e-05,
      "loss": 0.4909,
      "step": 9819
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5575214624404907,
      "learning_rate": 1.6193978174403045e-05,
      "loss": 0.44,
      "step": 9820
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1578774452209473,
      "learning_rate": 1.6193178611347587e-05,
      "loss": 0.9796,
      "step": 9821
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8597033023834229,
      "learning_rate": 1.6192378984058966e-05,
      "loss": 1.0705,
      "step": 9822
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7336581945419312,
      "learning_rate": 1.6191579292545473e-05,
      "loss": 0.5247,
      "step": 9823
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6491762399673462,
      "learning_rate": 1.61907795368154e-05,
      "loss": 0.4705,
      "step": 9824
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6709765195846558,
      "learning_rate": 1.6189979716877043e-05,
      "loss": 0.5333,
      "step": 9825
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6039702892303467,
      "learning_rate": 1.61891798327387e-05,
      "loss": 0.4757,
      "step": 9826
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.677032232284546,
      "learning_rate": 1.6188379884408664e-05,
      "loss": 0.526,
      "step": 9827
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5878723859786987,
      "learning_rate": 1.618757987189523e-05,
      "loss": 0.4628,
      "step": 9828
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5638607740402222,
      "learning_rate": 1.6186779795206703e-05,
      "loss": 0.4573,
      "step": 9829
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6958327293395996,
      "learning_rate": 1.6185979654351376e-05,
      "loss": 0.5357,
      "step": 9830
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6735615730285645,
      "learning_rate": 1.6185179449337546e-05,
      "loss": 0.5224,
      "step": 9831
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5448557138442993,
      "learning_rate": 1.6184379180173516e-05,
      "loss": 0.4645,
      "step": 9832
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8048759698867798,
      "learning_rate": 1.6183578846867585e-05,
      "loss": 0.5628,
      "step": 9833
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5595341920852661,
      "learning_rate": 1.6182778449428052e-05,
      "loss": 0.4471,
      "step": 9834
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6748559474945068,
      "learning_rate": 1.618197798786322e-05,
      "loss": 0.5207,
      "step": 9835
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6745858192443848,
      "learning_rate": 1.618117746218139e-05,
      "loss": 0.4691,
      "step": 9836
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.700862169265747,
      "learning_rate": 1.6180376872390867e-05,
      "loss": 0.5403,
      "step": 9837
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6503900289535522,
      "learning_rate": 1.6179576218499953e-05,
      "loss": 0.4652,
      "step": 9838
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7001826763153076,
      "learning_rate": 1.617877550051695e-05,
      "loss": 0.4846,
      "step": 9839
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7021735906600952,
      "learning_rate": 1.6177974718450163e-05,
      "loss": 0.4826,
      "step": 9840
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6059125661849976,
      "learning_rate": 1.6177173872307907e-05,
      "loss": 0.5498,
      "step": 9841
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5461763143539429,
      "learning_rate": 1.6176372962098473e-05,
      "loss": 0.4456,
      "step": 9842
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6540052890777588,
      "learning_rate": 1.6175571987830178e-05,
      "loss": 0.5988,
      "step": 9843
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5473487377166748,
      "learning_rate": 1.6174770949511324e-05,
      "loss": 0.4722,
      "step": 9844
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.487324595451355,
      "learning_rate": 1.6173969847150224e-05,
      "loss": 0.4861,
      "step": 9845
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.595889687538147,
      "learning_rate": 1.617316868075518e-05,
      "loss": 0.4714,
      "step": 9846
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5685715675354004,
      "learning_rate": 1.617236745033451e-05,
      "loss": 0.5126,
      "step": 9847
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6493884325027466,
      "learning_rate": 1.6171566155896515e-05,
      "loss": 0.5043,
      "step": 9848
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7024974822998047,
      "learning_rate": 1.6170764797449515e-05,
      "loss": 0.5184,
      "step": 9849
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7060478925704956,
      "learning_rate": 1.6169963375001813e-05,
      "loss": 0.4453,
      "step": 9850
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6009478569030762,
      "learning_rate": 1.6169161888561723e-05,
      "loss": 0.5982,
      "step": 9851
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5914236307144165,
      "learning_rate": 1.616836033813756e-05,
      "loss": 0.486,
      "step": 9852
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.720233678817749,
      "learning_rate": 1.616755872373764e-05,
      "loss": 0.5017,
      "step": 9853
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6521244049072266,
      "learning_rate": 1.6166757045370274e-05,
      "loss": 0.4715,
      "step": 9854
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.666244626045227,
      "learning_rate": 1.6165955303043775e-05,
      "loss": 0.5172,
      "step": 9855
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5845794677734375,
      "learning_rate": 1.6165153496766457e-05,
      "loss": 0.4421,
      "step": 9856
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.634561538696289,
      "learning_rate": 1.6164351626546642e-05,
      "loss": 0.6192,
      "step": 9857
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6027405261993408,
      "learning_rate": 1.616354969239264e-05,
      "loss": 0.489,
      "step": 9858
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.611274242401123,
      "learning_rate": 1.6162747694312774e-05,
      "loss": 0.5275,
      "step": 9859
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5535024404525757,
      "learning_rate": 1.616194563231536e-05,
      "loss": 0.4838,
      "step": 9860
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6355153322219849,
      "learning_rate": 1.6161143506408715e-05,
      "loss": 0.5354,
      "step": 9861
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5545902252197266,
      "learning_rate": 1.616034131660116e-05,
      "loss": 0.4615,
      "step": 9862
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7547852993011475,
      "learning_rate": 1.6159539062901016e-05,
      "loss": 0.5398,
      "step": 9863
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5642508268356323,
      "learning_rate": 1.61587367453166e-05,
      "loss": 0.4494,
      "step": 9864
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.550366759300232,
      "learning_rate": 1.615793436385624e-05,
      "loss": 0.4886,
      "step": 9865
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6363009214401245,
      "learning_rate": 1.6157131918528247e-05,
      "loss": 0.4968,
      "step": 9866
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6453828811645508,
      "learning_rate": 1.6156329409340953e-05,
      "loss": 0.5273,
      "step": 9867
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5308334827423096,
      "learning_rate": 1.6155526836302682e-05,
      "loss": 0.4384,
      "step": 9868
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5348705053329468,
      "learning_rate": 1.615472419942175e-05,
      "loss": 0.4961,
      "step": 9869
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.595180869102478,
      "learning_rate": 1.615392149870649e-05,
      "loss": 0.4665,
      "step": 9870
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.099346399307251,
      "learning_rate": 1.615311873416522e-05,
      "loss": 1.0012,
      "step": 9871
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8078289031982422,
      "learning_rate": 1.615231590580627e-05,
      "loss": 1.0554,
      "step": 9872
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.738439679145813,
      "learning_rate": 1.6151513013637967e-05,
      "loss": 0.5007,
      "step": 9873
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7527248859405518,
      "learning_rate": 1.6150710057668637e-05,
      "loss": 0.4743,
      "step": 9874
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.666893482208252,
      "learning_rate": 1.6149907037906607e-05,
      "loss": 0.4989,
      "step": 9875
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.591184139251709,
      "learning_rate": 1.614910395436021e-05,
      "loss": 0.446,
      "step": 9876
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5777034759521484,
      "learning_rate": 1.6148300807037765e-05,
      "loss": 0.5656,
      "step": 9877
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6785788536071777,
      "learning_rate": 1.6147497595947617e-05,
      "loss": 0.476,
      "step": 9878
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5999327898025513,
      "learning_rate": 1.6146694321098087e-05,
      "loss": 0.5098,
      "step": 9879
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5517295598983765,
      "learning_rate": 1.6145890982497507e-05,
      "loss": 0.4483,
      "step": 9880
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7366664409637451,
      "learning_rate": 1.6145087580154207e-05,
      "loss": 0.4972,
      "step": 9881
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5916438102722168,
      "learning_rate": 1.6144284114076527e-05,
      "loss": 0.4583,
      "step": 9882
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6337976455688477,
      "learning_rate": 1.6143480584272794e-05,
      "loss": 0.5279,
      "step": 9883
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.575839638710022,
      "learning_rate": 1.6142676990751343e-05,
      "loss": 0.4619,
      "step": 9884
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6288996934890747,
      "learning_rate": 1.6141873333520505e-05,
      "loss": 0.5202,
      "step": 9885
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5555857419967651,
      "learning_rate": 1.6141069612588626e-05,
      "loss": 0.4539,
      "step": 9886
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7484389543533325,
      "learning_rate": 1.614026582796403e-05,
      "loss": 0.5956,
      "step": 9887
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5369770526885986,
      "learning_rate": 1.613946197965506e-05,
      "loss": 0.4486,
      "step": 9888
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6978063583374023,
      "learning_rate": 1.6138658067670052e-05,
      "loss": 0.5542,
      "step": 9889
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.512149691581726,
      "learning_rate": 1.6137854092017344e-05,
      "loss": 0.4592,
      "step": 9890
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6095632314682007,
      "learning_rate": 1.613705005270527e-05,
      "loss": 0.5291,
      "step": 9891
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5460989475250244,
      "learning_rate": 1.6136245949742183e-05,
      "loss": 0.4959,
      "step": 9892
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5593618154525757,
      "learning_rate": 1.6135441783136404e-05,
      "loss": 0.5295,
      "step": 9893
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5594518184661865,
      "learning_rate": 1.6134637552896285e-05,
      "loss": 0.457,
      "step": 9894
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6185894012451172,
      "learning_rate": 1.6133833259030168e-05,
      "loss": 0.5375,
      "step": 9895
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6379613876342773,
      "learning_rate": 1.613302890154639e-05,
      "loss": 0.4646,
      "step": 9896
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7034045457839966,
      "learning_rate": 1.6132224480453293e-05,
      "loss": 0.5858,
      "step": 9897
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6591626405715942,
      "learning_rate": 1.6131419995759222e-05,
      "loss": 0.471,
      "step": 9898
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7155709266662598,
      "learning_rate": 1.6130615447472522e-05,
      "loss": 0.5568,
      "step": 9899
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6436092853546143,
      "learning_rate": 1.612981083560154e-05,
      "loss": 0.4757,
      "step": 9900
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7952351570129395,
      "learning_rate": 1.6129006160154615e-05,
      "loss": 0.576,
      "step": 9901
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6500409841537476,
      "learning_rate": 1.6128201421140094e-05,
      "loss": 0.4454,
      "step": 9902
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.708905816078186,
      "learning_rate": 1.6127396618566328e-05,
      "loss": 0.5261,
      "step": 9903
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6210993528366089,
      "learning_rate": 1.612659175244166e-05,
      "loss": 0.4535,
      "step": 9904
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7067325115203857,
      "learning_rate": 1.612578682277443e-05,
      "loss": 0.5954,
      "step": 9905
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5028417110443115,
      "learning_rate": 1.6124981829573007e-05,
      "loss": 0.4397,
      "step": 9906
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5936063528060913,
      "learning_rate": 1.6124176772845724e-05,
      "loss": 0.5798,
      "step": 9907
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5520437955856323,
      "learning_rate": 1.6123371652600935e-05,
      "loss": 0.4625,
      "step": 9908
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6206095218658447,
      "learning_rate": 1.6122566468846988e-05,
      "loss": 0.5117,
      "step": 9909
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6844884157180786,
      "learning_rate": 1.6121761221592237e-05,
      "loss": 0.536,
      "step": 9910
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6309672594070435,
      "learning_rate": 1.6120955910845032e-05,
      "loss": 0.5079,
      "step": 9911
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.642946720123291,
      "learning_rate": 1.6120150536613727e-05,
      "loss": 0.5143,
      "step": 9912
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6002401113510132,
      "learning_rate": 1.611934509890668e-05,
      "loss": 0.5221,
      "step": 9913
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6006346940994263,
      "learning_rate": 1.6118539597732233e-05,
      "loss": 0.4666,
      "step": 9914
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6062183380126953,
      "learning_rate": 1.6117734033098744e-05,
      "loss": 0.4991,
      "step": 9915
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.632885217666626,
      "learning_rate": 1.6116928405014575e-05,
      "loss": 0.4868,
      "step": 9916
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.3750736713409424,
      "learning_rate": 1.6116122713488076e-05,
      "loss": 1.0618,
      "step": 9917
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9546191692352295,
      "learning_rate": 1.6115316958527605e-05,
      "loss": 1.0561,
      "step": 9918
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9007166624069214,
      "learning_rate": 1.6114511140141516e-05,
      "loss": 0.5188,
      "step": 9919
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8164629936218262,
      "learning_rate": 1.611370525833817e-05,
      "loss": 0.5384,
      "step": 9920
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8078542947769165,
      "learning_rate": 1.6112899313125926e-05,
      "loss": 0.5672,
      "step": 9921
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.671498417854309,
      "learning_rate": 1.6112093304513137e-05,
      "loss": 0.4551,
      "step": 9922
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6239583492279053,
      "learning_rate": 1.611128723250817e-05,
      "loss": 0.5177,
      "step": 9923
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6333434581756592,
      "learning_rate": 1.611048109711938e-05,
      "loss": 0.4646,
      "step": 9924
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.742961049079895,
      "learning_rate": 1.6109674898355132e-05,
      "loss": 0.514,
      "step": 9925
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6044714450836182,
      "learning_rate": 1.6108868636223784e-05,
      "loss": 0.4331,
      "step": 9926
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7029237747192383,
      "learning_rate": 1.61080623107337e-05,
      "loss": 0.5548,
      "step": 9927
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.629533290863037,
      "learning_rate": 1.6107255921893242e-05,
      "loss": 0.4611,
      "step": 9928
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.859928011894226,
      "learning_rate": 1.6106449469710777e-05,
      "loss": 0.5065,
      "step": 9929
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7143523693084717,
      "learning_rate": 1.6105642954194665e-05,
      "loss": 0.4566,
      "step": 9930
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8092654943466187,
      "learning_rate": 1.610483637535327e-05,
      "loss": 0.4829,
      "step": 9931
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.654854416847229,
      "learning_rate": 1.610402973319496e-05,
      "loss": 0.4796,
      "step": 9932
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.626949429512024,
      "learning_rate": 1.61032230277281e-05,
      "loss": 0.4952,
      "step": 9933
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5904399156570435,
      "learning_rate": 1.6102416258961063e-05,
      "loss": 0.465,
      "step": 9934
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6462390422821045,
      "learning_rate": 1.6101609426902206e-05,
      "loss": 0.5034,
      "step": 9935
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7696361541748047,
      "learning_rate": 1.6100802531559906e-05,
      "loss": 0.4902,
      "step": 9936
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.326270818710327,
      "learning_rate": 1.6099995572942523e-05,
      "loss": 0.5654,
      "step": 9937
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5840611457824707,
      "learning_rate": 1.6099188551058436e-05,
      "loss": 0.4404,
      "step": 9938
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5465092658996582,
      "learning_rate": 1.6098381465916005e-05,
      "loss": 0.4888,
      "step": 9939
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7422127723693848,
      "learning_rate": 1.609757431752361e-05,
      "loss": 0.4789,
      "step": 9940
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7196977138519287,
      "learning_rate": 1.6096767105889623e-05,
      "loss": 0.6023,
      "step": 9941
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6281049251556396,
      "learning_rate": 1.6095959831022406e-05,
      "loss": 0.4813,
      "step": 9942
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.706833839416504,
      "learning_rate": 1.6095152492930343e-05,
      "loss": 0.5765,
      "step": 9943
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6710655689239502,
      "learning_rate": 1.6094345091621798e-05,
      "loss": 0.4785,
      "step": 9944
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6487963199615479,
      "learning_rate": 1.6093537627105148e-05,
      "loss": 0.4992,
      "step": 9945
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6618714332580566,
      "learning_rate": 1.609273009938877e-05,
      "loss": 0.4618,
      "step": 9946
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5645854473114014,
      "learning_rate": 1.609192250848104e-05,
      "loss": 0.5224,
      "step": 9947
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5582740306854248,
      "learning_rate": 1.609111485439033e-05,
      "loss": 0.4584,
      "step": 9948
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6423338651657104,
      "learning_rate": 1.609030713712502e-05,
      "loss": 0.5503,
      "step": 9949
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5018460750579834,
      "learning_rate": 1.6089499356693485e-05,
      "loss": 0.4499,
      "step": 9950
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6417043209075928,
      "learning_rate": 1.6088691513104105e-05,
      "loss": 0.5342,
      "step": 9951
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.59639573097229,
      "learning_rate": 1.6087883606365258e-05,
      "loss": 0.4573,
      "step": 9952
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9693409204483032,
      "learning_rate": 1.608707563648532e-05,
      "loss": 0.994,
      "step": 9953
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7898213863372803,
      "learning_rate": 1.6086267603472677e-05,
      "loss": 1.0852,
      "step": 9954
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0860445499420166,
      "learning_rate": 1.6085459507335705e-05,
      "loss": 1.0084,
      "step": 9955
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5546072721481323,
      "learning_rate": 1.608465134808279e-05,
      "loss": 1.0486,
      "step": 9956
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8623353242874146,
      "learning_rate": 1.6083843125722306e-05,
      "loss": 0.5318,
      "step": 9957
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.923142910003662,
      "learning_rate": 1.6083034840262645e-05,
      "loss": 0.4736,
      "step": 9958
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8330976963043213,
      "learning_rate": 1.608222649171218e-05,
      "loss": 0.5217,
      "step": 9959
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.602224349975586,
      "learning_rate": 1.6081418080079305e-05,
      "loss": 0.4985,
      "step": 9960
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7978427410125732,
      "learning_rate": 1.6080609605372395e-05,
      "loss": 0.5214,
      "step": 9961
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7280160188674927,
      "learning_rate": 1.6079801067599844e-05,
      "loss": 0.5371,
      "step": 9962
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8310377597808838,
      "learning_rate": 1.6078992466770033e-05,
      "loss": 0.5177,
      "step": 9963
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6109296083450317,
      "learning_rate": 1.6078183802891348e-05,
      "loss": 0.4528,
      "step": 9964
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6564080715179443,
      "learning_rate": 1.607737507597218e-05,
      "loss": 0.5372,
      "step": 9965
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7279462814331055,
      "learning_rate": 1.6076566286020914e-05,
      "loss": 0.4645,
      "step": 9966
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8358547687530518,
      "learning_rate": 1.607575743304594e-05,
      "loss": 0.5398,
      "step": 9967
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7205100059509277,
      "learning_rate": 1.607494851705564e-05,
      "loss": 0.4495,
      "step": 9968
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.700760841369629,
      "learning_rate": 1.6074139538058414e-05,
      "loss": 0.5591,
      "step": 9969
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.735826849937439,
      "learning_rate": 1.6073330496062646e-05,
      "loss": 0.5095,
      "step": 9970
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6476192474365234,
      "learning_rate": 1.607252139107673e-05,
      "loss": 0.5193,
      "step": 9971
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8220736980438232,
      "learning_rate": 1.6071712223109058e-05,
      "loss": 0.4738,
      "step": 9972
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7869478464126587,
      "learning_rate": 1.607090299216802e-05,
      "loss": 0.4951,
      "step": 9973
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5718187093734741,
      "learning_rate": 1.607009369826201e-05,
      "loss": 0.4665,
      "step": 9974
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7787880897521973,
      "learning_rate": 1.606928434139942e-05,
      "loss": 0.5621,
      "step": 9975
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.605296015739441,
      "learning_rate": 1.6068474921588648e-05,
      "loss": 0.4512,
      "step": 9976
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5622105598449707,
      "learning_rate": 1.6067665438838086e-05,
      "loss": 0.4974,
      "step": 9977
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7360503673553467,
      "learning_rate": 1.606685589315613e-05,
      "loss": 0.4951,
      "step": 9978
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5736565589904785,
      "learning_rate": 1.606604628455118e-05,
      "loss": 0.5386,
      "step": 9979
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5747191905975342,
      "learning_rate": 1.6065236613031627e-05,
      "loss": 0.494,
      "step": 9980
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5979516506195068,
      "learning_rate": 1.6064426878605873e-05,
      "loss": 0.5173,
      "step": 9981
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6643335819244385,
      "learning_rate": 1.6063617081282315e-05,
      "loss": 0.4504,
      "step": 9982
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8419671058654785,
      "learning_rate": 1.6062807221069352e-05,
      "loss": 0.5458,
      "step": 9983
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.517275094985962,
      "learning_rate": 1.606199729797538e-05,
      "loss": 0.4558,
      "step": 9984
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6402180194854736,
      "learning_rate": 1.6061187312008807e-05,
      "loss": 0.5452,
      "step": 9985
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7799880504608154,
      "learning_rate": 1.6060377263178026e-05,
      "loss": 0.4716,
      "step": 9986
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6644049882888794,
      "learning_rate": 1.6059567151491444e-05,
      "loss": 0.4754,
      "step": 9987
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6750857830047607,
      "learning_rate": 1.605875697695746e-05,
      "loss": 0.492,
      "step": 9988
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.650855302810669,
      "learning_rate": 1.6057946739584477e-05,
      "loss": 0.5783,
      "step": 9989
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5658001899719238,
      "learning_rate": 1.6057136439380897e-05,
      "loss": 0.4734,
      "step": 9990
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6096689701080322,
      "learning_rate": 1.605632607635513e-05,
      "loss": 0.5492,
      "step": 9991
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6123207807540894,
      "learning_rate": 1.6055515650515578e-05,
      "loss": 0.4929,
      "step": 9992
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7525479793548584,
      "learning_rate": 1.6054705161870643e-05,
      "loss": 0.5945,
      "step": 9993
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.582271933555603,
      "learning_rate": 1.6053894610428733e-05,
      "loss": 0.447,
      "step": 9994
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7333109378814697,
      "learning_rate": 1.6053083996198255e-05,
      "loss": 0.5374,
      "step": 9995
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5721049308776855,
      "learning_rate": 1.605227331918762e-05,
      "loss": 0.4806,
      "step": 9996
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7915513515472412,
      "learning_rate": 1.6051462579405227e-05,
      "loss": 0.6292,
      "step": 9997
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6759381294250488,
      "learning_rate": 1.6050651776859494e-05,
      "loss": 0.4733,
      "step": 9998
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7411435842514038,
      "learning_rate": 1.6049840911558826e-05,
      "loss": 0.5107,
      "step": 9999
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.612294316291809,
      "learning_rate": 1.6049029983511637e-05,
      "loss": 0.4721,
      "step": 10000
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7834123373031616,
      "learning_rate": 1.6048218992726327e-05,
      "loss": 0.526,
      "step": 10001
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.672507405281067,
      "learning_rate": 1.6047407939211322e-05,
      "loss": 0.4625,
      "step": 10002
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6986781358718872,
      "learning_rate": 1.604659682297502e-05,
      "loss": 0.5457,
      "step": 10003
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6359388828277588,
      "learning_rate": 1.6045785644025845e-05,
      "loss": 0.4374,
      "step": 10004
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6287822723388672,
      "learning_rate": 1.6044974402372202e-05,
      "loss": 0.5326,
      "step": 10005
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6895294189453125,
      "learning_rate": 1.604416309802251e-05,
      "loss": 0.4465,
      "step": 10006
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.7574615478515625,
      "learning_rate": 1.6043351730985177e-05,
      "loss": 1.0087,
      "step": 10007
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.417576789855957,
      "learning_rate": 1.604254030126863e-05,
      "loss": 1.1226,
      "step": 10008
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5816223621368408,
      "learning_rate": 1.604172880888127e-05,
      "loss": 0.5709,
      "step": 10009
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6481777429580688,
      "learning_rate": 1.6040917253831526e-05,
      "loss": 0.4555,
      "step": 10010
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6485861539840698,
      "learning_rate": 1.6040105636127808e-05,
      "loss": 0.5138,
      "step": 10011
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5898408889770508,
      "learning_rate": 1.603929395577854e-05,
      "loss": 0.4665,
      "step": 10012
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7105807065963745,
      "learning_rate": 1.6038482212792132e-05,
      "loss": 0.5434,
      "step": 10013
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5839475393295288,
      "learning_rate": 1.603767040717701e-05,
      "loss": 0.4447,
      "step": 10014
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9252078533172607,
      "learning_rate": 1.6036858538941586e-05,
      "loss": 0.5907,
      "step": 10015
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5633846521377563,
      "learning_rate": 1.6036046608094293e-05,
      "loss": 0.4645,
      "step": 10016
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.563413381576538,
      "learning_rate": 1.603523461464354e-05,
      "loss": 0.5058,
      "step": 10017
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5371050834655762,
      "learning_rate": 1.6034422558597755e-05,
      "loss": 0.4716,
      "step": 10018
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6992816925048828,
      "learning_rate": 1.6033610439965355e-05,
      "loss": 0.4887,
      "step": 10019
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5871096849441528,
      "learning_rate": 1.603279825875477e-05,
      "loss": 0.4714,
      "step": 10020
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.693895697593689,
      "learning_rate": 1.6031986014974418e-05,
      "loss": 0.5856,
      "step": 10021
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5423728227615356,
      "learning_rate": 1.6031173708632727e-05,
      "loss": 0.4479,
      "step": 10022
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8355382680892944,
      "learning_rate": 1.603036133973812e-05,
      "loss": 0.621,
      "step": 10023
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.53663969039917,
      "learning_rate": 1.6029548908299024e-05,
      "loss": 0.4506,
      "step": 10024
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6325088739395142,
      "learning_rate": 1.6028736414323862e-05,
      "loss": 0.5153,
      "step": 10025
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7185337543487549,
      "learning_rate": 1.6027923857821063e-05,
      "loss": 0.5132,
      "step": 10026
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6803998947143555,
      "learning_rate": 1.6027111238799057e-05,
      "loss": 0.5153,
      "step": 10027
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6141682863235474,
      "learning_rate": 1.602629855726627e-05,
      "loss": 0.4871,
      "step": 10028
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6382945775985718,
      "learning_rate": 1.6025485813231125e-05,
      "loss": 0.5186,
      "step": 10029
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6731352806091309,
      "learning_rate": 1.6024673006702062e-05,
      "loss": 0.4652,
      "step": 10030
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7169533967971802,
      "learning_rate": 1.6023860137687504e-05,
      "loss": 0.5638,
      "step": 10031
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5220890045166016,
      "learning_rate": 1.6023047206195884e-05,
      "loss": 0.4699,
      "step": 10032
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.621155023574829,
      "learning_rate": 1.602223421223563e-05,
      "loss": 0.5175,
      "step": 10033
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7411686182022095,
      "learning_rate": 1.6021421155815183e-05,
      "loss": 0.4899,
      "step": 10034
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.123382568359375,
      "learning_rate": 1.6020608036942965e-05,
      "loss": 0.5384,
      "step": 10035
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7343965768814087,
      "learning_rate": 1.601979485562742e-05,
      "loss": 0.485,
      "step": 10036
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7911486625671387,
      "learning_rate": 1.6018981611876972e-05,
      "loss": 0.5209,
      "step": 10037
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5812098979949951,
      "learning_rate": 1.601816830570006e-05,
      "loss": 0.4525,
      "step": 10038
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6302348375320435,
      "learning_rate": 1.6017354937105117e-05,
      "loss": 0.5214,
      "step": 10039
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.602812647819519,
      "learning_rate": 1.6016541506100587e-05,
      "loss": 0.4684,
      "step": 10040
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.725014090538025,
      "learning_rate": 1.6015728012694896e-05,
      "loss": 0.472,
      "step": 10041
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6577577590942383,
      "learning_rate": 1.6014914456896487e-05,
      "loss": 0.4744,
      "step": 10042
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6098514795303345,
      "learning_rate": 1.6014100838713796e-05,
      "loss": 0.5076,
      "step": 10043
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8125983476638794,
      "learning_rate": 1.6013287158155264e-05,
      "loss": 0.5096,
      "step": 10044
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6068086624145508,
      "learning_rate": 1.6012473415229326e-05,
      "loss": 0.4567,
      "step": 10045
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5740509033203125,
      "learning_rate": 1.6011659609944426e-05,
      "loss": 0.4689,
      "step": 10046
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6777745485305786,
      "learning_rate": 1.6010845742309e-05,
      "loss": 0.5548,
      "step": 10047
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.574130892753601,
      "learning_rate": 1.60100318123315e-05,
      "loss": 0.5026,
      "step": 10048
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6134390830993652,
      "learning_rate": 1.6009217820020354e-05,
      "loss": 0.5234,
      "step": 10049
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.646809458732605,
      "learning_rate": 1.6008403765384008e-05,
      "loss": 0.4848,
      "step": 10050
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5442217588424683,
      "learning_rate": 1.6007589648430906e-05,
      "loss": 0.5187,
      "step": 10051
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.512379765510559,
      "learning_rate": 1.60067754691695e-05,
      "loss": 0.4665,
      "step": 10052
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.795996904373169,
      "learning_rate": 1.600596122760822e-05,
      "loss": 0.5788,
      "step": 10053
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6674385070800781,
      "learning_rate": 1.600514692375552e-05,
      "loss": 0.4595,
      "step": 10054
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6520692110061646,
      "learning_rate": 1.6004332557619844e-05,
      "loss": 0.5564,
      "step": 10055
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5800116062164307,
      "learning_rate": 1.600351812920964e-05,
      "loss": 0.4619,
      "step": 10056
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6406344175338745,
      "learning_rate": 1.600270363853335e-05,
      "loss": 0.5108,
      "step": 10057
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8803867101669312,
      "learning_rate": 1.600188908559943e-05,
      "loss": 0.494,
      "step": 10058
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6157170534133911,
      "learning_rate": 1.6001074470416316e-05,
      "loss": 0.6118,
      "step": 10059
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6707439422607422,
      "learning_rate": 1.6000259792992467e-05,
      "loss": 0.5306,
      "step": 10060
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6913836002349854,
      "learning_rate": 1.5999445053336327e-05,
      "loss": 0.5503,
      "step": 10061
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5409729480743408,
      "learning_rate": 1.5998630251456352e-05,
      "loss": 0.4532,
      "step": 10062
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6873070001602173,
      "learning_rate": 1.5997815387360983e-05,
      "loss": 0.5179,
      "step": 10063
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.645294189453125,
      "learning_rate": 1.599700046105868e-05,
      "loss": 0.4824,
      "step": 10064
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.598500370979309,
      "learning_rate": 1.5996185472557892e-05,
      "loss": 0.543,
      "step": 10065
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6153925657272339,
      "learning_rate": 1.5995370421867078e-05,
      "loss": 0.4824,
      "step": 10066
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.620892882347107,
      "learning_rate": 1.599455530899468e-05,
      "loss": 0.4766,
      "step": 10067
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6466972827911377,
      "learning_rate": 1.5993740133949155e-05,
      "loss": 0.4423,
      "step": 10068
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7964799404144287,
      "learning_rate": 1.5992924896738965e-05,
      "loss": 0.5075,
      "step": 10069
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.48642098903656,
      "learning_rate": 1.599210959737256e-05,
      "loss": 0.4409,
      "step": 10070
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7275482416152954,
      "learning_rate": 1.59912942358584e-05,
      "loss": 0.5277,
      "step": 10071
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6095482110977173,
      "learning_rate": 1.5990478812204933e-05,
      "loss": 0.4582,
      "step": 10072
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5945874452590942,
      "learning_rate": 1.5989663326420622e-05,
      "loss": 0.4919,
      "step": 10073
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6971609592437744,
      "learning_rate": 1.598884777851393e-05,
      "loss": 0.4737,
      "step": 10074
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.68130362033844,
      "learning_rate": 1.59880321684933e-05,
      "loss": 0.5245,
      "step": 10075
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7754042148590088,
      "learning_rate": 1.5987216496367208e-05,
      "loss": 0.4847,
      "step": 10076
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7497881650924683,
      "learning_rate": 1.5986400762144105e-05,
      "loss": 0.5788,
      "step": 10077
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5879286527633667,
      "learning_rate": 1.5985584965832453e-05,
      "loss": 0.4634,
      "step": 10078
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7284263372421265,
      "learning_rate": 1.5984769107440713e-05,
      "loss": 0.5546,
      "step": 10079
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.662739634513855,
      "learning_rate": 1.598395318697735e-05,
      "loss": 0.4743,
      "step": 10080
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6792166233062744,
      "learning_rate": 1.598313720445082e-05,
      "loss": 0.4959,
      "step": 10081
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7326908111572266,
      "learning_rate": 1.598232115986959e-05,
      "loss": 0.4786,
      "step": 10082
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.674320101737976,
      "learning_rate": 1.598150505324213e-05,
      "loss": 0.5554,
      "step": 10083
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6051734685897827,
      "learning_rate": 1.5980688884576888e-05,
      "loss": 0.4598,
      "step": 10084
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7272288799285889,
      "learning_rate": 1.5979872653882343e-05,
      "loss": 0.547,
      "step": 10085
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7238198518753052,
      "learning_rate": 1.5979056361166958e-05,
      "loss": 0.5211,
      "step": 10086
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5307776927947998,
      "learning_rate": 1.5978240006439197e-05,
      "loss": 0.52,
      "step": 10087
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5599809885025024,
      "learning_rate": 1.5977423589707526e-05,
      "loss": 0.4186,
      "step": 10088
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7630488872528076,
      "learning_rate": 1.5976607110980413e-05,
      "loss": 0.5156,
      "step": 10089
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6488683223724365,
      "learning_rate": 1.597579057026633e-05,
      "loss": 0.5184,
      "step": 10090
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7126479148864746,
      "learning_rate": 1.597497396757374e-05,
      "loss": 0.501,
      "step": 10091
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5526115894317627,
      "learning_rate": 1.597415730291112e-05,
      "loss": 0.4481,
      "step": 10092
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6156026124954224,
      "learning_rate": 1.597334057628693e-05,
      "loss": 0.4973,
      "step": 10093
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7118717432022095,
      "learning_rate": 1.597252378770965e-05,
      "loss": 0.5038,
      "step": 10094
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8699262142181396,
      "learning_rate": 1.5971706937187747e-05,
      "loss": 0.6008,
      "step": 10095
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6560858488082886,
      "learning_rate": 1.5970890024729695e-05,
      "loss": 0.4518,
      "step": 10096
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7131755352020264,
      "learning_rate": 1.5970073050343963e-05,
      "loss": 0.5264,
      "step": 10097
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5319277048110962,
      "learning_rate": 1.596925601403903e-05,
      "loss": 0.4295,
      "step": 10098
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6588623523712158,
      "learning_rate": 1.5968438915823364e-05,
      "loss": 0.5342,
      "step": 10099
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.550620675086975,
      "learning_rate": 1.5967621755705445e-05,
      "loss": 0.4475,
      "step": 10100
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.7839598655700684,
      "learning_rate": 1.5966804533693744e-05,
      "loss": 0.5276,
      "step": 10101
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.619600772857666,
      "learning_rate": 1.596598724979674e-05,
      "loss": 0.4662,
      "step": 10102
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.800902009010315,
      "learning_rate": 1.5965169904022902e-05,
      "loss": 0.5545,
      "step": 10103
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5396965742111206,
      "learning_rate": 1.596435249638072e-05,
      "loss": 0.4392,
      "step": 10104
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6429243087768555,
      "learning_rate": 1.5963535026878662e-05,
      "loss": 0.5441,
      "step": 10105
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.598029613494873,
      "learning_rate": 1.5962717495525213e-05,
      "loss": 0.5085,
      "step": 10106
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6104849576950073,
      "learning_rate": 1.5961899902328845e-05,
      "loss": 0.4866,
      "step": 10107
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5732964277267456,
      "learning_rate": 1.5961082247298045e-05,
      "loss": 0.431,
      "step": 10108
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6624784469604492,
      "learning_rate": 1.5960264530441287e-05,
      "loss": 0.5288,
      "step": 10109
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7160321474075317,
      "learning_rate": 1.5959446751767054e-05,
      "loss": 0.4771,
      "step": 10110
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.687745451927185,
      "learning_rate": 1.5958628911283832e-05,
      "loss": 0.5261,
      "step": 10111
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5303821563720703,
      "learning_rate": 1.5957811009000096e-05,
      "loss": 0.47,
      "step": 10112
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7300430536270142,
      "learning_rate": 1.5956993044924334e-05,
      "loss": 0.543,
      "step": 10113
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6359424591064453,
      "learning_rate": 1.595617501906503e-05,
      "loss": 0.4632,
      "step": 10114
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6985408067703247,
      "learning_rate": 1.5955356931430666e-05,
      "loss": 0.5475,
      "step": 10115
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.669493317604065,
      "learning_rate": 1.5954538782029726e-05,
      "loss": 0.451,
      "step": 10116
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5479861497879028,
      "learning_rate": 1.59537205708707e-05,
      "loss": 0.479,
      "step": 10117
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6428214311599731,
      "learning_rate": 1.595290229796207e-05,
      "loss": 0.5199,
      "step": 10118
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.008601427078247,
      "learning_rate": 1.5952083963312325e-05,
      "loss": 1.0335,
      "step": 10119
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6724220514297485,
      "learning_rate": 1.5951265566929948e-05,
      "loss": 1.078,
      "step": 10120
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7292444705963135,
      "learning_rate": 1.5950447108823436e-05,
      "loss": 0.5164,
      "step": 10121
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7500584125518799,
      "learning_rate": 1.5949628589001264e-05,
      "loss": 0.4938,
      "step": 10122
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7408581972122192,
      "learning_rate": 1.5948810007471935e-05,
      "loss": 0.5471,
      "step": 10123
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7586851119995117,
      "learning_rate": 1.5947991364243933e-05,
      "loss": 0.5145,
      "step": 10124
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6201856136322021,
      "learning_rate": 1.594717265932575e-05,
      "loss": 0.5111,
      "step": 10125
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.678094744682312,
      "learning_rate": 1.5946353892725874e-05,
      "loss": 0.4721,
      "step": 10126
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.957004427909851,
      "learning_rate": 1.5945535064452805e-05,
      "loss": 0.5621,
      "step": 10127
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7942099571228027,
      "learning_rate": 1.5944716174515025e-05,
      "loss": 0.4426,
      "step": 10128
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6744868755340576,
      "learning_rate": 1.5943897222921033e-05,
      "loss": 0.5411,
      "step": 10129
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6917279958724976,
      "learning_rate": 1.5943078209679323e-05,
      "loss": 0.4519,
      "step": 10130
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.629891037940979,
      "learning_rate": 1.5942259134798388e-05,
      "loss": 0.4893,
      "step": 10131
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.534104585647583,
      "learning_rate": 1.5941439998286722e-05,
      "loss": 0.4476,
      "step": 10132
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6043027639389038,
      "learning_rate": 1.5940620800152824e-05,
      "loss": 0.4943,
      "step": 10133
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6114579439163208,
      "learning_rate": 1.593980154040519e-05,
      "loss": 0.4584,
      "step": 10134
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5644861459732056,
      "learning_rate": 1.5938982219052314e-05,
      "loss": 0.4885,
      "step": 10135
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6023344993591309,
      "learning_rate": 1.5938162836102694e-05,
      "loss": 0.4511,
      "step": 10136
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6332718133926392,
      "learning_rate": 1.5937343391564834e-05,
      "loss": 0.5453,
      "step": 10137
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5875307321548462,
      "learning_rate": 1.5936523885447225e-05,
      "loss": 0.4564,
      "step": 10138
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5767261981964111,
      "learning_rate": 1.593570431775837e-05,
      "loss": 0.4766,
      "step": 10139
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7227147817611694,
      "learning_rate": 1.5934884688506772e-05,
      "loss": 0.5223,
      "step": 10140
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6374455690383911,
      "learning_rate": 1.593406499770093e-05,
      "loss": 0.5853,
      "step": 10141
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7047446966171265,
      "learning_rate": 1.5933245245349343e-05,
      "loss": 0.5413,
      "step": 10142
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5614830255508423,
      "learning_rate": 1.593242543146052e-05,
      "loss": 0.4745,
      "step": 10143
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.567393183708191,
      "learning_rate": 1.593160555604295e-05,
      "loss": 0.439,
      "step": 10144
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5525068044662476,
      "learning_rate": 1.5930785619105152e-05,
      "loss": 0.4795,
      "step": 10145
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5803983211517334,
      "learning_rate": 1.5929965620655622e-05,
      "loss": 0.4357,
      "step": 10146
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6547017097473145,
      "learning_rate": 1.5929145560702864e-05,
      "loss": 0.5231,
      "step": 10147
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6051234006881714,
      "learning_rate": 1.5928325439255386e-05,
      "loss": 0.4824,
      "step": 10148
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6108192205429077,
      "learning_rate": 1.5927505256321694e-05,
      "loss": 0.5163,
      "step": 10149
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5924744606018066,
      "learning_rate": 1.5926685011910293e-05,
      "loss": 0.4519,
      "step": 10150
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7163742780685425,
      "learning_rate": 1.5925864706029694e-05,
      "loss": 0.5351,
      "step": 10151
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5278269052505493,
      "learning_rate": 1.59250443386884e-05,
      "loss": 0.4424,
      "step": 10152
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6636638641357422,
      "learning_rate": 1.592422390989492e-05,
      "loss": 0.5014,
      "step": 10153
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.619560956954956,
      "learning_rate": 1.5923403419657766e-05,
      "loss": 0.4765,
      "step": 10154
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7440720796585083,
      "learning_rate": 1.592258286798545e-05,
      "loss": 0.59,
      "step": 10155
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8322720527648926,
      "learning_rate": 1.592176225488647e-05,
      "loss": 0.4871,
      "step": 10156
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5349968671798706,
      "learning_rate": 1.5920941580369355e-05,
      "loss": 0.491,
      "step": 10157
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5822538137435913,
      "learning_rate": 1.5920120844442607e-05,
      "loss": 0.4895,
      "step": 10158
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.572849988937378,
      "learning_rate": 1.5919300047114736e-05,
      "loss": 0.4761,
      "step": 10159
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6323204040527344,
      "learning_rate": 1.5918479188394255e-05,
      "loss": 0.4345,
      "step": 10160
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9013636112213135,
      "learning_rate": 1.5917658268289687e-05,
      "loss": 1.0381,
      "step": 10161
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5907480716705322,
      "learning_rate": 1.5916837286809538e-05,
      "loss": 1.0927,
      "step": 10162
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7897335290908813,
      "learning_rate": 1.5916016243962323e-05,
      "loss": 0.5394,
      "step": 10163
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7046782970428467,
      "learning_rate": 1.591519513975656e-05,
      "loss": 0.489,
      "step": 10164
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7215150594711304,
      "learning_rate": 1.5914373974200766e-05,
      "loss": 0.5731,
      "step": 10165
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5745340585708618,
      "learning_rate": 1.5913552747303456e-05,
      "loss": 0.4395,
      "step": 10166
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5631321668624878,
      "learning_rate": 1.5912731459073147e-05,
      "loss": 0.4788,
      "step": 10167
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6937134265899658,
      "learning_rate": 1.5911910109518358e-05,
      "loss": 0.4736,
      "step": 10168
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.657598853111267,
      "learning_rate": 1.5911088698647606e-05,
      "loss": 0.5679,
      "step": 10169
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6223115921020508,
      "learning_rate": 1.5910267226469412e-05,
      "loss": 0.4658,
      "step": 10170
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6466059684753418,
      "learning_rate": 1.59094456929923e-05,
      "loss": 0.5743,
      "step": 10171
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.711835265159607,
      "learning_rate": 1.5908624098224784e-05,
      "loss": 0.4927,
      "step": 10172
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6130644083023071,
      "learning_rate": 1.590780244217539e-05,
      "loss": 0.4958,
      "step": 10173
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6590075492858887,
      "learning_rate": 1.5906980724852633e-05,
      "loss": 0.5157,
      "step": 10174
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6104975938796997,
      "learning_rate": 1.590615894626504e-05,
      "loss": 0.5403,
      "step": 10175
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5505179166793823,
      "learning_rate": 1.5905337106421138e-05,
      "loss": 0.4409,
      "step": 10176
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6031843423843384,
      "learning_rate": 1.5904515205329447e-05,
      "loss": 0.5135,
      "step": 10177
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7930480241775513,
      "learning_rate": 1.5903693242998488e-05,
      "loss": 0.5282,
      "step": 10178
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7593295574188232,
      "learning_rate": 1.5902871219436794e-05,
      "loss": 0.5088,
      "step": 10179
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8494296073913574,
      "learning_rate": 1.5902049134652884e-05,
      "loss": 0.4811,
      "step": 10180
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6065386533737183,
      "learning_rate": 1.5901226988655288e-05,
      "loss": 0.4996,
      "step": 10181
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5620310306549072,
      "learning_rate": 1.590040478145253e-05,
      "loss": 0.4658,
      "step": 10182
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6374843120574951,
      "learning_rate": 1.589958251305314e-05,
      "loss": 0.5004,
      "step": 10183
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6882579326629639,
      "learning_rate": 1.5898760183465644e-05,
      "loss": 0.4843,
      "step": 10184
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1550168991088867,
      "learning_rate": 1.5897937792698576e-05,
      "loss": 1.0123,
      "step": 10185
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7951760292053223,
      "learning_rate": 1.5897115340760455e-05,
      "loss": 1.0844,
      "step": 10186
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5896048545837402,
      "learning_rate": 1.5896292827659824e-05,
      "loss": 0.4939,
      "step": 10187
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.714677095413208,
      "learning_rate": 1.5895470253405205e-05,
      "loss": 0.5123,
      "step": 10188
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.73264741897583,
      "learning_rate": 1.5894647618005132e-05,
      "loss": 0.5493,
      "step": 10189
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.589143991470337,
      "learning_rate": 1.589382492146814e-05,
      "loss": 0.4569,
      "step": 10190
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.593988299369812,
      "learning_rate": 1.5893002163802756e-05,
      "loss": 0.4964,
      "step": 10191
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6088584661483765,
      "learning_rate": 1.589217934501752e-05,
      "loss": 0.5219,
      "step": 10192
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.607759714126587,
      "learning_rate": 1.5891356465120958e-05,
      "loss": 0.4877,
      "step": 10193
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7357949018478394,
      "learning_rate": 1.5890533524121606e-05,
      "loss": 0.5503,
      "step": 10194
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7047202587127686,
      "learning_rate": 1.5889710522028007e-05,
      "loss": 0.5572,
      "step": 10195
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.682813286781311,
      "learning_rate": 1.588888745884869e-05,
      "loss": 0.464,
      "step": 10196
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.936113715171814,
      "learning_rate": 1.5888064334592196e-05,
      "loss": 0.9996,
      "step": 10197
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7035480737686157,
      "learning_rate": 1.5887241149267058e-05,
      "loss": 1.0838,
      "step": 10198
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7206714153289795,
      "learning_rate": 1.588641790288181e-05,
      "loss": 0.4935,
      "step": 10199
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6593488454818726,
      "learning_rate": 1.5885594595445e-05,
      "loss": 0.4649,
      "step": 10200
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9556275606155396,
      "learning_rate": 1.5884771226965166e-05,
      "loss": 0.6153,
      "step": 10201
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6220686435699463,
      "learning_rate": 1.5883947797450838e-05,
      "loss": 0.4736,
      "step": 10202
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5281940698623657,
      "learning_rate": 1.5883124306910563e-05,
      "loss": 0.4455,
      "step": 10203
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6273928880691528,
      "learning_rate": 1.5882300755352885e-05,
      "loss": 0.4812,
      "step": 10204
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6265074014663696,
      "learning_rate": 1.588147714278634e-05,
      "loss": 0.5215,
      "step": 10205
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.614687204360962,
      "learning_rate": 1.5880653469219475e-05,
      "loss": 0.5414,
      "step": 10206
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6399973630905151,
      "learning_rate": 1.5879829734660826e-05,
      "loss": 0.5757,
      "step": 10207
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5816222429275513,
      "learning_rate": 1.587900593911894e-05,
      "loss": 0.4304,
      "step": 10208
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5909347534179688,
      "learning_rate": 1.5878182082602362e-05,
      "loss": 0.4895,
      "step": 10209
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6350274085998535,
      "learning_rate": 1.5877358165119637e-05,
      "loss": 0.5495,
      "step": 10210
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6562235355377197,
      "learning_rate": 1.587653418667931e-05,
      "loss": 0.5152,
      "step": 10211
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5193504095077515,
      "learning_rate": 1.5875710147289927e-05,
      "loss": 0.4709,
      "step": 10212
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6091424226760864,
      "learning_rate": 1.5874886046960033e-05,
      "loss": 0.5046,
      "step": 10213
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5588287115097046,
      "learning_rate": 1.5874061885698176e-05,
      "loss": 0.4405,
      "step": 10214
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6669999361038208,
      "learning_rate": 1.5873237663512905e-05,
      "loss": 0.5586,
      "step": 10215
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6088809967041016,
      "learning_rate": 1.5872413380412766e-05,
      "loss": 0.4651,
      "step": 10216
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6853708028793335,
      "learning_rate": 1.587158903640631e-05,
      "loss": 0.5788,
      "step": 10217
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.705653190612793,
      "learning_rate": 1.5870764631502085e-05,
      "loss": 0.4853,
      "step": 10218
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.290862083435059,
      "learning_rate": 1.5869940165708648e-05,
      "loss": 0.9918,
      "step": 10219
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.693740963935852,
      "learning_rate": 1.5869115639034542e-05,
      "loss": 1.0873,
      "step": 10220
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.64723801612854,
      "learning_rate": 1.586829105148832e-05,
      "loss": 0.5024,
      "step": 10221
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6926543712615967,
      "learning_rate": 1.5867466403078538e-05,
      "loss": 0.4577,
      "step": 10222
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.680738091468811,
      "learning_rate": 1.5866641693813747e-05,
      "loss": 0.5247,
      "step": 10223
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5652241706848145,
      "learning_rate": 1.58658169237025e-05,
      "loss": 0.4871,
      "step": 10224
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4938703775405884,
      "learning_rate": 1.5864992092753353e-05,
      "loss": 0.487,
      "step": 10225
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6514232158660889,
      "learning_rate": 1.5864167200974856e-05,
      "loss": 0.4743,
      "step": 10226
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.605315089225769,
      "learning_rate": 1.5863342248375572e-05,
      "loss": 0.5252,
      "step": 10227
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5757108926773071,
      "learning_rate": 1.586251723496405e-05,
      "loss": 0.4649,
      "step": 10228
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5335098505020142,
      "learning_rate": 1.586169216074885e-05,
      "loss": 0.4984,
      "step": 10229
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5457453727722168,
      "learning_rate": 1.5860867025738528e-05,
      "loss": 0.4727,
      "step": 10230
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7694088220596313,
      "learning_rate": 1.5860041829941645e-05,
      "loss": 0.5784,
      "step": 10231
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5747957229614258,
      "learning_rate": 1.585921657336676e-05,
      "loss": 0.4579,
      "step": 10232
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6357849836349487,
      "learning_rate": 1.5858391256022426e-05,
      "loss": 0.5191,
      "step": 10233
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5806870460510254,
      "learning_rate": 1.5857565877917205e-05,
      "loss": 0.5136,
      "step": 10234
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6580027341842651,
      "learning_rate": 1.585674043905966e-05,
      "loss": 0.5448,
      "step": 10235
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.623314380645752,
      "learning_rate": 1.5855914939458353e-05,
      "loss": 0.4682,
      "step": 10236
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.683274507522583,
      "learning_rate": 1.5855089379121844e-05,
      "loss": 0.5405,
      "step": 10237
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.52704656124115,
      "learning_rate": 1.5854263758058696e-05,
      "loss": 0.4773,
      "step": 10238
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5701751708984375,
      "learning_rate": 1.585343807627747e-05,
      "loss": 0.5145,
      "step": 10239
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.521929383277893,
      "learning_rate": 1.585261233378673e-05,
      "loss": 0.4715,
      "step": 10240
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.743493914604187,
      "learning_rate": 1.5851786530595043e-05,
      "loss": 0.5182,
      "step": 10241
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5800983905792236,
      "learning_rate": 1.585096066671097e-05,
      "loss": 0.4382,
      "step": 10242
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.642896056175232,
      "learning_rate": 1.5850134742143084e-05,
      "loss": 0.5228,
      "step": 10243
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6755436658859253,
      "learning_rate": 1.584930875689994e-05,
      "loss": 0.4786,
      "step": 10244
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.700948715209961,
      "learning_rate": 1.5848482710990115e-05,
      "loss": 0.5236,
      "step": 10245
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5531307458877563,
      "learning_rate": 1.584765660442217e-05,
      "loss": 0.4558,
      "step": 10246
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.896139144897461,
      "learning_rate": 1.5846830437204676e-05,
      "loss": 0.5318,
      "step": 10247
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5417536497116089,
      "learning_rate": 1.5846004209346197e-05,
      "loss": 0.4642,
      "step": 10248
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.706957459449768,
      "learning_rate": 1.5845177920855313e-05,
      "loss": 0.5326,
      "step": 10249
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6833447217941284,
      "learning_rate": 1.584435157174058e-05,
      "loss": 0.4672,
      "step": 10250
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6510295867919922,
      "learning_rate": 1.584352516201058e-05,
      "loss": 0.5524,
      "step": 10251
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6435238122940063,
      "learning_rate": 1.584269869167388e-05,
      "loss": 0.5048,
      "step": 10252
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6482475996017456,
      "learning_rate": 1.5841872160739052e-05,
      "loss": 0.6023,
      "step": 10253
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5635288953781128,
      "learning_rate": 1.5841045569214664e-05,
      "loss": 0.4505,
      "step": 10254
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6532357931137085,
      "learning_rate": 1.5840218917109295e-05,
      "loss": 0.543,
      "step": 10255
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5514544248580933,
      "learning_rate": 1.5839392204431518e-05,
      "loss": 0.451,
      "step": 10256
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5534956455230713,
      "learning_rate": 1.5838565431189906e-05,
      "loss": 0.5543,
      "step": 10257
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6226551532745361,
      "learning_rate": 1.5837738597393032e-05,
      "loss": 0.5081,
      "step": 10258
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.681150197982788,
      "learning_rate": 1.5836911703049477e-05,
      "loss": 0.5434,
      "step": 10259
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6234989166259766,
      "learning_rate": 1.5836084748167813e-05,
      "loss": 0.4921,
      "step": 10260
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6076819896697998,
      "learning_rate": 1.583525773275662e-05,
      "loss": 0.5475,
      "step": 10261
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5282829999923706,
      "learning_rate": 1.5834430656824466e-05,
      "loss": 0.4524,
      "step": 10262
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5970536470413208,
      "learning_rate": 1.5833603520379945e-05,
      "loss": 0.5228,
      "step": 10263
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6061581373214722,
      "learning_rate": 1.583277632343162e-05,
      "loss": 0.4959,
      "step": 10264
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.506402969360352,
      "learning_rate": 1.583194906598808e-05,
      "loss": 0.5023,
      "step": 10265
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6099674701690674,
      "learning_rate": 1.5831121748057902e-05,
      "loss": 0.5004,
      "step": 10266
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6847915649414062,
      "learning_rate": 1.5830294369649668e-05,
      "loss": 0.5371,
      "step": 10267
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6177207231521606,
      "learning_rate": 1.582946693077196e-05,
      "loss": 0.4825,
      "step": 10268
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6731914281845093,
      "learning_rate": 1.5828639431433358e-05,
      "loss": 0.5218,
      "step": 10269
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.666956901550293,
      "learning_rate": 1.582781187164244e-05,
      "loss": 0.4417,
      "step": 10270
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6108072996139526,
      "learning_rate": 1.5826984251407797e-05,
      "loss": 0.5212,
      "step": 10271
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5811207294464111,
      "learning_rate": 1.582615657073801e-05,
      "loss": 0.5015,
      "step": 10272
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6095499992370605,
      "learning_rate": 1.582532882964166e-05,
      "loss": 0.5227,
      "step": 10273
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.613734245300293,
      "learning_rate": 1.582450102812734e-05,
      "loss": 0.4415,
      "step": 10274
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7358664274215698,
      "learning_rate": 1.5823673166203624e-05,
      "loss": 0.4931,
      "step": 10275
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5472462177276611,
      "learning_rate": 1.582284524387911e-05,
      "loss": 0.4625,
      "step": 10276
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6930320262908936,
      "learning_rate": 1.5822017261162377e-05,
      "loss": 0.4906,
      "step": 10277
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7032537460327148,
      "learning_rate": 1.5821189218062015e-05,
      "loss": 0.4532,
      "step": 10278
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5992177724838257,
      "learning_rate": 1.5820361114586612e-05,
      "loss": 0.5225,
      "step": 10279
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6301405429840088,
      "learning_rate": 1.5819532950744756e-05,
      "loss": 0.4729,
      "step": 10280
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7226874828338623,
      "learning_rate": 1.5818704726545044e-05,
      "loss": 0.5299,
      "step": 10281
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.592334508895874,
      "learning_rate": 1.5817876441996053e-05,
      "loss": 0.444,
      "step": 10282
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7058534622192383,
      "learning_rate": 1.5817048097106384e-05,
      "loss": 0.5335,
      "step": 10283
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7045572996139526,
      "learning_rate": 1.581621969188462e-05,
      "loss": 0.4969,
      "step": 10284
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6303906440734863,
      "learning_rate": 1.581539122633936e-05,
      "loss": 0.5003,
      "step": 10285
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6522632837295532,
      "learning_rate": 1.5814562700479194e-05,
      "loss": 0.4591,
      "step": 10286
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.69056236743927,
      "learning_rate": 1.5813734114312714e-05,
      "loss": 0.5347,
      "step": 10287
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.75936758518219,
      "learning_rate": 1.5812905467848514e-05,
      "loss": 0.574,
      "step": 10288
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.685603141784668,
      "learning_rate": 1.581207676109519e-05,
      "loss": 0.5345,
      "step": 10289
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5625711679458618,
      "learning_rate": 1.5811247994061333e-05,
      "loss": 0.4385,
      "step": 10290
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5761111974716187,
      "learning_rate": 1.5810419166755545e-05,
      "loss": 0.4642,
      "step": 10291
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5604289770126343,
      "learning_rate": 1.5809590279186418e-05,
      "loss": 0.4715,
      "step": 10292
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6575706005096436,
      "learning_rate": 1.580876133136255e-05,
      "loss": 0.5082,
      "step": 10293
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5685464143753052,
      "learning_rate": 1.5807932323292536e-05,
      "loss": 0.4389,
      "step": 10294
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8545266389846802,
      "learning_rate": 1.5807103254984982e-05,
      "loss": 0.6512,
      "step": 10295
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5968619585037231,
      "learning_rate": 1.5806274126448477e-05,
      "loss": 0.4519,
      "step": 10296
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7570115327835083,
      "learning_rate": 1.5805444937691625e-05,
      "loss": 0.545,
      "step": 10297
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5680562257766724,
      "learning_rate": 1.5804615688723027e-05,
      "loss": 0.462,
      "step": 10298
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6882036924362183,
      "learning_rate": 1.580378637955128e-05,
      "loss": 0.5577,
      "step": 10299
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.553152084350586,
      "learning_rate": 1.580295701018499e-05,
      "loss": 0.4465,
      "step": 10300
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5962986946105957,
      "learning_rate": 1.5802127580632753e-05,
      "loss": 0.4873,
      "step": 10301
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6917102336883545,
      "learning_rate": 1.5801298090903178e-05,
      "loss": 0.5055,
      "step": 10302
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6604382991790771,
      "learning_rate": 1.5800468541004864e-05,
      "loss": 0.5349,
      "step": 10303
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5762439966201782,
      "learning_rate": 1.579963893094642e-05,
      "loss": 0.4785,
      "step": 10304
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1717169284820557,
      "learning_rate": 1.579880926073644e-05,
      "loss": 0.5136,
      "step": 10305
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6890966892242432,
      "learning_rate": 1.5797979530383537e-05,
      "loss": 0.4764,
      "step": 10306
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6276735067367554,
      "learning_rate": 1.5797149739896316e-05,
      "loss": 0.5469,
      "step": 10307
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6590825319290161,
      "learning_rate": 1.579631988928338e-05,
      "loss": 0.4515,
      "step": 10308
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5854305028915405,
      "learning_rate": 1.579548997855334e-05,
      "loss": 0.498,
      "step": 10309
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5470092296600342,
      "learning_rate": 1.57946600077148e-05,
      "loss": 0.4531,
      "step": 10310
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.523079514503479,
      "learning_rate": 1.5793829976776372e-05,
      "loss": 0.5245,
      "step": 10311
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6105164289474487,
      "learning_rate": 1.579299988574666e-05,
      "loss": 0.5007,
      "step": 10312
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.632022738456726,
      "learning_rate": 1.5792169734634277e-05,
      "loss": 0.5692,
      "step": 10313
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5788483619689941,
      "learning_rate": 1.579133952344783e-05,
      "loss": 0.474,
      "step": 10314
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5776902437210083,
      "learning_rate": 1.579050925219593e-05,
      "loss": 0.472,
      "step": 10315
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5444462299346924,
      "learning_rate": 1.5789678920887195e-05,
      "loss": 0.4757,
      "step": 10316
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5613685846328735,
      "learning_rate": 1.578884852953023e-05,
      "loss": 0.5262,
      "step": 10317
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5699937343597412,
      "learning_rate": 1.5788018078133647e-05,
      "loss": 0.4394,
      "step": 10318
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7803720235824585,
      "learning_rate": 1.578718756670606e-05,
      "loss": 0.5494,
      "step": 10319
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5450810194015503,
      "learning_rate": 1.5786356995256084e-05,
      "loss": 0.448,
      "step": 10320
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5943676233291626,
      "learning_rate": 1.5785526363792337e-05,
      "loss": 0.5654,
      "step": 10321
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.523394227027893,
      "learning_rate": 1.5784695672323428e-05,
      "loss": 0.4533,
      "step": 10322
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5260423421859741,
      "learning_rate": 1.5783864920857972e-05,
      "loss": 0.5085,
      "step": 10323
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5097479820251465,
      "learning_rate": 1.578303410940459e-05,
      "loss": 0.4579,
      "step": 10324
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8105868101119995,
      "learning_rate": 1.5782203237971895e-05,
      "loss": 0.9701,
      "step": 10325
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5395400524139404,
      "learning_rate": 1.578137230656851e-05,
      "loss": 1.055,
      "step": 10326
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6727303266525269,
      "learning_rate": 1.5780541315203045e-05,
      "loss": 0.5172,
      "step": 10327
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6733752489089966,
      "learning_rate": 1.5779710263884127e-05,
      "loss": 0.449,
      "step": 10328
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6587356328964233,
      "learning_rate": 1.577887915262037e-05,
      "loss": 0.4849,
      "step": 10329
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5377615690231323,
      "learning_rate": 1.577804798142039e-05,
      "loss": 0.4942,
      "step": 10330
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5791208744049072,
      "learning_rate": 1.5777216750292822e-05,
      "loss": 0.4833,
      "step": 10331
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6366267204284668,
      "learning_rate": 1.5776385459246274e-05,
      "loss": 0.4589,
      "step": 10332
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.914374589920044,
      "learning_rate": 1.577555410828937e-05,
      "loss": 0.5788,
      "step": 10333
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.649903416633606,
      "learning_rate": 1.5774722697430737e-05,
      "loss": 0.4644,
      "step": 10334
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6708202362060547,
      "learning_rate": 1.5773891226678993e-05,
      "loss": 0.5075,
      "step": 10335
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5926315784454346,
      "learning_rate": 1.5773059696042766e-05,
      "loss": 0.4487,
      "step": 10336
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.778686761856079,
      "learning_rate": 1.5772228105530677e-05,
      "loss": 0.5641,
      "step": 10337
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6702896356582642,
      "learning_rate": 1.5771396455151354e-05,
      "loss": 0.4453,
      "step": 10338
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9843356609344482,
      "learning_rate": 1.5770564744913418e-05,
      "loss": 0.6089,
      "step": 10339
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5321327447891235,
      "learning_rate": 1.57697329748255e-05,
      "loss": 0.4428,
      "step": 10340
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.622351884841919,
      "learning_rate": 1.5768901144896224e-05,
      "loss": 0.5048,
      "step": 10341
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6233232021331787,
      "learning_rate": 1.576806925513422e-05,
      "loss": 0.4439,
      "step": 10342
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6358938217163086,
      "learning_rate": 1.5767237305548113e-05,
      "loss": 0.553,
      "step": 10343
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6844967603683472,
      "learning_rate": 1.5766405296146533e-05,
      "loss": 0.5008,
      "step": 10344
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.614688754081726,
      "learning_rate": 1.576557322693811e-05,
      "loss": 0.5301,
      "step": 10345
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5797771215438843,
      "learning_rate": 1.576474109793147e-05,
      "loss": 0.4561,
      "step": 10346
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.687915325164795,
      "learning_rate": 1.576390890913525e-05,
      "loss": 0.5157,
      "step": 10347
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5166118144989014,
      "learning_rate": 1.5763076660558074e-05,
      "loss": 0.4476,
      "step": 10348
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5986056327819824,
      "learning_rate": 1.576224435220858e-05,
      "loss": 0.5619,
      "step": 10349
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6089009046554565,
      "learning_rate": 1.57614119840954e-05,
      "loss": 0.4526,
      "step": 10350
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.540781855583191,
      "learning_rate": 1.5760579556227162e-05,
      "loss": 0.5154,
      "step": 10351
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.599524736404419,
      "learning_rate": 1.57597470686125e-05,
      "loss": 0.4619,
      "step": 10352
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.682951807975769,
      "learning_rate": 1.5758914521260055e-05,
      "loss": 0.4885,
      "step": 10353
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6509093046188354,
      "learning_rate": 1.5758081914178457e-05,
      "loss": 0.4677,
      "step": 10354
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7371394634246826,
      "learning_rate": 1.5757249247376338e-05,
      "loss": 0.5769,
      "step": 10355
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6370176076889038,
      "learning_rate": 1.5756416520862342e-05,
      "loss": 0.4307,
      "step": 10356
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.655914545059204,
      "learning_rate": 1.5755583734645098e-05,
      "loss": 0.5001,
      "step": 10357
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5925382375717163,
      "learning_rate": 1.575475088873325e-05,
      "loss": 0.4458,
      "step": 10358
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7151004076004028,
      "learning_rate": 1.5753917983135433e-05,
      "loss": 0.5447,
      "step": 10359
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5396233797073364,
      "learning_rate": 1.5753085017860285e-05,
      "loss": 0.4622,
      "step": 10360
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.595626711845398,
      "learning_rate": 1.5752251992916443e-05,
      "loss": 0.5314,
      "step": 10361
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6352194547653198,
      "learning_rate": 1.575141890831255e-05,
      "loss": 0.4638,
      "step": 10362
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8588941097259521,
      "learning_rate": 1.5750585764057248e-05,
      "loss": 0.5508,
      "step": 10363
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.668738603591919,
      "learning_rate": 1.5749752560159176e-05,
      "loss": 0.5093,
      "step": 10364
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6751569509506226,
      "learning_rate": 1.5748919296626977e-05,
      "loss": 0.5457,
      "step": 10365
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5760549306869507,
      "learning_rate": 1.5748085973469293e-05,
      "loss": 0.4552,
      "step": 10366
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.580854892730713,
      "learning_rate": 1.5747252590694763e-05,
      "loss": 0.5184,
      "step": 10367
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5740400552749634,
      "learning_rate": 1.574641914831203e-05,
      "loss": 0.4506,
      "step": 10368
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.73966646194458,
      "learning_rate": 1.5745585646329747e-05,
      "loss": 0.5331,
      "step": 10369
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5524444580078125,
      "learning_rate": 1.574475208475655e-05,
      "loss": 0.4642,
      "step": 10370
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5691540241241455,
      "learning_rate": 1.5743918463601088e-05,
      "loss": 0.5209,
      "step": 10371
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4613456726074219,
      "learning_rate": 1.574308478287201e-05,
      "loss": 0.4396,
      "step": 10372
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5437753200531006,
      "learning_rate": 1.5742251042577954e-05,
      "loss": 0.5134,
      "step": 10373
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6426606178283691,
      "learning_rate": 1.5741417242727575e-05,
      "loss": 0.4575,
      "step": 10374
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5477070808410645,
      "learning_rate": 1.5740583383329517e-05,
      "loss": 0.533,
      "step": 10375
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6201671361923218,
      "learning_rate": 1.5739749464392432e-05,
      "loss": 0.5097,
      "step": 10376
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6197277307510376,
      "learning_rate": 1.5738915485924967e-05,
      "loss": 0.531,
      "step": 10377
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6625490188598633,
      "learning_rate": 1.573808144793577e-05,
      "loss": 0.5369,
      "step": 10378
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.597995400428772,
      "learning_rate": 1.573724735043349e-05,
      "loss": 0.4931,
      "step": 10379
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7444367408752441,
      "learning_rate": 1.5736413193426787e-05,
      "loss": 0.493,
      "step": 10380
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.57816743850708,
      "learning_rate": 1.57355789769243e-05,
      "loss": 0.5085,
      "step": 10381
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.540221095085144,
      "learning_rate": 1.573474470093469e-05,
      "loss": 0.4819,
      "step": 10382
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6124906539916992,
      "learning_rate": 1.5733910365466605e-05,
      "loss": 0.5357,
      "step": 10383
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5469485521316528,
      "learning_rate": 1.5733075970528703e-05,
      "loss": 0.4629,
      "step": 10384
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6593434810638428,
      "learning_rate": 1.5732241516129633e-05,
      "loss": 0.5127,
      "step": 10385
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6608822345733643,
      "learning_rate": 1.5731407002278057e-05,
      "loss": 0.4659,
      "step": 10386
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.615122675895691,
      "learning_rate": 1.573057242898262e-05,
      "loss": 0.5075,
      "step": 10387
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8082258701324463,
      "learning_rate": 1.5729737796251984e-05,
      "loss": 0.5096,
      "step": 10388
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.640006184577942,
      "learning_rate": 1.572890310409481e-05,
      "loss": 0.4972,
      "step": 10389
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6890780925750732,
      "learning_rate": 1.5728068352519746e-05,
      "loss": 0.4741,
      "step": 10390
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8125187158584595,
      "learning_rate": 1.5727233541535452e-05,
      "loss": 0.5578,
      "step": 10391
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5734387636184692,
      "learning_rate": 1.572639867115059e-05,
      "loss": 0.4603,
      "step": 10392
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9209243059158325,
      "learning_rate": 1.5725563741373816e-05,
      "loss": 0.5779,
      "step": 10393
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4980664253234863,
      "learning_rate": 1.572472875221379e-05,
      "loss": 0.4301,
      "step": 10394
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5109212398529053,
      "learning_rate": 1.5723893703679175e-05,
      "loss": 0.4577,
      "step": 10395
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5561107397079468,
      "learning_rate": 1.5723058595778625e-05,
      "loss": 0.5003,
      "step": 10396
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.556822419166565,
      "learning_rate": 1.572222342852081e-05,
      "loss": 0.5045,
      "step": 10397
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5981053113937378,
      "learning_rate": 1.5721388201914388e-05,
      "loss": 0.4434,
      "step": 10398
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6460870504379272,
      "learning_rate": 1.5720552915968017e-05,
      "loss": 0.5621,
      "step": 10399
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.528468132019043,
      "learning_rate": 1.571971757069037e-05,
      "loss": 0.4437,
      "step": 10400
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6547565460205078,
      "learning_rate": 1.57188821660901e-05,
      "loss": 0.5283,
      "step": 10401
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6073143482208252,
      "learning_rate": 1.571804670217588e-05,
      "loss": 0.4891,
      "step": 10402
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6333484649658203,
      "learning_rate": 1.571721117895637e-05,
      "loss": 0.4995,
      "step": 10403
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6738736629486084,
      "learning_rate": 1.571637559644024e-05,
      "loss": 0.4394,
      "step": 10404
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6842454671859741,
      "learning_rate": 1.5715539954636153e-05,
      "loss": 0.5151,
      "step": 10405
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6187503337860107,
      "learning_rate": 1.571470425355278e-05,
      "loss": 0.4678,
      "step": 10406
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6517330408096313,
      "learning_rate": 1.5713868493198782e-05,
      "loss": 0.5395,
      "step": 10407
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6866552829742432,
      "learning_rate": 1.5713032673582834e-05,
      "loss": 0.4525,
      "step": 10408
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5886123180389404,
      "learning_rate": 1.57121967947136e-05,
      "loss": 0.5248,
      "step": 10409
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5359570980072021,
      "learning_rate": 1.5711360856599748e-05,
      "loss": 0.4268,
      "step": 10410
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.642510175704956,
      "learning_rate": 1.5710524859249955e-05,
      "loss": 0.5369,
      "step": 10411
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5593734979629517,
      "learning_rate": 1.5709688802672887e-05,
      "loss": 0.4403,
      "step": 10412
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5992743968963623,
      "learning_rate": 1.5708852686877215e-05,
      "loss": 0.535,
      "step": 10413
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6264206171035767,
      "learning_rate": 1.5708016511871612e-05,
      "loss": 0.4722,
      "step": 10414
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5664819478988647,
      "learning_rate": 1.570718027766475e-05,
      "loss": 0.5596,
      "step": 10415
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.518876552581787,
      "learning_rate": 1.5706343984265302e-05,
      "loss": 0.4565,
      "step": 10416
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5795557498931885,
      "learning_rate": 1.570550763168194e-05,
      "loss": 0.533,
      "step": 10417
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.581282138824463,
      "learning_rate": 1.5704671219923343e-05,
      "loss": 0.4772,
      "step": 10418
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5283045768737793,
      "learning_rate": 1.5703834748998184e-05,
      "loss": 0.4867,
      "step": 10419
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.573305368423462,
      "learning_rate": 1.5702998218915136e-05,
      "loss": 0.4577,
      "step": 10420
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6370052099227905,
      "learning_rate": 1.5702161629682877e-05,
      "loss": 0.4921,
      "step": 10421
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5654010772705078,
      "learning_rate": 1.5701324981310086e-05,
      "loss": 0.4556,
      "step": 10422
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.803943395614624,
      "learning_rate": 1.5700488273805433e-05,
      "loss": 0.5327,
      "step": 10423
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5721626281738281,
      "learning_rate": 1.5699651507177604e-05,
      "loss": 0.4562,
      "step": 10424
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.589508295059204,
      "learning_rate": 1.5698814681435276e-05,
      "loss": 0.5127,
      "step": 10425
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6463223695755005,
      "learning_rate": 1.5697977796587126e-05,
      "loss": 0.4831,
      "step": 10426
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7387194633483887,
      "learning_rate": 1.5697140852641835e-05,
      "loss": 0.5543,
      "step": 10427
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6718895435333252,
      "learning_rate": 1.5696303849608082e-05,
      "loss": 0.4854,
      "step": 10428
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6775141954421997,
      "learning_rate": 1.569546678749455e-05,
      "loss": 0.5533,
      "step": 10429
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5880571603775024,
      "learning_rate": 1.569462966630992e-05,
      "loss": 0.4617,
      "step": 10430
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6401872634887695,
      "learning_rate": 1.5693792486062874e-05,
      "loss": 0.5096,
      "step": 10431
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5938323736190796,
      "learning_rate": 1.5692955246762098e-05,
      "loss": 0.4785,
      "step": 10432
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5858018398284912,
      "learning_rate": 1.5692117948416267e-05,
      "loss": 0.4869,
      "step": 10433
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5517776012420654,
      "learning_rate": 1.5691280591034073e-05,
      "loss": 0.4516,
      "step": 10434
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.753596305847168,
      "learning_rate": 1.56904431746242e-05,
      "loss": 0.5841,
      "step": 10435
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6217986345291138,
      "learning_rate": 1.568960569919533e-05,
      "loss": 0.4539,
      "step": 10436
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7343140840530396,
      "learning_rate": 1.568876816475615e-05,
      "loss": 0.5659,
      "step": 10437
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6114599704742432,
      "learning_rate": 1.5687930571315352e-05,
      "loss": 0.4508,
      "step": 10438
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6577036380767822,
      "learning_rate": 1.568709291888161e-05,
      "loss": 0.5409,
      "step": 10439
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.571070909500122,
      "learning_rate": 1.5686255207463627e-05,
      "loss": 0.4585,
      "step": 10440
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6455563306808472,
      "learning_rate": 1.5685417437070082e-05,
      "loss": 0.5388,
      "step": 10441
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5945265293121338,
      "learning_rate": 1.5684579607709668e-05,
      "loss": 0.4538,
      "step": 10442
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.674443244934082,
      "learning_rate": 1.5683741719391075e-05,
      "loss": 0.5063,
      "step": 10443
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5840165615081787,
      "learning_rate": 1.568290377212299e-05,
      "loss": 0.4456,
      "step": 10444
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5561994314193726,
      "learning_rate": 1.5682065765914107e-05,
      "loss": 0.5194,
      "step": 10445
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.570905089378357,
      "learning_rate": 1.5681227700773114e-05,
      "loss": 0.471,
      "step": 10446
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.699928641319275,
      "learning_rate": 1.5680389576708704e-05,
      "loss": 0.5201,
      "step": 10447
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6801598072052002,
      "learning_rate": 1.567955139372957e-05,
      "loss": 0.4752,
      "step": 10448
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6734286546707153,
      "learning_rate": 1.567871315184441e-05,
      "loss": 0.5158,
      "step": 10449
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.565731406211853,
      "learning_rate": 1.5677874851061914e-05,
      "loss": 0.4653,
      "step": 10450
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6674610376358032,
      "learning_rate": 1.567703649139077e-05,
      "loss": 0.5333,
      "step": 10451
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6242156028747559,
      "learning_rate": 1.567619807283969e-05,
      "loss": 0.461,
      "step": 10452
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7982655763626099,
      "learning_rate": 1.567535959541735e-05,
      "loss": 0.5031,
      "step": 10453
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6991183757781982,
      "learning_rate": 1.5674521059132455e-05,
      "loss": 0.4669,
      "step": 10454
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6741999387741089,
      "learning_rate": 1.5673682463993708e-05,
      "loss": 0.5482,
      "step": 10455
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5607070922851562,
      "learning_rate": 1.5672843810009802e-05,
      "loss": 0.4602,
      "step": 10456
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5632612705230713,
      "learning_rate": 1.567200509718943e-05,
      "loss": 0.4861,
      "step": 10457
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6378310918807983,
      "learning_rate": 1.5671166325541298e-05,
      "loss": 0.4637,
      "step": 10458
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.640005350112915,
      "learning_rate": 1.5670327495074104e-05,
      "loss": 0.5205,
      "step": 10459
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7630250453948975,
      "learning_rate": 1.5669488605796544e-05,
      "loss": 0.4826,
      "step": 10460
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6784459352493286,
      "learning_rate": 1.5668649657717318e-05,
      "loss": 0.5578,
      "step": 10461
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6157690286636353,
      "learning_rate": 1.5667810650845138e-05,
      "loss": 0.4379,
      "step": 10462
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8306541442871094,
      "learning_rate": 1.566697158518869e-05,
      "loss": 0.9583,
      "step": 10463
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6299502849578857,
      "learning_rate": 1.566613246075669e-05,
      "loss": 1.1407,
      "step": 10464
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7495920658111572,
      "learning_rate": 1.5665293277557835e-05,
      "loss": 0.5236,
      "step": 10465
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7633849382400513,
      "learning_rate": 1.566445403560083e-05,
      "loss": 0.4472,
      "step": 10466
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.663953185081482,
      "learning_rate": 1.5663614734894376e-05,
      "loss": 0.5516,
      "step": 10467
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.577579379081726,
      "learning_rate": 1.566277537544718e-05,
      "loss": 0.4679,
      "step": 10468
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.627803921699524,
      "learning_rate": 1.5661935957267952e-05,
      "loss": 0.5379,
      "step": 10469
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6975700855255127,
      "learning_rate": 1.5661096480365395e-05,
      "loss": 0.5046,
      "step": 10470
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7108606100082397,
      "learning_rate": 1.5660256944748212e-05,
      "loss": 0.5215,
      "step": 10471
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5835899114608765,
      "learning_rate": 1.5659417350425114e-05,
      "loss": 0.4694,
      "step": 10472
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6790363788604736,
      "learning_rate": 1.5658577697404807e-05,
      "loss": 0.5305,
      "step": 10473
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.723239541053772,
      "learning_rate": 1.5657737985696004e-05,
      "loss": 0.4674,
      "step": 10474
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.745947241783142,
      "learning_rate": 1.5656898215307407e-05,
      "loss": 0.961,
      "step": 10475
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.516388177871704,
      "learning_rate": 1.565605838624773e-05,
      "loss": 1.015,
      "step": 10476
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.845365047454834,
      "learning_rate": 1.5655218498525684e-05,
      "loss": 0.6082,
      "step": 10477
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7345895767211914,
      "learning_rate": 1.5654378552149984e-05,
      "loss": 0.4262,
      "step": 10478
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6424659490585327,
      "learning_rate": 1.565353854712933e-05,
      "loss": 0.4806,
      "step": 10479
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6390490531921387,
      "learning_rate": 1.5652698483472445e-05,
      "loss": 0.5352,
      "step": 10480
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.598546028137207,
      "learning_rate": 1.5651858361188032e-05,
      "loss": 0.4997,
      "step": 10481
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6470367908477783,
      "learning_rate": 1.5651018180284812e-05,
      "loss": 0.5047,
      "step": 10482
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7499878406524658,
      "learning_rate": 1.56501779407715e-05,
      "loss": 0.5477,
      "step": 10483
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5844415426254272,
      "learning_rate": 1.5649337642656806e-05,
      "loss": 0.4624,
      "step": 10484
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6492030620574951,
      "learning_rate": 1.5648497285949446e-05,
      "loss": 0.5448,
      "step": 10485
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6561728715896606,
      "learning_rate": 1.5647656870658142e-05,
      "loss": 0.4445,
      "step": 10486
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6349903345108032,
      "learning_rate": 1.5646816396791596e-05,
      "loss": 0.5354,
      "step": 10487
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6792970895767212,
      "learning_rate": 1.5645975864358543e-05,
      "loss": 0.4863,
      "step": 10488
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7305474281311035,
      "learning_rate": 1.5645135273367686e-05,
      "loss": 0.5552,
      "step": 10489
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6611415147781372,
      "learning_rate": 1.5644294623827754e-05,
      "loss": 0.4977,
      "step": 10490
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.688520908355713,
      "learning_rate": 1.5643453915747457e-05,
      "loss": 0.509,
      "step": 10491
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6481983661651611,
      "learning_rate": 1.564261314913552e-05,
      "loss": 0.4536,
      "step": 10492
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.615971326828003,
      "learning_rate": 1.564177232400066e-05,
      "loss": 0.4811,
      "step": 10493
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.578386664390564,
      "learning_rate": 1.5640931440351602e-05,
      "loss": 0.4484,
      "step": 10494
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6963963508605957,
      "learning_rate": 1.5640090498197064e-05,
      "loss": 0.5161,
      "step": 10495
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.624740481376648,
      "learning_rate": 1.563924949754577e-05,
      "loss": 0.4368,
      "step": 10496
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6234369277954102,
      "learning_rate": 1.5638408438406436e-05,
      "loss": 0.5173,
      "step": 10497
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6180557012557983,
      "learning_rate": 1.5637567320787797e-05,
      "loss": 0.4527,
      "step": 10498
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6472426652908325,
      "learning_rate": 1.5636726144698564e-05,
      "loss": 0.5039,
      "step": 10499
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5139845609664917,
      "learning_rate": 1.563588491014747e-05,
      "loss": 0.4405,
      "step": 10500
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.671848177909851,
      "learning_rate": 1.5635043617143234e-05,
      "loss": 0.5152,
      "step": 10501
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5731232166290283,
      "learning_rate": 1.563420226569459e-05,
      "loss": 0.4418,
      "step": 10502
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9246920347213745,
      "learning_rate": 1.5633360855810257e-05,
      "loss": 1.0222,
      "step": 10503
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5661370754241943,
      "learning_rate": 1.5632519387498964e-05,
      "loss": 1.0579,
      "step": 10504
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6901791095733643,
      "learning_rate": 1.5631677860769435e-05,
      "loss": 0.4995,
      "step": 10505
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7381715774536133,
      "learning_rate": 1.5630836275630404e-05,
      "loss": 0.513,
      "step": 10506
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5551862716674805,
      "learning_rate": 1.5629994632090596e-05,
      "loss": 0.4836,
      "step": 10507
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.674235463142395,
      "learning_rate": 1.5629152930158743e-05,
      "loss": 0.4937,
      "step": 10508
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5840014219284058,
      "learning_rate": 1.562831116984357e-05,
      "loss": 0.5357,
      "step": 10509
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6308927536010742,
      "learning_rate": 1.5627469351153813e-05,
      "loss": 0.4881,
      "step": 10510
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5149165391921997,
      "learning_rate": 1.5626627474098196e-05,
      "loss": 0.4831,
      "step": 10511
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5589431524276733,
      "learning_rate": 1.5625785538685457e-05,
      "loss": 0.4563,
      "step": 10512
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5375779867172241,
      "learning_rate": 1.5624943544924326e-05,
      "loss": 0.5057,
      "step": 10513
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.611841082572937,
      "learning_rate": 1.5624101492823536e-05,
      "loss": 0.4619,
      "step": 10514
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7097055912017822,
      "learning_rate": 1.562325938239182e-05,
      "loss": 0.5211,
      "step": 10515
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7414289712905884,
      "learning_rate": 1.562241721363791e-05,
      "loss": 0.4629,
      "step": 10516
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5803070068359375,
      "learning_rate": 1.5621574986570545e-05,
      "loss": 0.4899,
      "step": 10517
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6394315958023071,
      "learning_rate": 1.5620732701198457e-05,
      "loss": 0.4791,
      "step": 10518
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5594170093536377,
      "learning_rate": 1.5619890357530383e-05,
      "loss": 0.4631,
      "step": 10519
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6108533143997192,
      "learning_rate": 1.561904795557506e-05,
      "loss": 0.4497,
      "step": 10520
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.743693232536316,
      "learning_rate": 1.5618205495341225e-05,
      "loss": 0.5608,
      "step": 10521
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5733106136322021,
      "learning_rate": 1.5617362976837612e-05,
      "loss": 0.4464,
      "step": 10522
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7609506845474243,
      "learning_rate": 1.5616520400072963e-05,
      "loss": 0.5452,
      "step": 10523
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7668815851211548,
      "learning_rate": 1.561567776505602e-05,
      "loss": 0.4411,
      "step": 10524
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6188442707061768,
      "learning_rate": 1.5614835071795513e-05,
      "loss": 0.5167,
      "step": 10525
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.639631748199463,
      "learning_rate": 1.5613992320300194e-05,
      "loss": 0.4841,
      "step": 10526
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.652001142501831,
      "learning_rate": 1.561314951057879e-05,
      "loss": 0.5911,
      "step": 10527
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.53661048412323,
      "learning_rate": 1.5612306642640058e-05,
      "loss": 0.437,
      "step": 10528
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6904641389846802,
      "learning_rate": 1.561146371649273e-05,
      "loss": 0.5218,
      "step": 10529
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5832093954086304,
      "learning_rate": 1.5610620732145543e-05,
      "loss": 0.4587,
      "step": 10530
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6283618211746216,
      "learning_rate": 1.5609777689607254e-05,
      "loss": 0.4729,
      "step": 10531
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6296955347061157,
      "learning_rate": 1.5608934588886597e-05,
      "loss": 0.4481,
      "step": 10532
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.732678771018982,
      "learning_rate": 1.560809142999232e-05,
      "loss": 0.6254,
      "step": 10533
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5836158990859985,
      "learning_rate": 1.5607248212933166e-05,
      "loss": 0.4608,
      "step": 10534
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5755327939987183,
      "learning_rate": 1.560640493771788e-05,
      "loss": 0.4769,
      "step": 10535
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6951853036880493,
      "learning_rate": 1.5605561604355214e-05,
      "loss": 0.4746,
      "step": 10536
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6101220846176147,
      "learning_rate": 1.560471821285391e-05,
      "loss": 0.5101,
      "step": 10537
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5633491277694702,
      "learning_rate": 1.5603874763222714e-05,
      "loss": 0.4962,
      "step": 10538
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6939697265625,
      "learning_rate": 1.5603031255470376e-05,
      "loss": 0.5599,
      "step": 10539
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5660794973373413,
      "learning_rate": 1.5602187689605642e-05,
      "loss": 0.4397,
      "step": 10540
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.624728798866272,
      "learning_rate": 1.5601344065637264e-05,
      "loss": 0.5454,
      "step": 10541
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6175156831741333,
      "learning_rate": 1.5600500383573995e-05,
      "loss": 0.4773,
      "step": 10542
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6966677904129028,
      "learning_rate": 1.559965664342458e-05,
      "loss": 0.5288,
      "step": 10543
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5479185581207275,
      "learning_rate": 1.559881284519777e-05,
      "loss": 0.435,
      "step": 10544
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6515233516693115,
      "learning_rate": 1.5597968988902317e-05,
      "loss": 0.5193,
      "step": 10545
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5612183809280396,
      "learning_rate": 1.559712507454698e-05,
      "loss": 0.4735,
      "step": 10546
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6790603399276733,
      "learning_rate": 1.5596281102140494e-05,
      "loss": 0.507,
      "step": 10547
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.566686987876892,
      "learning_rate": 1.5595437071691633e-05,
      "loss": 0.4612,
      "step": 10548
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.563504934310913,
      "learning_rate": 1.559459298320914e-05,
      "loss": 0.4898,
      "step": 10549
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8425079584121704,
      "learning_rate": 1.5593748836701772e-05,
      "loss": 0.4778,
      "step": 10550
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7190113067626953,
      "learning_rate": 1.5592904632178282e-05,
      "loss": 0.5734,
      "step": 10551
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5318125486373901,
      "learning_rate": 1.5592060369647428e-05,
      "loss": 0.4615,
      "step": 10552
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5941945314407349,
      "learning_rate": 1.5591216049117965e-05,
      "loss": 0.4839,
      "step": 10553
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8024532794952393,
      "learning_rate": 1.559037167059865e-05,
      "loss": 0.4771,
      "step": 10554
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.636648416519165,
      "learning_rate": 1.5589527234098247e-05,
      "loss": 0.5384,
      "step": 10555
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5786634683609009,
      "learning_rate": 1.5588682739625506e-05,
      "loss": 0.465,
      "step": 10556
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6003342866897583,
      "learning_rate": 1.5587838187189186e-05,
      "loss": 0.5074,
      "step": 10557
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7105355262756348,
      "learning_rate": 1.558699357679805e-05,
      "loss": 0.4644,
      "step": 10558
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.876296043395996,
      "learning_rate": 1.5586148908460855e-05,
      "loss": 0.4748,
      "step": 10559
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5739672183990479,
      "learning_rate": 1.5585304182186364e-05,
      "loss": 0.4928,
      "step": 10560
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.602768063545227,
      "learning_rate": 1.5584459397983337e-05,
      "loss": 0.4811,
      "step": 10561
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5803031921386719,
      "learning_rate": 1.558361455586054e-05,
      "loss": 0.4557,
      "step": 10562
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5795611143112183,
      "learning_rate": 1.5582769655826723e-05,
      "loss": 0.5667,
      "step": 10563
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6324470043182373,
      "learning_rate": 1.5581924697890665e-05,
      "loss": 0.4862,
      "step": 10564
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6600724458694458,
      "learning_rate": 1.558107968206112e-05,
      "loss": 0.5254,
      "step": 10565
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6532034873962402,
      "learning_rate": 1.5580234608346852e-05,
      "loss": 0.4607,
      "step": 10566
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.642255425453186,
      "learning_rate": 1.557938947675663e-05,
      "loss": 0.4623,
      "step": 10567
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6724753379821777,
      "learning_rate": 1.5578544287299217e-05,
      "loss": 0.465,
      "step": 10568
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5752333402633667,
      "learning_rate": 1.557769903998338e-05,
      "loss": 0.5197,
      "step": 10569
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5611635446548462,
      "learning_rate": 1.557685373481788e-05,
      "loss": 0.4792,
      "step": 10570
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6434447765350342,
      "learning_rate": 1.5576008371811492e-05,
      "loss": 0.5129,
      "step": 10571
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5908461809158325,
      "learning_rate": 1.5575162950972983e-05,
      "loss": 0.4632,
      "step": 10572
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5943219661712646,
      "learning_rate": 1.5574317472311115e-05,
      "loss": 0.488,
      "step": 10573
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5167878866195679,
      "learning_rate": 1.5573471935834665e-05,
      "loss": 0.4399,
      "step": 10574
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5833227634429932,
      "learning_rate": 1.5572626341552397e-05,
      "loss": 0.4739,
      "step": 10575
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6094963550567627,
      "learning_rate": 1.5571780689473083e-05,
      "loss": 0.436,
      "step": 10576
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.57863187789917,
      "learning_rate": 1.5570934979605496e-05,
      "loss": 0.5142,
      "step": 10577
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5517005920410156,
      "learning_rate": 1.55700892119584e-05,
      "loss": 0.4588,
      "step": 10578
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.624099612236023,
      "learning_rate": 1.5569243386540573e-05,
      "loss": 0.5508,
      "step": 10579
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.583301067352295,
      "learning_rate": 1.5568397503360788e-05,
      "loss": 0.4661,
      "step": 10580
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.677135705947876,
      "learning_rate": 1.5567551562427813e-05,
      "loss": 0.5859,
      "step": 10581
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.69483482837677,
      "learning_rate": 1.556670556375043e-05,
      "loss": 0.5308,
      "step": 10582
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6349332332611084,
      "learning_rate": 1.5565859507337408e-05,
      "loss": 0.484,
      "step": 10583
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5818891525268555,
      "learning_rate": 1.556501339319752e-05,
      "loss": 0.4975,
      "step": 10584
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.814003586769104,
      "learning_rate": 1.5564167221339547e-05,
      "loss": 0.5569,
      "step": 10585
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5848264694213867,
      "learning_rate": 1.556332099177226e-05,
      "loss": 0.4583,
      "step": 10586
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5764336585998535,
      "learning_rate": 1.556247470450444e-05,
      "loss": 0.5119,
      "step": 10587
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7654122114181519,
      "learning_rate": 1.556162835954486e-05,
      "loss": 0.5513,
      "step": 10588
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5742794275283813,
      "learning_rate": 1.5560781956902302e-05,
      "loss": 0.5217,
      "step": 10589
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.632225751876831,
      "learning_rate": 1.5559935496585543e-05,
      "loss": 0.4785,
      "step": 10590
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6132903099060059,
      "learning_rate": 1.5559088978603363e-05,
      "loss": 0.5021,
      "step": 10591
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.601798176765442,
      "learning_rate": 1.5558242402964538e-05,
      "loss": 0.465,
      "step": 10592
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5611985921859741,
      "learning_rate": 1.5557395769677854e-05,
      "loss": 0.5169,
      "step": 10593
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5989952087402344,
      "learning_rate": 1.555654907875209e-05,
      "loss": 0.472,
      "step": 10594
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6006091833114624,
      "learning_rate": 1.5555702330196024e-05,
      "loss": 0.4952,
      "step": 10595
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6887401342391968,
      "learning_rate": 1.555485552401844e-05,
      "loss": 0.48,
      "step": 10596
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5951261520385742,
      "learning_rate": 1.5554008660228127e-05,
      "loss": 0.4975,
      "step": 10597
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5585196018218994,
      "learning_rate": 1.555316173883386e-05,
      "loss": 0.4445,
      "step": 10598
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6591295003890991,
      "learning_rate": 1.5552314759844422e-05,
      "loss": 0.5648,
      "step": 10599
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5533117055892944,
      "learning_rate": 1.5551467723268604e-05,
      "loss": 0.484,
      "step": 10600
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6745210886001587,
      "learning_rate": 1.555062062911519e-05,
      "loss": 0.521,
      "step": 10601
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6512256860733032,
      "learning_rate": 1.554977347739296e-05,
      "loss": 0.4939,
      "step": 10602
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.921298623085022,
      "learning_rate": 1.554892626811071e-05,
      "loss": 0.5534,
      "step": 10603
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5395745038986206,
      "learning_rate": 1.5548079001277218e-05,
      "loss": 0.4468,
      "step": 10604
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6707580089569092,
      "learning_rate": 1.5547231676901278e-05,
      "loss": 0.5126,
      "step": 10605
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.58840012550354,
      "learning_rate": 1.5546384294991672e-05,
      "loss": 0.4489,
      "step": 10606
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6980751752853394,
      "learning_rate": 1.5545536855557193e-05,
      "loss": 0.5125,
      "step": 10607
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6739810705184937,
      "learning_rate": 1.5544689358606628e-05,
      "loss": 0.4856,
      "step": 10608
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6746348142623901,
      "learning_rate": 1.554384180414877e-05,
      "loss": 0.4884,
      "step": 10609
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5611778497695923,
      "learning_rate": 1.5542994192192403e-05,
      "loss": 0.4366,
      "step": 10610
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.755589246749878,
      "learning_rate": 1.5542146522746326e-05,
      "loss": 0.5702,
      "step": 10611
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5459345579147339,
      "learning_rate": 1.5541298795819327e-05,
      "loss": 0.4244,
      "step": 10612
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7097384929656982,
      "learning_rate": 1.55404510114202e-05,
      "loss": 0.5357,
      "step": 10613
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6154968738555908,
      "learning_rate": 1.553960316955773e-05,
      "loss": 0.4597,
      "step": 10614
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6796514987945557,
      "learning_rate": 1.5538755270240727e-05,
      "loss": 0.5175,
      "step": 10615
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.583823800086975,
      "learning_rate": 1.5537907313477968e-05,
      "loss": 0.4477,
      "step": 10616
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5974581241607666,
      "learning_rate": 1.5537059299278257e-05,
      "loss": 0.5215,
      "step": 10617
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4887571334838867,
      "learning_rate": 1.5536211227650383e-05,
      "loss": 0.4739,
      "step": 10618
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5952824354171753,
      "learning_rate": 1.5535363098603152e-05,
      "loss": 0.4847,
      "step": 10619
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5969183444976807,
      "learning_rate": 1.553451491214535e-05,
      "loss": 0.476,
      "step": 10620
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5737196207046509,
      "learning_rate": 1.5533666668285777e-05,
      "loss": 0.4995,
      "step": 10621
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6021183729171753,
      "learning_rate": 1.5532818367033234e-05,
      "loss": 0.4417,
      "step": 10622
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6626033782958984,
      "learning_rate": 1.553197000839652e-05,
      "loss": 0.5284,
      "step": 10623
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6312403678894043,
      "learning_rate": 1.5531121592384427e-05,
      "loss": 0.4449,
      "step": 10624
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6597504615783691,
      "learning_rate": 1.553027311900576e-05,
      "loss": 0.5297,
      "step": 10625
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.6666877269744873,
      "learning_rate": 1.5529424588269315e-05,
      "loss": 0.4464,
      "step": 10626
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7054427862167358,
      "learning_rate": 1.5528576000183898e-05,
      "loss": 0.5704,
      "step": 10627
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8373944759368896,
      "learning_rate": 1.5527727354758307e-05,
      "loss": 0.4777,
      "step": 10628
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7053310871124268,
      "learning_rate": 1.552687865200134e-05,
      "loss": 0.571,
      "step": 10629
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5017268657684326,
      "learning_rate": 1.5526029891921807e-05,
      "loss": 0.4459,
      "step": 10630
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6286165714263916,
      "learning_rate": 1.5525181074528508e-05,
      "loss": 0.5391,
      "step": 10631
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5702118873596191,
      "learning_rate": 1.5524332199830245e-05,
      "loss": 0.4374,
      "step": 10632
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.76698899269104,
      "learning_rate": 1.5523483267835824e-05,
      "loss": 0.5562,
      "step": 10633
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5791956186294556,
      "learning_rate": 1.5522634278554048e-05,
      "loss": 0.4566,
      "step": 10634
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5637534856796265,
      "learning_rate": 1.5521785231993725e-05,
      "loss": 0.4913,
      "step": 10635
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5527347326278687,
      "learning_rate": 1.5520936128163655e-05,
      "loss": 0.4552,
      "step": 10636
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6827213764190674,
      "learning_rate": 1.5520086967072653e-05,
      "loss": 0.5463,
      "step": 10637
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.565916657447815,
      "learning_rate": 1.5519237748729522e-05,
      "loss": 0.4772,
      "step": 10638
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.56169593334198,
      "learning_rate": 1.5518388473143068e-05,
      "loss": 0.4956,
      "step": 10639
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6416667699813843,
      "learning_rate": 1.5517539140322102e-05,
      "loss": 0.4635,
      "step": 10640
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6136075258255005,
      "learning_rate": 1.5516689750275432e-05,
      "loss": 0.4844,
      "step": 10641
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6861751079559326,
      "learning_rate": 1.551584030301187e-05,
      "loss": 0.5059,
      "step": 10642
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.982611060142517,
      "learning_rate": 1.551499079854022e-05,
      "loss": 0.5199,
      "step": 10643
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5594115257263184,
      "learning_rate": 1.5514141236869298e-05,
      "loss": 0.4502,
      "step": 10644
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6677287817001343,
      "learning_rate": 1.5513291618007915e-05,
      "loss": 0.5014,
      "step": 10645
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6157479286193848,
      "learning_rate": 1.551244194196488e-05,
      "loss": 0.5485,
      "step": 10646
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.764349102973938,
      "learning_rate": 1.551159220874901e-05,
      "loss": 0.489,
      "step": 10647
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5872184038162231,
      "learning_rate": 1.551074241836911e-05,
      "loss": 0.4336,
      "step": 10648
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6653565168380737,
      "learning_rate": 1.5509892570834006e-05,
      "loss": 0.501,
      "step": 10649
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6099753379821777,
      "learning_rate": 1.55090426661525e-05,
      "loss": 0.4368,
      "step": 10650
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8217434883117676,
      "learning_rate": 1.5508192704333415e-05,
      "loss": 0.5369,
      "step": 10651
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.620958924293518,
      "learning_rate": 1.5507342685385557e-05,
      "loss": 0.4927,
      "step": 10652
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6487575769424438,
      "learning_rate": 1.5506492609317758e-05,
      "loss": 0.5421,
      "step": 10653
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6196297407150269,
      "learning_rate": 1.550564247613882e-05,
      "loss": 0.4522,
      "step": 10654
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6092329025268555,
      "learning_rate": 1.550479228585757e-05,
      "loss": 0.4914,
      "step": 10655
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6249269247055054,
      "learning_rate": 1.5503942038482815e-05,
      "loss": 0.5264,
      "step": 10656
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7093312740325928,
      "learning_rate": 1.5503091734023383e-05,
      "loss": 0.5487,
      "step": 10657
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6484333276748657,
      "learning_rate": 1.550224137248809e-05,
      "loss": 0.4784,
      "step": 10658
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6828176975250244,
      "learning_rate": 1.5501390953885757e-05,
      "loss": 0.5181,
      "step": 10659
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5516260862350464,
      "learning_rate": 1.5500540478225198e-05,
      "loss": 0.4509,
      "step": 10660
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6740238666534424,
      "learning_rate": 1.5499689945515242e-05,
      "loss": 0.5224,
      "step": 10661
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5916517972946167,
      "learning_rate": 1.5498839355764705e-05,
      "loss": 0.4578,
      "step": 10662
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5304491519927979,
      "learning_rate": 1.5497988708982413e-05,
      "loss": 0.4938,
      "step": 10663
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6647037267684937,
      "learning_rate": 1.5497138005177186e-05,
      "loss": 0.478,
      "step": 10664
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6038237810134888,
      "learning_rate": 1.5496287244357845e-05,
      "loss": 0.4977,
      "step": 10665
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5175689458847046,
      "learning_rate": 1.5495436426533218e-05,
      "loss": 0.47,
      "step": 10666
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6267247200012207,
      "learning_rate": 1.549458555171213e-05,
      "loss": 0.5401,
      "step": 10667
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6378686428070068,
      "learning_rate": 1.54937346199034e-05,
      "loss": 0.4962,
      "step": 10668
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6402199268341064,
      "learning_rate": 1.549288363111586e-05,
      "loss": 0.5558,
      "step": 10669
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5919504165649414,
      "learning_rate": 1.5492032585358328e-05,
      "loss": 0.4451,
      "step": 10670
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.845550298690796,
      "learning_rate": 1.549118148263964e-05,
      "loss": 0.504,
      "step": 10671
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6169408559799194,
      "learning_rate": 1.549033032296862e-05,
      "loss": 0.4481,
      "step": 10672
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6116005182266235,
      "learning_rate": 1.5489479106354094e-05,
      "loss": 0.5285,
      "step": 10673
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5754612684249878,
      "learning_rate": 1.548862783280489e-05,
      "loss": 0.4522,
      "step": 10674
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6418777704238892,
      "learning_rate": 1.548777650232984e-05,
      "loss": 0.5366,
      "step": 10675
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5891356468200684,
      "learning_rate": 1.548692511493777e-05,
      "loss": 0.425,
      "step": 10676
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7505072355270386,
      "learning_rate": 1.5486073670637516e-05,
      "loss": 0.5772,
      "step": 10677
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5924791097640991,
      "learning_rate": 1.5485222169437905e-05,
      "loss": 0.4672,
      "step": 10678
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.637709379196167,
      "learning_rate": 1.5484370611347766e-05,
      "loss": 0.521,
      "step": 10679
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5907328128814697,
      "learning_rate": 1.5483518996375936e-05,
      "loss": 0.4902,
      "step": 10680
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6704738140106201,
      "learning_rate": 1.5482667324531244e-05,
      "loss": 0.5046,
      "step": 10681
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.656130075454712,
      "learning_rate": 1.5481815595822527e-05,
      "loss": 0.4892,
      "step": 10682
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6413567066192627,
      "learning_rate": 1.5480963810258614e-05,
      "loss": 0.5197,
      "step": 10683
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5707801580429077,
      "learning_rate": 1.548011196784834e-05,
      "loss": 0.4739,
      "step": 10684
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6045417785644531,
      "learning_rate": 1.5479260068600548e-05,
      "loss": 0.5222,
      "step": 10685
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.578145146369934,
      "learning_rate": 1.547840811252406e-05,
      "loss": 0.4584,
      "step": 10686
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5901339054107666,
      "learning_rate": 1.5477556099627726e-05,
      "loss": 0.497,
      "step": 10687
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6689547300338745,
      "learning_rate": 1.5476704029920373e-05,
      "loss": 0.4531,
      "step": 10688
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.656368374824524,
      "learning_rate": 1.5475851903410845e-05,
      "loss": 0.5112,
      "step": 10689
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5852129459381104,
      "learning_rate": 1.547499972010797e-05,
      "loss": 0.4503,
      "step": 10690
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5430210828781128,
      "learning_rate": 1.5474147480020598e-05,
      "loss": 0.4844,
      "step": 10691
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7128626108169556,
      "learning_rate": 1.547329518315756e-05,
      "loss": 0.49,
      "step": 10692
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6096038818359375,
      "learning_rate": 1.54724428295277e-05,
      "loss": 0.4808,
      "step": 10693
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6647289991378784,
      "learning_rate": 1.5471590419139856e-05,
      "loss": 0.4966,
      "step": 10694
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7294303178787231,
      "learning_rate": 1.5470737952002873e-05,
      "loss": 0.5393,
      "step": 10695
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6758735179901123,
      "learning_rate": 1.5469885428125584e-05,
      "loss": 0.5187,
      "step": 10696
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.693617582321167,
      "learning_rate": 1.546903284751684e-05,
      "loss": 0.5496,
      "step": 10697
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6656454801559448,
      "learning_rate": 1.546818021018548e-05,
      "loss": 0.5063,
      "step": 10698
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.725093126296997,
      "learning_rate": 1.5467327516140347e-05,
      "loss": 0.5399,
      "step": 10699
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5867431163787842,
      "learning_rate": 1.5466474765390285e-05,
      "loss": 0.4444,
      "step": 10700
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.619779348373413,
      "learning_rate": 1.546562195794414e-05,
      "loss": 0.4803,
      "step": 10701
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5624135732650757,
      "learning_rate": 1.5464769093810752e-05,
      "loss": 0.4484,
      "step": 10702
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5652540922164917,
      "learning_rate": 1.5463916172998968e-05,
      "loss": 0.4901,
      "step": 10703
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5949461460113525,
      "learning_rate": 1.5463063195517644e-05,
      "loss": 0.4496,
      "step": 10704
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5864251852035522,
      "learning_rate": 1.5462210161375612e-05,
      "loss": 0.49,
      "step": 10705
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4630619287490845,
      "learning_rate": 1.5461357070581727e-05,
      "loss": 0.4275,
      "step": 10706
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.739735722541809,
      "learning_rate": 1.5460503923144838e-05,
      "loss": 0.5596,
      "step": 10707
    },
    {
      "epoch": 0.34,
      "grad_norm": 11.363396644592285,
      "learning_rate": 1.545965071907379e-05,
      "loss": 0.4632,
      "step": 10708
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6130354404449463,
      "learning_rate": 1.5458797458377436e-05,
      "loss": 0.497,
      "step": 10709
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6145063638687134,
      "learning_rate": 1.545794414106462e-05,
      "loss": 0.4471,
      "step": 10710
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5887197256088257,
      "learning_rate": 1.54570907671442e-05,
      "loss": 0.5026,
      "step": 10711
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6169402599334717,
      "learning_rate": 1.5456237336625016e-05,
      "loss": 0.4413,
      "step": 10712
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5946199893951416,
      "learning_rate": 1.545538384951593e-05,
      "loss": 0.535,
      "step": 10713
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8635996580123901,
      "learning_rate": 1.5454530305825787e-05,
      "loss": 0.4889,
      "step": 10714
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6207553148269653,
      "learning_rate": 1.5453676705563444e-05,
      "loss": 0.5167,
      "step": 10715
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6315418481826782,
      "learning_rate": 1.5452823048737756e-05,
      "loss": 0.4733,
      "step": 10716
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6357282400131226,
      "learning_rate": 1.545196933535757e-05,
      "loss": 0.5406,
      "step": 10717
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6645219326019287,
      "learning_rate": 1.5451115565431744e-05,
      "loss": 0.5133,
      "step": 10718
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6361472606658936,
      "learning_rate": 1.5450261738969134e-05,
      "loss": 0.4982,
      "step": 10719
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4939004182815552,
      "learning_rate": 1.544940785597859e-05,
      "loss": 0.4407,
      "step": 10720
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6942945718765259,
      "learning_rate": 1.5448553916468983e-05,
      "loss": 0.5367,
      "step": 10721
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.66720712184906,
      "learning_rate": 1.5447699920449147e-05,
      "loss": 0.4447,
      "step": 10722
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.057383060455322,
      "learning_rate": 1.5446845867927958e-05,
      "loss": 0.6443,
      "step": 10723
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.644471526145935,
      "learning_rate": 1.5445991758914266e-05,
      "loss": 0.4643,
      "step": 10724
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.590032935142517,
      "learning_rate": 1.5445137593416934e-05,
      "loss": 0.5269,
      "step": 10725
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6826752424240112,
      "learning_rate": 1.5444283371444816e-05,
      "loss": 0.5302,
      "step": 10726
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6216381788253784,
      "learning_rate": 1.5443429093006774e-05,
      "loss": 0.5421,
      "step": 10727
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6415531635284424,
      "learning_rate": 1.5442574758111665e-05,
      "loss": 0.4689,
      "step": 10728
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.632236361503601,
      "learning_rate": 1.5441720366768356e-05,
      "loss": 0.4988,
      "step": 10729
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5479718446731567,
      "learning_rate": 1.54408659189857e-05,
      "loss": 0.4542,
      "step": 10730
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6285028457641602,
      "learning_rate": 1.5440011414772567e-05,
      "loss": 0.5181,
      "step": 10731
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6451058387756348,
      "learning_rate": 1.543915685413782e-05,
      "loss": 0.4602,
      "step": 10732
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6008198261260986,
      "learning_rate": 1.5438302237090315e-05,
      "loss": 0.4719,
      "step": 10733
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5561639070510864,
      "learning_rate": 1.543744756363892e-05,
      "loss": 0.4512,
      "step": 10734
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.671964168548584,
      "learning_rate": 1.54365928337925e-05,
      "loss": 0.5098,
      "step": 10735
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5623486042022705,
      "learning_rate": 1.5435738047559917e-05,
      "loss": 0.4676,
      "step": 10736
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7393909692764282,
      "learning_rate": 1.543488320495004e-05,
      "loss": 0.521,
      "step": 10737
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5594515800476074,
      "learning_rate": 1.543402830597173e-05,
      "loss": 0.4709,
      "step": 10738
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6152418851852417,
      "learning_rate": 1.5433173350633862e-05,
      "loss": 0.5923,
      "step": 10739
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5621262788772583,
      "learning_rate": 1.5432318338945296e-05,
      "loss": 0.4718,
      "step": 10740
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5890406370162964,
      "learning_rate": 1.54314632709149e-05,
      "loss": 0.5154,
      "step": 10741
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5251535177230835,
      "learning_rate": 1.5430608146551546e-05,
      "loss": 0.4328,
      "step": 10742
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5815670490264893,
      "learning_rate": 1.5429752965864105e-05,
      "loss": 0.4727,
      "step": 10743
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.651175618171692,
      "learning_rate": 1.542889772886144e-05,
      "loss": 0.4962,
      "step": 10744
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6070560216903687,
      "learning_rate": 1.5428042435552422e-05,
      "loss": 0.5358,
      "step": 10745
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5232594013214111,
      "learning_rate": 1.5427187085945927e-05,
      "loss": 0.4366,
      "step": 10746
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6297980546951294,
      "learning_rate": 1.5426331680050823e-05,
      "loss": 0.5636,
      "step": 10747
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.60358464717865,
      "learning_rate": 1.5425476217875984e-05,
      "loss": 0.4769,
      "step": 10748
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.751099586486816,
      "learning_rate": 1.542462069943028e-05,
      "loss": 0.5662,
      "step": 10749
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.740472435951233,
      "learning_rate": 1.542376512472258e-05,
      "loss": 0.4954,
      "step": 10750
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7310723066329956,
      "learning_rate": 1.542290949376177e-05,
      "loss": 0.5134,
      "step": 10751
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5463589429855347,
      "learning_rate": 1.5422053806556713e-05,
      "loss": 0.4603,
      "step": 10752
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6052391529083252,
      "learning_rate": 1.5421198063116286e-05,
      "loss": 0.5139,
      "step": 10753
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.547370433807373,
      "learning_rate": 1.5420342263449368e-05,
      "loss": 0.4417,
      "step": 10754
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.560306191444397,
      "learning_rate": 1.5419486407564833e-05,
      "loss": 0.4732,
      "step": 10755
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6165497303009033,
      "learning_rate": 1.541863049547156e-05,
      "loss": 0.4654,
      "step": 10756
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.765647292137146,
      "learning_rate": 1.541777452717842e-05,
      "loss": 0.5113,
      "step": 10757
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.529310703277588,
      "learning_rate": 1.54169185026943e-05,
      "loss": 0.4282,
      "step": 10758
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6499114036560059,
      "learning_rate": 1.541606242202807e-05,
      "loss": 0.5534,
      "step": 10759
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6475180387496948,
      "learning_rate": 1.5415206285188608e-05,
      "loss": 0.4839,
      "step": 10760
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.649487018585205,
      "learning_rate": 1.5414350092184804e-05,
      "loss": 0.5469,
      "step": 10761
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5889312028884888,
      "learning_rate": 1.5413493843025528e-05,
      "loss": 0.4594,
      "step": 10762
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7383995056152344,
      "learning_rate": 1.5412637537719666e-05,
      "loss": 0.5737,
      "step": 10763
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6083528995513916,
      "learning_rate": 1.5411781176276096e-05,
      "loss": 0.4204,
      "step": 10764
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5802488327026367,
      "learning_rate": 1.5410924758703702e-05,
      "loss": 0.4773,
      "step": 10765
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.635737419128418,
      "learning_rate": 1.5410068285011366e-05,
      "loss": 0.4624,
      "step": 10766
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6594605445861816,
      "learning_rate": 1.540921175520797e-05,
      "loss": 0.5315,
      "step": 10767
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.597885012626648,
      "learning_rate": 1.54083551693024e-05,
      "loss": 0.4438,
      "step": 10768
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6344667673110962,
      "learning_rate": 1.5407498527303538e-05,
      "loss": 0.5086,
      "step": 10769
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.720167636871338,
      "learning_rate": 1.5406641829220266e-05,
      "loss": 0.487,
      "step": 10770
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.665952205657959,
      "learning_rate": 1.5405785075061478e-05,
      "loss": 0.5296,
      "step": 10771
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6056632995605469,
      "learning_rate": 1.540492826483605e-05,
      "loss": 0.4408,
      "step": 10772
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.827365756034851,
      "learning_rate": 1.5404071398552876e-05,
      "loss": 1.0278,
      "step": 10773
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.619212031364441,
      "learning_rate": 1.5403214476220836e-05,
      "loss": 1.0947,
      "step": 10774
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.658145546913147,
      "learning_rate": 1.5402357497848828e-05,
      "loss": 0.4773,
      "step": 10775
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7741812467575073,
      "learning_rate": 1.5401500463445727e-05,
      "loss": 0.4727,
      "step": 10776
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6887974739074707,
      "learning_rate": 1.5400643373020436e-05,
      "loss": 0.497,
      "step": 10777
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6610057353973389,
      "learning_rate": 1.539978622658183e-05,
      "loss": 0.4553,
      "step": 10778
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5756115913391113,
      "learning_rate": 1.5398929024138808e-05,
      "loss": 0.5218,
      "step": 10779
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6079373359680176,
      "learning_rate": 1.539807176570026e-05,
      "loss": 0.4548,
      "step": 10780
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6502071619033813,
      "learning_rate": 1.539721445127508e-05,
      "loss": 0.4708,
      "step": 10781
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.564616322517395,
      "learning_rate": 1.5396357080872147e-05,
      "loss": 0.4238,
      "step": 10782
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6202377080917358,
      "learning_rate": 1.539549965450037e-05,
      "loss": 0.5034,
      "step": 10783
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5688596963882446,
      "learning_rate": 1.5394642172168627e-05,
      "loss": 0.4612,
      "step": 10784
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.050992488861084,
      "learning_rate": 1.5393784633885825e-05,
      "loss": 0.5808,
      "step": 10785
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6798388957977295,
      "learning_rate": 1.5392927039660844e-05,
      "loss": 0.4466,
      "step": 10786
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6889002323150635,
      "learning_rate": 1.539206938950259e-05,
      "loss": 0.5094,
      "step": 10787
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.663475513458252,
      "learning_rate": 1.5391211683419954e-05,
      "loss": 0.4781,
      "step": 10788
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.57681405544281,
      "learning_rate": 1.539035392142183e-05,
      "loss": 0.5366,
      "step": 10789
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5381029844284058,
      "learning_rate": 1.5389496103517118e-05,
      "loss": 0.4441,
      "step": 10790
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6727651357650757,
      "learning_rate": 1.5388638229714713e-05,
      "loss": 0.5538,
      "step": 10791
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6064702272415161,
      "learning_rate": 1.5387780300023512e-05,
      "loss": 0.462,
      "step": 10792
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5509490966796875,
      "learning_rate": 1.5386922314452413e-05,
      "loss": 0.5276,
      "step": 10793
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8632811307907104,
      "learning_rate": 1.5386064273010316e-05,
      "loss": 0.5372,
      "step": 10794
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6258786916732788,
      "learning_rate": 1.5385206175706122e-05,
      "loss": 0.488,
      "step": 10795
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6398504972457886,
      "learning_rate": 1.5384348022548724e-05,
      "loss": 0.4767,
      "step": 10796
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5918145179748535,
      "learning_rate": 1.5383489813547034e-05,
      "loss": 0.4914,
      "step": 10797
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5506467819213867,
      "learning_rate": 1.538263154870994e-05,
      "loss": 0.4355,
      "step": 10798
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.626216173171997,
      "learning_rate": 1.5381773228046354e-05,
      "loss": 0.4819,
      "step": 10799
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.544864296913147,
      "learning_rate": 1.538091485156517e-05,
      "loss": 0.4617,
      "step": 10800
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.693134069442749,
      "learning_rate": 1.5380056419275297e-05,
      "loss": 0.5277,
      "step": 10801
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6225255727767944,
      "learning_rate": 1.537919793118563e-05,
      "loss": 0.4818,
      "step": 10802
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6579055786132812,
      "learning_rate": 1.5378339387305085e-05,
      "loss": 0.5397,
      "step": 10803
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6028143167495728,
      "learning_rate": 1.537748078764256e-05,
      "loss": 0.45,
      "step": 10804
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5533721446990967,
      "learning_rate": 1.5376622132206957e-05,
      "loss": 0.4711,
      "step": 10805
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6722681522369385,
      "learning_rate": 1.5375763421007188e-05,
      "loss": 0.47,
      "step": 10806
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6082302331924438,
      "learning_rate": 1.5374904654052155e-05,
      "loss": 0.5497,
      "step": 10807
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5286617279052734,
      "learning_rate": 1.5374045831350764e-05,
      "loss": 0.4544,
      "step": 10808
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5370240211486816,
      "learning_rate": 1.5373186952911926e-05,
      "loss": 0.5017,
      "step": 10809
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6151143312454224,
      "learning_rate": 1.5372328018744546e-05,
      "loss": 0.4507,
      "step": 10810
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6211713552474976,
      "learning_rate": 1.5371469028857534e-05,
      "loss": 0.5403,
      "step": 10811
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6240689754486084,
      "learning_rate": 1.5370609983259795e-05,
      "loss": 0.4699,
      "step": 10812
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6011841297149658,
      "learning_rate": 1.5369750881960252e-05,
      "loss": 0.4996,
      "step": 10813
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6792864799499512,
      "learning_rate": 1.5368891724967795e-05,
      "loss": 0.4573,
      "step": 10814
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5991376638412476,
      "learning_rate": 1.5368032512291354e-05,
      "loss": 0.5353,
      "step": 10815
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.57034170627594,
      "learning_rate": 1.5367173243939828e-05,
      "loss": 0.4681,
      "step": 10816
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.738729476928711,
      "learning_rate": 1.5366313919922133e-05,
      "loss": 0.5827,
      "step": 10817
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5915790796279907,
      "learning_rate": 1.5365454540247183e-05,
      "loss": 0.4409,
      "step": 10818
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6283984184265137,
      "learning_rate": 1.5364595104923887e-05,
      "loss": 0.4998,
      "step": 10819
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5774046182632446,
      "learning_rate": 1.536373561396116e-05,
      "loss": 0.457,
      "step": 10820
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6984801292419434,
      "learning_rate": 1.5362876067367928e-05,
      "loss": 0.5217,
      "step": 10821
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6391500234603882,
      "learning_rate": 1.5362016465153083e-05,
      "loss": 0.461,
      "step": 10822
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6677253246307373,
      "learning_rate": 1.5361156807325558e-05,
      "loss": 0.5063,
      "step": 10823
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5118818283081055,
      "learning_rate": 1.5360297093894264e-05,
      "loss": 0.4045,
      "step": 10824
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6348118782043457,
      "learning_rate": 1.5359437324868118e-05,
      "loss": 0.5015,
      "step": 10825
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5777170658111572,
      "learning_rate": 1.535857750025604e-05,
      "loss": 0.4791,
      "step": 10826
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7187944650650024,
      "learning_rate": 1.535771762006694e-05,
      "loss": 0.5151,
      "step": 10827
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6010161638259888,
      "learning_rate": 1.535685768430974e-05,
      "loss": 0.4448,
      "step": 10828
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5478774309158325,
      "learning_rate": 1.5355997692993364e-05,
      "loss": 0.4892,
      "step": 10829
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.626845359802246,
      "learning_rate": 1.535513764612672e-05,
      "loss": 0.4564,
      "step": 10830
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.55415678024292,
      "learning_rate": 1.5354277543718743e-05,
      "loss": 0.4876,
      "step": 10831
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6699974536895752,
      "learning_rate": 1.535341738577834e-05,
      "loss": 0.4758,
      "step": 10832
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6520657539367676,
      "learning_rate": 1.5352557172314445e-05,
      "loss": 0.517,
      "step": 10833
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6354384422302246,
      "learning_rate": 1.5351696903335966e-05,
      "loss": 0.493,
      "step": 10834
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6453498601913452,
      "learning_rate": 1.5350836578851834e-05,
      "loss": 0.4988,
      "step": 10835
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5129894018173218,
      "learning_rate": 1.5349976198870974e-05,
      "loss": 0.4193,
      "step": 10836
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6323094367980957,
      "learning_rate": 1.5349115763402304e-05,
      "loss": 0.5269,
      "step": 10837
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5604454278945923,
      "learning_rate": 1.5348255272454748e-05,
      "loss": 0.4362,
      "step": 10838
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6642794609069824,
      "learning_rate": 1.5347394726037233e-05,
      "loss": 0.5349,
      "step": 10839
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5979506969451904,
      "learning_rate": 1.5346534124158685e-05,
      "loss": 0.4576,
      "step": 10840
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.750791072845459,
      "learning_rate": 1.534567346682803e-05,
      "loss": 0.5812,
      "step": 10841
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5909557342529297,
      "learning_rate": 1.534481275405419e-05,
      "loss": 0.4307,
      "step": 10842
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.632704257965088,
      "learning_rate": 1.5343951985846096e-05,
      "loss": 0.5249,
      "step": 10843
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5358339548110962,
      "learning_rate": 1.5343091162212675e-05,
      "loss": 0.4486,
      "step": 10844
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5529778003692627,
      "learning_rate": 1.5342230283162852e-05,
      "loss": 0.5056,
      "step": 10845
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5684103965759277,
      "learning_rate": 1.5341369348705565e-05,
      "loss": 0.4656,
      "step": 10846
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.643106460571289,
      "learning_rate": 1.534050835884973e-05,
      "loss": 0.5028,
      "step": 10847
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6379389762878418,
      "learning_rate": 1.5339647313604286e-05,
      "loss": 0.475,
      "step": 10848
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5963425636291504,
      "learning_rate": 1.5338786212978162e-05,
      "loss": 0.4757,
      "step": 10849
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5719411373138428,
      "learning_rate": 1.5337925056980285e-05,
      "loss": 0.4669,
      "step": 10850
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.553369164466858,
      "learning_rate": 1.533706384561959e-05,
      "loss": 0.4697,
      "step": 10851
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6680206060409546,
      "learning_rate": 1.533620257890501e-05,
      "loss": 0.4965,
      "step": 10852
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5999805927276611,
      "learning_rate": 1.5335341256845473e-05,
      "loss": 0.4884,
      "step": 10853
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6362429857254028,
      "learning_rate": 1.533447987944992e-05,
      "loss": 0.4814,
      "step": 10854
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6613861322402954,
      "learning_rate": 1.5333618446727278e-05,
      "loss": 0.5434,
      "step": 10855
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.581109881401062,
      "learning_rate": 1.5332756958686482e-05,
      "loss": 0.4858,
      "step": 10856
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.699820876121521,
      "learning_rate": 1.533189541533647e-05,
      "loss": 0.5302,
      "step": 10857
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6091126203536987,
      "learning_rate": 1.533103381668618e-05,
      "loss": 0.4391,
      "step": 10858
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5574941635131836,
      "learning_rate": 1.533017216274454e-05,
      "loss": 0.4753,
      "step": 10859
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6610673666000366,
      "learning_rate": 1.5329310453520495e-05,
      "loss": 0.4711,
      "step": 10860
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5727430582046509,
      "learning_rate": 1.5328448689022974e-05,
      "loss": 0.523,
      "step": 10861
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5528630018234253,
      "learning_rate": 1.5327586869260925e-05,
      "loss": 0.45,
      "step": 10862
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.712338924407959,
      "learning_rate": 1.5326724994243275e-05,
      "loss": 0.5322,
      "step": 10863
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6000882387161255,
      "learning_rate": 1.532586306397897e-05,
      "loss": 0.4636,
      "step": 10864
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6802952289581299,
      "learning_rate": 1.5325001078476955e-05,
      "loss": 0.5133,
      "step": 10865
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5439056158065796,
      "learning_rate": 1.5324139037746158e-05,
      "loss": 0.4503,
      "step": 10866
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5992841720581055,
      "learning_rate": 1.5323276941795528e-05,
      "loss": 0.5413,
      "step": 10867
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5731501579284668,
      "learning_rate": 1.5322414790634003e-05,
      "loss": 0.4216,
      "step": 10868
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6336430311203003,
      "learning_rate": 1.5321552584270524e-05,
      "loss": 0.51,
      "step": 10869
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.561352014541626,
      "learning_rate": 1.5320690322714037e-05,
      "loss": 0.454,
      "step": 10870
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6332042217254639,
      "learning_rate": 1.531982800597348e-05,
      "loss": 0.5352,
      "step": 10871
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4872182607650757,
      "learning_rate": 1.5318965634057806e-05,
      "loss": 0.4196,
      "step": 10872
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6172412633895874,
      "learning_rate": 1.5318103206975946e-05,
      "loss": 0.5233,
      "step": 10873
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5762050151824951,
      "learning_rate": 1.5317240724736856e-05,
      "loss": 0.4769,
      "step": 10874
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.624349594116211,
      "learning_rate": 1.5316378187349476e-05,
      "loss": 0.5167,
      "step": 10875
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6215786933898926,
      "learning_rate": 1.531551559482275e-05,
      "loss": 0.453,
      "step": 10876
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6474922895431519,
      "learning_rate": 1.531465294716563e-05,
      "loss": 0.5168,
      "step": 10877
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6175849437713623,
      "learning_rate": 1.5313790244387062e-05,
      "loss": 0.4674,
      "step": 10878
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0530428886413574,
      "learning_rate": 1.531292748649599e-05,
      "loss": 1.0657,
      "step": 10879
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.735713005065918,
      "learning_rate": 1.5312064673501365e-05,
      "loss": 1.1167,
      "step": 10880
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.631449818611145,
      "learning_rate": 1.5311201805412133e-05,
      "loss": 0.5059,
      "step": 10881
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7008060216903687,
      "learning_rate": 1.5310338882237248e-05,
      "loss": 0.4697,
      "step": 10882
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.671558141708374,
      "learning_rate": 1.5309475903985655e-05,
      "loss": 0.5135,
      "step": 10883
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.535696268081665,
      "learning_rate": 1.530861287066631e-05,
      "loss": 0.4331,
      "step": 10884
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7264106273651123,
      "learning_rate": 1.5307749782288154e-05,
      "loss": 0.5838,
      "step": 10885
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.673336386680603,
      "learning_rate": 1.5306886638860153e-05,
      "loss": 0.4689,
      "step": 10886
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6698819398880005,
      "learning_rate": 1.5306023440391245e-05,
      "loss": 0.5309,
      "step": 10887
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5498100519180298,
      "learning_rate": 1.5305160186890394e-05,
      "loss": 0.4524,
      "step": 10888
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7040692567825317,
      "learning_rate": 1.5304296878366543e-05,
      "loss": 0.529,
      "step": 10889
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5885826349258423,
      "learning_rate": 1.530343351482866e-05,
      "loss": 0.464,
      "step": 10890
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6148451566696167,
      "learning_rate": 1.530257009628568e-05,
      "loss": 0.5035,
      "step": 10891
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6115851402282715,
      "learning_rate": 1.5301706622746574e-05,
      "loss": 0.4317,
      "step": 10892
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6274840831756592,
      "learning_rate": 1.5300843094220295e-05,
      "loss": 0.5129,
      "step": 10893
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.532000184059143,
      "learning_rate": 1.5299979510715793e-05,
      "loss": 0.4422,
      "step": 10894
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6856389045715332,
      "learning_rate": 1.5299115872242032e-05,
      "loss": 0.5258,
      "step": 10895
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5950649976730347,
      "learning_rate": 1.5298252178807965e-05,
      "loss": 0.4464,
      "step": 10896
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6523594856262207,
      "learning_rate": 1.5297388430422547e-05,
      "loss": 0.5384,
      "step": 10897
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6521307229995728,
      "learning_rate": 1.5296524627094745e-05,
      "loss": 0.456,
      "step": 10898
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.659065842628479,
      "learning_rate": 1.5295660768833507e-05,
      "loss": 0.5135,
      "step": 10899
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6962841749191284,
      "learning_rate": 1.52947968556478e-05,
      "loss": 0.49,
      "step": 10900
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9674643278121948,
      "learning_rate": 1.5293932887546588e-05,
      "loss": 0.5458,
      "step": 10901
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6880075931549072,
      "learning_rate": 1.529306886453882e-05,
      "loss": 0.4702,
      "step": 10902
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6145716905593872,
      "learning_rate": 1.5292204786633467e-05,
      "loss": 0.5139,
      "step": 10903
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5506881475448608,
      "learning_rate": 1.529134065383949e-05,
      "loss": 0.4573,
      "step": 10904
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5985058546066284,
      "learning_rate": 1.5290476466165846e-05,
      "loss": 0.5126,
      "step": 10905
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5478276014328003,
      "learning_rate": 1.5289612223621505e-05,
      "loss": 0.4617,
      "step": 10906
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.673166036605835,
      "learning_rate": 1.528874792621542e-05,
      "loss": 0.5507,
      "step": 10907
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.682849407196045,
      "learning_rate": 1.528788357395657e-05,
      "loss": 0.4676,
      "step": 10908
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6357141733169556,
      "learning_rate": 1.5287019166853903e-05,
      "loss": 0.4752,
      "step": 10909
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6012605428695679,
      "learning_rate": 1.5286154704916398e-05,
      "loss": 0.4472,
      "step": 10910
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7663534879684448,
      "learning_rate": 1.528529018815301e-05,
      "loss": 0.6043,
      "step": 10911
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.544959306716919,
      "learning_rate": 1.528442561657272e-05,
      "loss": 0.471,
      "step": 10912
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7514604330062866,
      "learning_rate": 1.5283560990184484e-05,
      "loss": 0.5109,
      "step": 10913
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5917503833770752,
      "learning_rate": 1.5282696308997268e-05,
      "loss": 0.4405,
      "step": 10914
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0299088954925537,
      "learning_rate": 1.5281831573020043e-05,
      "loss": 1.0301,
      "step": 10915
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7135275602340698,
      "learning_rate": 1.5280966782261783e-05,
      "loss": 0.9818,
      "step": 10916
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6432796716690063,
      "learning_rate": 1.528010193673145e-05,
      "loss": 0.4614,
      "step": 10917
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.586705207824707,
      "learning_rate": 1.5279237036438018e-05,
      "loss": 0.4367,
      "step": 10918
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5993813276290894,
      "learning_rate": 1.5278372081390457e-05,
      "loss": 0.5118,
      "step": 10919
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.573447346687317,
      "learning_rate": 1.5277507071597733e-05,
      "loss": 0.4836,
      "step": 10920
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6028553247451782,
      "learning_rate": 1.5276642007068826e-05,
      "loss": 0.4919,
      "step": 10921
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5807262659072876,
      "learning_rate": 1.52757768878127e-05,
      "loss": 0.4581,
      "step": 10922
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.596055030822754,
      "learning_rate": 1.5274911713838333e-05,
      "loss": 0.4948,
      "step": 10923
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6318868398666382,
      "learning_rate": 1.5274046485154698e-05,
      "loss": 0.4568,
      "step": 10924
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.694517731666565,
      "learning_rate": 1.5273181201770762e-05,
      "loss": 0.5936,
      "step": 10925
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5503153800964355,
      "learning_rate": 1.527231586369551e-05,
      "loss": 0.4447,
      "step": 10926
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6994596719741821,
      "learning_rate": 1.527145047093791e-05,
      "loss": 0.5256,
      "step": 10927
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.566595435142517,
      "learning_rate": 1.5270585023506938e-05,
      "loss": 0.4364,
      "step": 10928
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.695024013519287,
      "learning_rate": 1.5269719521411574e-05,
      "loss": 0.5512,
      "step": 10929
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5480067729949951,
      "learning_rate": 1.526885396466079e-05,
      "loss": 0.4558,
      "step": 10930
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6941602230072021,
      "learning_rate": 1.5267988353263565e-05,
      "loss": 0.5084,
      "step": 10931
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6795365810394287,
      "learning_rate": 1.5267122687228877e-05,
      "loss": 0.4439,
      "step": 10932
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6124505996704102,
      "learning_rate": 1.52662569665657e-05,
      "loss": 0.4927,
      "step": 10933
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6607658863067627,
      "learning_rate": 1.5265391191283025e-05,
      "loss": 0.4453,
      "step": 10934
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.631192684173584,
      "learning_rate": 1.5264525361389817e-05,
      "loss": 0.5196,
      "step": 10935
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6779159307479858,
      "learning_rate": 1.526365947689507e-05,
      "loss": 0.5017,
      "step": 10936
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5273417234420776,
      "learning_rate": 1.526279353780775e-05,
      "loss": 0.4694,
      "step": 10937
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6391725540161133,
      "learning_rate": 1.526192754413685e-05,
      "loss": 0.4781,
      "step": 10938
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7095348834991455,
      "learning_rate": 1.5261061495891345e-05,
      "loss": 0.5389,
      "step": 10939
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6736907958984375,
      "learning_rate": 1.5260195393080218e-05,
      "loss": 0.5085,
      "step": 10940
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6352232694625854,
      "learning_rate": 1.5259329235712455e-05,
      "loss": 0.512,
      "step": 10941
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5725146532058716,
      "learning_rate": 1.5258463023797037e-05,
      "loss": 0.4674,
      "step": 10942
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6571218967437744,
      "learning_rate": 1.5257596757342948e-05,
      "loss": 0.4668,
      "step": 10943
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5368194580078125,
      "learning_rate": 1.5256730436359177e-05,
      "loss": 0.4263,
      "step": 10944
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6376365423202515,
      "learning_rate": 1.5255864060854702e-05,
      "loss": 0.5998,
      "step": 10945
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.577504277229309,
      "learning_rate": 1.5254997630838513e-05,
      "loss": 0.4213,
      "step": 10946
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7827569246292114,
      "learning_rate": 1.5254131146319595e-05,
      "loss": 0.5376,
      "step": 10947
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5592808723449707,
      "learning_rate": 1.5253264607306936e-05,
      "loss": 0.4211,
      "step": 10948
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9571738243103027,
      "learning_rate": 1.5252398013809525e-05,
      "loss": 0.5026,
      "step": 10949
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5502259731292725,
      "learning_rate": 1.5251531365836343e-05,
      "loss": 0.4494,
      "step": 10950
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5054926872253418,
      "learning_rate": 1.5250664663396383e-05,
      "loss": 0.4754,
      "step": 10951
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.642242193222046,
      "learning_rate": 1.524979790649864e-05,
      "loss": 0.4543,
      "step": 10952
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6873559951782227,
      "learning_rate": 1.524893109515209e-05,
      "loss": 0.5251,
      "step": 10953
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5959138870239258,
      "learning_rate": 1.5248064229365738e-05,
      "loss": 0.4315,
      "step": 10954
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6106716394424438,
      "learning_rate": 1.5247197309148564e-05,
      "loss": 0.5246,
      "step": 10955
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.590765357017517,
      "learning_rate": 1.5246330334509564e-05,
      "loss": 0.4435,
      "step": 10956
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.660776972770691,
      "learning_rate": 1.524546330545773e-05,
      "loss": 0.5009,
      "step": 10957
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5434064865112305,
      "learning_rate": 1.5244596222002056e-05,
      "loss": 0.4431,
      "step": 10958
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.57459557056427,
      "learning_rate": 1.5243729084151526e-05,
      "loss": 0.5035,
      "step": 10959
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5505857467651367,
      "learning_rate": 1.5242861891915146e-05,
      "loss": 0.4539,
      "step": 10960
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7174830436706543,
      "learning_rate": 1.5241994645301903e-05,
      "loss": 0.5253,
      "step": 10961
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.694068193435669,
      "learning_rate": 1.5241127344320795e-05,
      "loss": 0.5181,
      "step": 10962
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6659307479858398,
      "learning_rate": 1.5240259988980807e-05,
      "loss": 0.5734,
      "step": 10963
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5698307752609253,
      "learning_rate": 1.523939257929095e-05,
      "loss": 0.4787,
      "step": 10964
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6454157829284668,
      "learning_rate": 1.5238525115260215e-05,
      "loss": 0.5511,
      "step": 10965
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5152543783187866,
      "learning_rate": 1.5237657596897596e-05,
      "loss": 0.4464,
      "step": 10966
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6560314893722534,
      "learning_rate": 1.5236790024212092e-05,
      "loss": 0.5037,
      "step": 10967
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.720552921295166,
      "learning_rate": 1.5235922397212703e-05,
      "loss": 0.4914,
      "step": 10968
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.603201150894165,
      "learning_rate": 1.5235054715908422e-05,
      "loss": 0.532,
      "step": 10969
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.591327428817749,
      "learning_rate": 1.5234186980308256e-05,
      "loss": 0.4771,
      "step": 10970
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5450794696807861,
      "learning_rate": 1.5233319190421198e-05,
      "loss": 0.5006,
      "step": 10971
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6103222370147705,
      "learning_rate": 1.5232451346256257e-05,
      "loss": 0.4895,
      "step": 10972
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1163971424102783,
      "learning_rate": 1.5231583447822423e-05,
      "loss": 1.0435,
      "step": 10973
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8605741262435913,
      "learning_rate": 1.5230715495128706e-05,
      "loss": 1.0632,
      "step": 10974
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7572413682937622,
      "learning_rate": 1.5229847488184102e-05,
      "loss": 0.5355,
      "step": 10975
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5850797891616821,
      "learning_rate": 1.5228979426997618e-05,
      "loss": 0.4437,
      "step": 10976
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5898518562316895,
      "learning_rate": 1.5228111311578255e-05,
      "loss": 0.5309,
      "step": 10977
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6776312589645386,
      "learning_rate": 1.5227243141935022e-05,
      "loss": 0.4915,
      "step": 10978
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.590783953666687,
      "learning_rate": 1.5226374918076913e-05,
      "loss": 0.4756,
      "step": 10979
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6598299741744995,
      "learning_rate": 1.5225506640012943e-05,
      "loss": 0.476,
      "step": 10980
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6748895645141602,
      "learning_rate": 1.522463830775211e-05,
      "loss": 0.5596,
      "step": 10981
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.565319538116455,
      "learning_rate": 1.5223769921303424e-05,
      "loss": 0.4489,
      "step": 10982
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5204392671585083,
      "learning_rate": 1.5222901480675894e-05,
      "loss": 0.5157,
      "step": 10983
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.542434811592102,
      "learning_rate": 1.522203298587852e-05,
      "loss": 0.4423,
      "step": 10984
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.739366054534912,
      "learning_rate": 1.5221164436920315e-05,
      "loss": 0.5577,
      "step": 10985
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5745424032211304,
      "learning_rate": 1.5220295833810284e-05,
      "loss": 0.4543,
      "step": 10986
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6198363304138184,
      "learning_rate": 1.5219427176557437e-05,
      "loss": 0.5019,
      "step": 10987
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6103229522705078,
      "learning_rate": 1.5218558465170788e-05,
      "loss": 0.479,
      "step": 10988
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6407936811447144,
      "learning_rate": 1.521768969965934e-05,
      "loss": 0.5039,
      "step": 10989
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.544044017791748,
      "learning_rate": 1.5216820880032105e-05,
      "loss": 0.4795,
      "step": 10990
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6147081851959229,
      "learning_rate": 1.5215952006298098e-05,
      "loss": 0.5214,
      "step": 10991
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6778188943862915,
      "learning_rate": 1.5215083078466326e-05,
      "loss": 0.4773,
      "step": 10992
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.553776741027832,
      "learning_rate": 1.5214214096545807e-05,
      "loss": 0.4826,
      "step": 10993
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5507934093475342,
      "learning_rate": 1.5213345060545548e-05,
      "loss": 0.4426,
      "step": 10994
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6355191469192505,
      "learning_rate": 1.5212475970474559e-05,
      "loss": 0.5211,
      "step": 10995
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.53421151638031,
      "learning_rate": 1.5211606826341867e-05,
      "loss": 0.4324,
      "step": 10996
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5492241382598877,
      "learning_rate": 1.521073762815647e-05,
      "loss": 0.4771,
      "step": 10997
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5821877717971802,
      "learning_rate": 1.5209868375927399e-05,
      "loss": 0.4516,
      "step": 10998
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6041715145111084,
      "learning_rate": 1.5208999069663659e-05,
      "loss": 0.5002,
      "step": 10999
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5570886135101318,
      "learning_rate": 1.520812970937427e-05,
      "loss": 0.4472,
      "step": 11000
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7060797214508057,
      "learning_rate": 1.5207260295068245e-05,
      "loss": 0.5382,
      "step": 11001
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5953389406204224,
      "learning_rate": 1.5206390826754607e-05,
      "loss": 0.4652,
      "step": 11002
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6556113958358765,
      "learning_rate": 1.5205521304442367e-05,
      "loss": 0.5037,
      "step": 11003
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5513852834701538,
      "learning_rate": 1.5204651728140551e-05,
      "loss": 0.4381,
      "step": 11004
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5770996809005737,
      "learning_rate": 1.5203782097858173e-05,
      "loss": 0.5413,
      "step": 11005
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5584816932678223,
      "learning_rate": 1.5202912413604254e-05,
      "loss": 0.4849,
      "step": 11006
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.696333885192871,
      "learning_rate": 1.520204267538781e-05,
      "loss": 0.4998,
      "step": 11007
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6434210538864136,
      "learning_rate": 1.5201172883217866e-05,
      "loss": 0.4898,
      "step": 11008
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7034626007080078,
      "learning_rate": 1.5200303037103444e-05,
      "loss": 0.5967,
      "step": 11009
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.504278302192688,
      "learning_rate": 1.5199433137053564e-05,
      "loss": 0.4274,
      "step": 11010
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5794657468795776,
      "learning_rate": 1.5198563183077246e-05,
      "loss": 0.5224,
      "step": 11011
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5511677265167236,
      "learning_rate": 1.5197693175183516e-05,
      "loss": 0.455,
      "step": 11012
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.514037847518921,
      "learning_rate": 1.5196823113381394e-05,
      "loss": 0.4902,
      "step": 11013
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5289160013198853,
      "learning_rate": 1.5195952997679912e-05,
      "loss": 0.4526,
      "step": 11014
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5848661661148071,
      "learning_rate": 1.5195082828088082e-05,
      "loss": 0.5304,
      "step": 11015
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6003472805023193,
      "learning_rate": 1.5194212604614937e-05,
      "loss": 0.4551,
      "step": 11016
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6256930828094482,
      "learning_rate": 1.5193342327269504e-05,
      "loss": 0.5617,
      "step": 11017
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6015897989273071,
      "learning_rate": 1.5192471996060806e-05,
      "loss": 0.5039,
      "step": 11018
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9453015327453613,
      "learning_rate": 1.519160161099787e-05,
      "loss": 1.0402,
      "step": 11019
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8935436010360718,
      "learning_rate": 1.5190731172089725e-05,
      "loss": 1.0282,
      "step": 11020
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6289530992507935,
      "learning_rate": 1.5189860679345393e-05,
      "loss": 0.4728,
      "step": 11021
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5504628419876099,
      "learning_rate": 1.5188990132773911e-05,
      "loss": 0.4532,
      "step": 11022
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5762779712677002,
      "learning_rate": 1.51881195323843e-05,
      "loss": 0.5169,
      "step": 11023
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5835890769958496,
      "learning_rate": 1.5187248878185598e-05,
      "loss": 0.4803,
      "step": 11024
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6371989250183105,
      "learning_rate": 1.5186378170186826e-05,
      "loss": 0.4957,
      "step": 11025
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.605016827583313,
      "learning_rate": 1.518550740839702e-05,
      "loss": 0.4832,
      "step": 11026
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5810608863830566,
      "learning_rate": 1.5184636592825211e-05,
      "loss": 0.5085,
      "step": 11027
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.635993480682373,
      "learning_rate": 1.5183765723480428e-05,
      "loss": 0.4686,
      "step": 11028
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6496303081512451,
      "learning_rate": 1.5182894800371708e-05,
      "loss": 0.5192,
      "step": 11029
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4972684383392334,
      "learning_rate": 1.518202382350808e-05,
      "loss": 0.4179,
      "step": 11030
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6353858709335327,
      "learning_rate": 1.5181152792898577e-05,
      "loss": 0.5656,
      "step": 11031
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6044999361038208,
      "learning_rate": 1.5180281708552236e-05,
      "loss": 0.4575,
      "step": 11032
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5172358751296997,
      "learning_rate": 1.5179410570478087e-05,
      "loss": 0.4878,
      "step": 11033
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.099561929702759,
      "learning_rate": 1.5178539378685173e-05,
      "loss": 0.4253,
      "step": 11034
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.633937954902649,
      "learning_rate": 1.5177668133182524e-05,
      "loss": 0.4931,
      "step": 11035
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6311157941818237,
      "learning_rate": 1.5176796833979173e-05,
      "loss": 0.4555,
      "step": 11036
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.647826910018921,
      "learning_rate": 1.5175925481084167e-05,
      "loss": 0.5215,
      "step": 11037
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6008949279785156,
      "learning_rate": 1.5175054074506531e-05,
      "loss": 0.4716,
      "step": 11038
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8161474466323853,
      "learning_rate": 1.517418261425531e-05,
      "loss": 0.5136,
      "step": 11039
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.584831714630127,
      "learning_rate": 1.5173311100339544e-05,
      "loss": 0.4577,
      "step": 11040
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5497711896896362,
      "learning_rate": 1.5172439532768266e-05,
      "loss": 0.5091,
      "step": 11041
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5956369638442993,
      "learning_rate": 1.5171567911550522e-05,
      "loss": 0.4624,
      "step": 11042
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6938096284866333,
      "learning_rate": 1.5170696236695345e-05,
      "loss": 0.55,
      "step": 11043
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.570640206336975,
      "learning_rate": 1.5169824508211783e-05,
      "loss": 0.4549,
      "step": 11044
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.661210060119629,
      "learning_rate": 1.5168952726108874e-05,
      "loss": 0.5305,
      "step": 11045
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5964797735214233,
      "learning_rate": 1.5168080890395657e-05,
      "loss": 0.4484,
      "step": 11046
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5360709428787231,
      "learning_rate": 1.5167209001081182e-05,
      "loss": 0.4905,
      "step": 11047
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.602068543434143,
      "learning_rate": 1.5166337058174482e-05,
      "loss": 0.4857,
      "step": 11048
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6157790422439575,
      "learning_rate": 1.5165465061684605e-05,
      "loss": 0.4854,
      "step": 11049
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6350239515304565,
      "learning_rate": 1.51645930116206e-05,
      "loss": 0.434,
      "step": 11050
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5961236953735352,
      "learning_rate": 1.5163720907991504e-05,
      "loss": 0.5197,
      "step": 11051
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6659483909606934,
      "learning_rate": 1.5162848750806365e-05,
      "loss": 0.4614,
      "step": 11052
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7158044576644897,
      "learning_rate": 1.516197654007423e-05,
      "loss": 0.5316,
      "step": 11053
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.657036542892456,
      "learning_rate": 1.5161104275804142e-05,
      "loss": 0.4665,
      "step": 11054
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6703238487243652,
      "learning_rate": 1.5160231958005152e-05,
      "loss": 0.5129,
      "step": 11055
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6886450052261353,
      "learning_rate": 1.5159359586686302e-05,
      "loss": 0.5011,
      "step": 11056
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6087957620620728,
      "learning_rate": 1.5158487161856643e-05,
      "loss": 0.5254,
      "step": 11057
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.481999158859253,
      "learning_rate": 1.5157614683525226e-05,
      "loss": 0.422,
      "step": 11058
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.2230288982391357,
      "learning_rate": 1.5156742151701094e-05,
      "loss": 1.162,
      "step": 11059
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8144268989562988,
      "learning_rate": 1.5155869566393304e-05,
      "loss": 1.0676,
      "step": 11060
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6236457824707031,
      "learning_rate": 1.5154996927610897e-05,
      "loss": 0.5129,
      "step": 11061
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7988839149475098,
      "learning_rate": 1.5154124235362933e-05,
      "loss": 0.4699,
      "step": 11062
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6437509059906006,
      "learning_rate": 1.5153251489658455e-05,
      "loss": 0.5213,
      "step": 11063
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.58286452293396,
      "learning_rate": 1.515237869050652e-05,
      "loss": 0.4516,
      "step": 11064
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6218842267990112,
      "learning_rate": 1.515150583791618e-05,
      "loss": 0.5068,
      "step": 11065
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5452927350997925,
      "learning_rate": 1.5150632931896487e-05,
      "loss": 0.4678,
      "step": 11066
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6049957275390625,
      "learning_rate": 1.514975997245649e-05,
      "loss": 0.4895,
      "step": 11067
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5507502555847168,
      "learning_rate": 1.5148886959605254e-05,
      "loss": 0.4307,
      "step": 11068
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6427128314971924,
      "learning_rate": 1.5148013893351823e-05,
      "loss": 0.5053,
      "step": 11069
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5559459924697876,
      "learning_rate": 1.5147140773705256e-05,
      "loss": 0.4606,
      "step": 11070
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6106094121932983,
      "learning_rate": 1.514626760067461e-05,
      "loss": 0.5318,
      "step": 11071
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5983233451843262,
      "learning_rate": 1.5145394374268936e-05,
      "loss": 0.4723,
      "step": 11072
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6421959400177002,
      "learning_rate": 1.5144521094497297e-05,
      "loss": 0.4979,
      "step": 11073
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5575242042541504,
      "learning_rate": 1.514364776136875e-05,
      "loss": 0.4325,
      "step": 11074
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.547875165939331,
      "learning_rate": 1.5142774374892346e-05,
      "loss": 0.5085,
      "step": 11075
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4881930351257324,
      "learning_rate": 1.5141900935077151e-05,
      "loss": 0.4503,
      "step": 11076
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6501219272613525,
      "learning_rate": 1.5141027441932217e-05,
      "loss": 0.5125,
      "step": 11077
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5730923414230347,
      "learning_rate": 1.5140153895466614e-05,
      "loss": 0.4468,
      "step": 11078
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.61137855052948,
      "learning_rate": 1.513928029568939e-05,
      "loss": 0.5105,
      "step": 11079
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.583749532699585,
      "learning_rate": 1.5138406642609615e-05,
      "loss": 0.4607,
      "step": 11080
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5916589498519897,
      "learning_rate": 1.5137532936236343e-05,
      "loss": 0.5039,
      "step": 11081
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5368331670761108,
      "learning_rate": 1.513665917657864e-05,
      "loss": 0.4408,
      "step": 11082
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6794202327728271,
      "learning_rate": 1.5135785363645572e-05,
      "loss": 0.5418,
      "step": 11083
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.508182406425476,
      "learning_rate": 1.5134911497446192e-05,
      "loss": 0.4686,
      "step": 11084
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6033419370651245,
      "learning_rate": 1.5134037577989568e-05,
      "loss": 0.5184,
      "step": 11085
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6311055421829224,
      "learning_rate": 1.5133163605284768e-05,
      "loss": 0.4575,
      "step": 11086
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6590360403060913,
      "learning_rate": 1.5132289579340853e-05,
      "loss": 0.5375,
      "step": 11087
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7078688144683838,
      "learning_rate": 1.5131415500166885e-05,
      "loss": 0.4794,
      "step": 11088
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5924705266952515,
      "learning_rate": 1.5130541367771937e-05,
      "loss": 0.5377,
      "step": 11089
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5474656820297241,
      "learning_rate": 1.5129667182165067e-05,
      "loss": 0.472,
      "step": 11090
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.757986068725586,
      "learning_rate": 1.5128792943355348e-05,
      "loss": 0.5153,
      "step": 11091
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5033046007156372,
      "learning_rate": 1.5127918651351844e-05,
      "loss": 0.4359,
      "step": 11092
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6376464366912842,
      "learning_rate": 1.5127044306163621e-05,
      "loss": 0.4931,
      "step": 11093
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5374679565429688,
      "learning_rate": 1.5126169907799754e-05,
      "loss": 0.432,
      "step": 11094
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.612361192703247,
      "learning_rate": 1.5125295456269307e-05,
      "loss": 0.5388,
      "step": 11095
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5684709548950195,
      "learning_rate": 1.5124420951581348e-05,
      "loss": 0.4794,
      "step": 11096
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.555282711982727,
      "learning_rate": 1.5123546393744954e-05,
      "loss": 0.4801,
      "step": 11097
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5868515968322754,
      "learning_rate": 1.5122671782769187e-05,
      "loss": 0.4844,
      "step": 11098
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6338813304901123,
      "learning_rate": 1.5121797118663124e-05,
      "loss": 0.5038,
      "step": 11099
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.578464150428772,
      "learning_rate": 1.5120922401435832e-05,
      "loss": 0.4705,
      "step": 11100
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5507668256759644,
      "learning_rate": 1.5120047631096388e-05,
      "loss": 0.464,
      "step": 11101
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5905203819274902,
      "learning_rate": 1.5119172807653861e-05,
      "loss": 0.4553,
      "step": 11102
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6570734977722168,
      "learning_rate": 1.511829793111733e-05,
      "loss": 0.478,
      "step": 11103
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.099656105041504,
      "learning_rate": 1.5117423001495862e-05,
      "loss": 0.4942,
      "step": 11104
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.550535798072815,
      "learning_rate": 1.5116548018798534e-05,
      "loss": 0.5026,
      "step": 11105
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6830109357833862,
      "learning_rate": 1.5115672983034424e-05,
      "loss": 0.5011,
      "step": 11106
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5931034088134766,
      "learning_rate": 1.5114797894212601e-05,
      "loss": 0.5089,
      "step": 11107
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5673860311508179,
      "learning_rate": 1.5113922752342144e-05,
      "loss": 0.4495,
      "step": 11108
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8508220911026,
      "learning_rate": 1.5113047557432136e-05,
      "loss": 0.5727,
      "step": 11109
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6716240644454956,
      "learning_rate": 1.5112172309491647e-05,
      "loss": 0.4902,
      "step": 11110
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.616540789604187,
      "learning_rate": 1.5111297008529753e-05,
      "loss": 0.4909,
      "step": 11111
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.563286304473877,
      "learning_rate": 1.5110421654555538e-05,
      "loss": 0.4764,
      "step": 11112
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7009060382843018,
      "learning_rate": 1.5109546247578077e-05,
      "loss": 0.6378,
      "step": 11113
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5610202550888062,
      "learning_rate": 1.5108670787606453e-05,
      "loss": 0.4388,
      "step": 11114
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.950893759727478,
      "learning_rate": 1.5107795274649743e-05,
      "loss": 0.9985,
      "step": 11115
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.673146367073059,
      "learning_rate": 1.5106919708717027e-05,
      "loss": 1.0795,
      "step": 11116
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6989513635635376,
      "learning_rate": 1.5106044089817391e-05,
      "loss": 0.5222,
      "step": 11117
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6041090488433838,
      "learning_rate": 1.5105168417959908e-05,
      "loss": 0.4295,
      "step": 11118
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7614141702651978,
      "learning_rate": 1.5104292693153668e-05,
      "loss": 0.5131,
      "step": 11119
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6504815816879272,
      "learning_rate": 1.510341691540775e-05,
      "loss": 0.4726,
      "step": 11120
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6550343036651611,
      "learning_rate": 1.5102541084731237e-05,
      "loss": 0.5771,
      "step": 11121
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.512555480003357,
      "learning_rate": 1.5101665201133214e-05,
      "loss": 0.4265,
      "step": 11122
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6439186334609985,
      "learning_rate": 1.5100789264622764e-05,
      "loss": 0.5216,
      "step": 11123
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6056227684020996,
      "learning_rate": 1.5099913275208975e-05,
      "loss": 0.4603,
      "step": 11124
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6013089418411255,
      "learning_rate": 1.5099037232900928e-05,
      "loss": 0.4884,
      "step": 11125
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5787200927734375,
      "learning_rate": 1.509816113770771e-05,
      "loss": 0.525,
      "step": 11126
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6473381519317627,
      "learning_rate": 1.5097284989638415e-05,
      "loss": 0.5141,
      "step": 11127
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7728497982025146,
      "learning_rate": 1.5096408788702115e-05,
      "loss": 0.4645,
      "step": 11128
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.614313006401062,
      "learning_rate": 1.509553253490791e-05,
      "loss": 0.5135,
      "step": 11129
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6505327224731445,
      "learning_rate": 1.5094656228264887e-05,
      "loss": 0.4766,
      "step": 11130
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.618411898612976,
      "learning_rate": 1.509377986878213e-05,
      "loss": 0.5084,
      "step": 11131
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5425418615341187,
      "learning_rate": 1.5092903456468731e-05,
      "loss": 0.4452,
      "step": 11132
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.2302074432373047,
      "learning_rate": 1.5092026991333777e-05,
      "loss": 1.0613,
      "step": 11133
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.800959587097168,
      "learning_rate": 1.5091150473386362e-05,
      "loss": 1.0919,
      "step": 11134
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8216314315795898,
      "learning_rate": 1.5090273902635578e-05,
      "loss": 0.5342,
      "step": 11135
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6822582483291626,
      "learning_rate": 1.508939727909051e-05,
      "loss": 0.4952,
      "step": 11136
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6727800369262695,
      "learning_rate": 1.5088520602760257e-05,
      "loss": 0.5556,
      "step": 11137
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.631772756576538,
      "learning_rate": 1.5087643873653904e-05,
      "loss": 0.4813,
      "step": 11138
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5831079483032227,
      "learning_rate": 1.5086767091780552e-05,
      "loss": 0.506,
      "step": 11139
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7763748168945312,
      "learning_rate": 1.508589025714929e-05,
      "loss": 0.4839,
      "step": 11140
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.581946849822998,
      "learning_rate": 1.5085013369769215e-05,
      "loss": 0.4997,
      "step": 11141
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.704188585281372,
      "learning_rate": 1.508413642964942e-05,
      "loss": 0.4768,
      "step": 11142
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.709043264389038,
      "learning_rate": 1.5083259436798998e-05,
      "loss": 0.5624,
      "step": 11143
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5666000843048096,
      "learning_rate": 1.5082382391227046e-05,
      "loss": 0.4552,
      "step": 11144
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7140175104141235,
      "learning_rate": 1.5081505292942663e-05,
      "loss": 0.5116,
      "step": 11145
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6407408714294434,
      "learning_rate": 1.5080628141954945e-05,
      "loss": 0.4543,
      "step": 11146
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6960062980651855,
      "learning_rate": 1.507975093827299e-05,
      "loss": 0.566,
      "step": 11147
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.591340184211731,
      "learning_rate": 1.5078873681905895e-05,
      "loss": 0.447,
      "step": 11148
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6782146692276,
      "learning_rate": 1.5077996372862756e-05,
      "loss": 0.4929,
      "step": 11149
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5711824893951416,
      "learning_rate": 1.5077119011152675e-05,
      "loss": 0.461,
      "step": 11150
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.975329875946045,
      "learning_rate": 1.5076241596784753e-05,
      "loss": 1.0639,
      "step": 11151
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7870774269104004,
      "learning_rate": 1.5075364129768085e-05,
      "loss": 1.0743,
      "step": 11152
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.817234754562378,
      "learning_rate": 1.5074486610111778e-05,
      "loss": 0.558,
      "step": 11153
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.666465401649475,
      "learning_rate": 1.5073609037824928e-05,
      "loss": 0.4518,
      "step": 11154
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6300041675567627,
      "learning_rate": 1.5072731412916641e-05,
      "loss": 0.4872,
      "step": 11155
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.611695647239685,
      "learning_rate": 1.5071853735396016e-05,
      "loss": 0.4543,
      "step": 11156
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8707040548324585,
      "learning_rate": 1.507097600527216e-05,
      "loss": 0.5908,
      "step": 11157
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5523664951324463,
      "learning_rate": 1.5070098222554173e-05,
      "loss": 0.4547,
      "step": 11158
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.639561653137207,
      "learning_rate": 1.5069220387251157e-05,
      "loss": 0.5267,
      "step": 11159
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6224437952041626,
      "learning_rate": 1.5068342499372222e-05,
      "loss": 0.4485,
      "step": 11160
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6088857650756836,
      "learning_rate": 1.506746455892647e-05,
      "loss": 0.5014,
      "step": 11161
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6475224494934082,
      "learning_rate": 1.5066586565923008e-05,
      "loss": 0.4709,
      "step": 11162
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5695713758468628,
      "learning_rate": 1.5065708520370943e-05,
      "loss": 0.4979,
      "step": 11163
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.817649245262146,
      "learning_rate": 1.5064830422279376e-05,
      "loss": 0.4872,
      "step": 11164
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6001384258270264,
      "learning_rate": 1.506395227165742e-05,
      "loss": 0.4956,
      "step": 11165
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6514283418655396,
      "learning_rate": 1.5063074068514181e-05,
      "loss": 0.4521,
      "step": 11166
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5641151666641235,
      "learning_rate": 1.506219581285877e-05,
      "loss": 0.5361,
      "step": 11167
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.520477533340454,
      "learning_rate": 1.5061317504700293e-05,
      "loss": 0.4414,
      "step": 11168
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5266672372817993,
      "learning_rate": 1.5060439144047859e-05,
      "loss": 0.4963,
      "step": 11169
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6198731660842896,
      "learning_rate": 1.5059560730910576e-05,
      "loss": 0.5558,
      "step": 11170
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6908626556396484,
      "learning_rate": 1.5058682265297564e-05,
      "loss": 0.5312,
      "step": 11171
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5967706441879272,
      "learning_rate": 1.5057803747217922e-05,
      "loss": 0.4628,
      "step": 11172
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5456757545471191,
      "learning_rate": 1.5056925176680773e-05,
      "loss": 0.5311,
      "step": 11173
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5922685861587524,
      "learning_rate": 1.5056046553695218e-05,
      "loss": 0.4821,
      "step": 11174
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.5570015907287598,
      "learning_rate": 1.5055167878270376e-05,
      "loss": 1.0158,
      "step": 11175
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.205233335494995,
      "learning_rate": 1.5054289150415363e-05,
      "loss": 1.0619,
      "step": 11176
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.667951226234436,
      "learning_rate": 1.5053410370139286e-05,
      "loss": 0.5507,
      "step": 11177
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.717956781387329,
      "learning_rate": 1.5052531537451265e-05,
      "loss": 0.4787,
      "step": 11178
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5556837320327759,
      "learning_rate": 1.505165265236041e-05,
      "loss": 0.495,
      "step": 11179
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7023845911026,
      "learning_rate": 1.5050773714875839e-05,
      "loss": 0.456,
      "step": 11180
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.616647720336914,
      "learning_rate": 1.5049894725006674e-05,
      "loss": 0.4713,
      "step": 11181
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.467068076133728,
      "learning_rate": 1.5049015682762016e-05,
      "loss": 0.4283,
      "step": 11182
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6589103937149048,
      "learning_rate": 1.5048136588150999e-05,
      "loss": 0.5114,
      "step": 11183
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.982234239578247,
      "learning_rate": 1.504725744118273e-05,
      "loss": 0.4447,
      "step": 11184
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5172559022903442,
      "learning_rate": 1.504637824186633e-05,
      "loss": 0.4692,
      "step": 11185
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5388685464859009,
      "learning_rate": 1.504549899021092e-05,
      "loss": 0.4705,
      "step": 11186
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6410337686538696,
      "learning_rate": 1.5044619686225616e-05,
      "loss": 0.527,
      "step": 11187
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6704472303390503,
      "learning_rate": 1.5043740329919539e-05,
      "loss": 0.5174,
      "step": 11188
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7545350790023804,
      "learning_rate": 1.5042860921301811e-05,
      "loss": 0.5089,
      "step": 11189
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5691994428634644,
      "learning_rate": 1.5041981460381547e-05,
      "loss": 0.4552,
      "step": 11190
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6385923624038696,
      "learning_rate": 1.5041101947167878e-05,
      "loss": 0.4958,
      "step": 11191
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5582060813903809,
      "learning_rate": 1.504022238166992e-05,
      "loss": 0.4544,
      "step": 11192
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6638673543930054,
      "learning_rate": 1.5039342763896794e-05,
      "loss": 0.5368,
      "step": 11193
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.581583857536316,
      "learning_rate": 1.5038463093857628e-05,
      "loss": 0.4588,
      "step": 11194
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5712060928344727,
      "learning_rate": 1.5037583371561538e-05,
      "loss": 0.5093,
      "step": 11195
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.520882248878479,
      "learning_rate": 1.5036703597017657e-05,
      "loss": 0.4281,
      "step": 11196
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5610568523406982,
      "learning_rate": 1.5035823770235105e-05,
      "loss": 0.4903,
      "step": 11197
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.515783429145813,
      "learning_rate": 1.5034943891223003e-05,
      "loss": 0.4378,
      "step": 11198
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6507591009140015,
      "learning_rate": 1.5034063959990487e-05,
      "loss": 0.5141,
      "step": 11199
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6069746017456055,
      "learning_rate": 1.5033183976546677e-05,
      "loss": 0.4452,
      "step": 11200
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5946998596191406,
      "learning_rate": 1.5032303940900701e-05,
      "loss": 0.4494,
      "step": 11201
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5562015771865845,
      "learning_rate": 1.5031423853061683e-05,
      "loss": 0.5041,
      "step": 11202
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.67539644241333,
      "learning_rate": 1.503054371303876e-05,
      "loss": 0.4994,
      "step": 11203
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6079671382904053,
      "learning_rate": 1.5029663520841049e-05,
      "loss": 0.4695,
      "step": 11204
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.562889814376831,
      "learning_rate": 1.5028783276477686e-05,
      "loss": 0.5188,
      "step": 11205
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.962233543395996,
      "learning_rate": 1.5027902979957798e-05,
      "loss": 0.4654,
      "step": 11206
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7308975458145142,
      "learning_rate": 1.502702263129052e-05,
      "loss": 0.518,
      "step": 11207
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6171278953552246,
      "learning_rate": 1.5026142230484975e-05,
      "loss": 0.4547,
      "step": 11208
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.002614974975586,
      "learning_rate": 1.50252617775503e-05,
      "loss": 1.0371,
      "step": 11209
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8287631273269653,
      "learning_rate": 1.5024381272495623e-05,
      "loss": 1.0173,
      "step": 11210
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7084784507751465,
      "learning_rate": 1.5023500715330077e-05,
      "loss": 0.5338,
      "step": 11211
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6725293397903442,
      "learning_rate": 1.5022620106062798e-05,
      "loss": 0.4501,
      "step": 11212
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.675844669342041,
      "learning_rate": 1.5021739444702915e-05,
      "loss": 0.5186,
      "step": 11213
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.577983021736145,
      "learning_rate": 1.5020858731259565e-05,
      "loss": 0.4597,
      "step": 11214
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7798255681991577,
      "learning_rate": 1.5019977965741881e-05,
      "loss": 0.5318,
      "step": 11215
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5726033449172974,
      "learning_rate": 1.5019097148158995e-05,
      "loss": 0.4466,
      "step": 11216
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5576386451721191,
      "learning_rate": 1.5018216278520052e-05,
      "loss": 0.4686,
      "step": 11217
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5767958164215088,
      "learning_rate": 1.5017335356834174e-05,
      "loss": 0.4775,
      "step": 11218
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6546356678009033,
      "learning_rate": 1.501645438311051e-05,
      "loss": 0.5561,
      "step": 11219
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5157781839370728,
      "learning_rate": 1.5015573357358193e-05,
      "loss": 0.4379,
      "step": 11220
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6141587495803833,
      "learning_rate": 1.5014692279586359e-05,
      "loss": 0.4996,
      "step": 11221
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6314793825149536,
      "learning_rate": 1.5013811149804147e-05,
      "loss": 0.4727,
      "step": 11222
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.531419277191162,
      "learning_rate": 1.5012929968020695e-05,
      "loss": 0.5575,
      "step": 11223
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.650229811668396,
      "learning_rate": 1.5012048734245141e-05,
      "loss": 0.5164,
      "step": 11224
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.544748306274414,
      "learning_rate": 1.5011167448486635e-05,
      "loss": 0.459,
      "step": 11225
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6162123680114746,
      "learning_rate": 1.5010286110754302e-05,
      "loss": 0.4563,
      "step": 11226
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6212855577468872,
      "learning_rate": 1.5009404721057293e-05,
      "loss": 0.5049,
      "step": 11227
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6094688177108765,
      "learning_rate": 1.5008523279404746e-05,
      "loss": 0.4516,
      "step": 11228
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6668647527694702,
      "learning_rate": 1.5007641785805801e-05,
      "loss": 0.5261,
      "step": 11229
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6448272466659546,
      "learning_rate": 1.5006760240269607e-05,
      "loss": 0.4642,
      "step": 11230
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5760278701782227,
      "learning_rate": 1.5005878642805302e-05,
      "loss": 0.5071,
      "step": 11231
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5108691453933716,
      "learning_rate": 1.500499699342203e-05,
      "loss": 0.4289,
      "step": 11232
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.59291672706604,
      "learning_rate": 1.5004115292128938e-05,
      "loss": 0.5048,
      "step": 11233
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5487593412399292,
      "learning_rate": 1.5003233538935163e-05,
      "loss": 0.4643,
      "step": 11234
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6921807527542114,
      "learning_rate": 1.500235173384986e-05,
      "loss": 0.474,
      "step": 11235
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4860374927520752,
      "learning_rate": 1.500146987688217e-05,
      "loss": 0.4312,
      "step": 11236
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6997876167297363,
      "learning_rate": 1.5000587968041237e-05,
      "loss": 0.4895,
      "step": 11237
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5875554084777832,
      "learning_rate": 1.4999706007336215e-05,
      "loss": 0.4568,
      "step": 11238
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6061530113220215,
      "learning_rate": 1.4998823994776243e-05,
      "loss": 0.4926,
      "step": 11239
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5991438627243042,
      "learning_rate": 1.4997941930370474e-05,
      "loss": 0.4964,
      "step": 11240
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6211024522781372,
      "learning_rate": 1.4997059814128057e-05,
      "loss": 0.4921,
      "step": 11241
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6595003604888916,
      "learning_rate": 1.4996177646058134e-05,
      "loss": 0.5537,
      "step": 11242
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5579694509506226,
      "learning_rate": 1.4995295426169865e-05,
      "loss": 0.479,
      "step": 11243
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5613293647766113,
      "learning_rate": 1.4994413154472391e-05,
      "loss": 0.4497,
      "step": 11244
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6233649253845215,
      "learning_rate": 1.4993530830974868e-05,
      "loss": 0.54,
      "step": 11245
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6291229724884033,
      "learning_rate": 1.4992648455686443e-05,
      "loss": 0.4506,
      "step": 11246
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7944440841674805,
      "learning_rate": 1.4991766028616273e-05,
      "loss": 0.5478,
      "step": 11247
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4968169927597046,
      "learning_rate": 1.4990883549773505e-05,
      "loss": 0.4369,
      "step": 11248
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6950244903564453,
      "learning_rate": 1.4990001019167295e-05,
      "loss": 0.5584,
      "step": 11249
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.590842604637146,
      "learning_rate": 1.4989118436806791e-05,
      "loss": 0.4368,
      "step": 11250
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5572363138198853,
      "learning_rate": 1.4988235802701156e-05,
      "loss": 0.4847,
      "step": 11251
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.541914701461792,
      "learning_rate": 1.4987353116859535e-05,
      "loss": 0.4971,
      "step": 11252
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6640031337738037,
      "learning_rate": 1.4986470379291092e-05,
      "loss": 0.4762,
      "step": 11253
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.550166368484497,
      "learning_rate": 1.4985587590004972e-05,
      "loss": 0.4386,
      "step": 11254
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5927717685699463,
      "learning_rate": 1.4984704749010339e-05,
      "loss": 0.4944,
      "step": 11255
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.484938621520996,
      "learning_rate": 1.4983821856316347e-05,
      "loss": 0.4508,
      "step": 11256
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.553151249885559,
      "learning_rate": 1.4982938911932152e-05,
      "loss": 0.5324,
      "step": 11257
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4924774169921875,
      "learning_rate": 1.4982055915866914e-05,
      "loss": 0.4424,
      "step": 11258
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6125905513763428,
      "learning_rate": 1.4981172868129787e-05,
      "loss": 0.4965,
      "step": 11259
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.725809097290039,
      "learning_rate": 1.4980289768729931e-05,
      "loss": 0.4554,
      "step": 11260
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.506089448928833,
      "learning_rate": 1.4979406617676512e-05,
      "loss": 0.4736,
      "step": 11261
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6937216520309448,
      "learning_rate": 1.4978523414978678e-05,
      "loss": 0.5079,
      "step": 11262
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5511455535888672,
      "learning_rate": 1.4977640160645599e-05,
      "loss": 0.4607,
      "step": 11263
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.695565938949585,
      "learning_rate": 1.497675685468643e-05,
      "loss": 0.5321,
      "step": 11264
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7263661623001099,
      "learning_rate": 1.4975873497110334e-05,
      "loss": 0.4926,
      "step": 11265
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6224018335342407,
      "learning_rate": 1.4974990087926476e-05,
      "loss": 0.4297,
      "step": 11266
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7147414684295654,
      "learning_rate": 1.4974106627144012e-05,
      "loss": 0.524,
      "step": 11267
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7486201524734497,
      "learning_rate": 1.497322311477211e-05,
      "loss": 0.4764,
      "step": 11268
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.547711730003357,
      "learning_rate": 1.4972339550819929e-05,
      "loss": 0.5041,
      "step": 11269
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5869531631469727,
      "learning_rate": 1.4971455935296637e-05,
      "loss": 0.4656,
      "step": 11270
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7159183025360107,
      "learning_rate": 1.4970572268211401e-05,
      "loss": 0.5334,
      "step": 11271
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5922621488571167,
      "learning_rate": 1.4969688549573378e-05,
      "loss": 0.4924,
      "step": 11272
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.649807095527649,
      "learning_rate": 1.4968804779391736e-05,
      "loss": 0.4836,
      "step": 11273
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.561449408531189,
      "learning_rate": 1.4967920957675648e-05,
      "loss": 0.4366,
      "step": 11274
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5422130823135376,
      "learning_rate": 1.4967037084434272e-05,
      "loss": 0.4853,
      "step": 11275
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.567888617515564,
      "learning_rate": 1.4966153159676782e-05,
      "loss": 0.4568,
      "step": 11276
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5940148830413818,
      "learning_rate": 1.4965269183412341e-05,
      "loss": 0.4983,
      "step": 11277
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.501184105873108,
      "learning_rate": 1.4964385155650116e-05,
      "loss": 0.4603,
      "step": 11278
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.630334734916687,
      "learning_rate": 1.4963501076399283e-05,
      "loss": 0.4772,
      "step": 11279
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6241062879562378,
      "learning_rate": 1.4962616945669002e-05,
      "loss": 0.4593,
      "step": 11280
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5514743328094482,
      "learning_rate": 1.4961732763468451e-05,
      "loss": 0.506,
      "step": 11281
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7178467512130737,
      "learning_rate": 1.4960848529806795e-05,
      "loss": 0.4894,
      "step": 11282
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6374415159225464,
      "learning_rate": 1.495996424469321e-05,
      "loss": 0.53,
      "step": 11283
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.61408269405365,
      "learning_rate": 1.4959079908136862e-05,
      "loss": 0.5045,
      "step": 11284
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5941073894500732,
      "learning_rate": 1.4958195520146926e-05,
      "loss": 0.5172,
      "step": 11285
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6318669319152832,
      "learning_rate": 1.4957311080732572e-05,
      "loss": 0.4824,
      "step": 11286
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.739409327507019,
      "learning_rate": 1.495642658990298e-05,
      "loss": 0.5198,
      "step": 11287
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7232580184936523,
      "learning_rate": 1.4955542047667314e-05,
      "loss": 0.4481,
      "step": 11288
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.525129795074463,
      "learning_rate": 1.4954657454034754e-05,
      "loss": 0.5085,
      "step": 11289
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6047946214675903,
      "learning_rate": 1.495377280901447e-05,
      "loss": 0.5088,
      "step": 11290
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7973964214324951,
      "learning_rate": 1.4952888112615647e-05,
      "loss": 0.5274,
      "step": 11291
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5957720279693604,
      "learning_rate": 1.495200336484745e-05,
      "loss": 0.4668,
      "step": 11292
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6169333457946777,
      "learning_rate": 1.4951118565719061e-05,
      "loss": 0.4972,
      "step": 11293
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5766875743865967,
      "learning_rate": 1.4950233715239655e-05,
      "loss": 0.4503,
      "step": 11294
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7370463609695435,
      "learning_rate": 1.494934881341841e-05,
      "loss": 0.5373,
      "step": 11295
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.523982286453247,
      "learning_rate": 1.49484638602645e-05,
      "loss": 0.4307,
      "step": 11296
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5867477655410767,
      "learning_rate": 1.4947578855787111e-05,
      "loss": 0.4857,
      "step": 11297
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6199880838394165,
      "learning_rate": 1.4946693799995413e-05,
      "loss": 0.4761,
      "step": 11298
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8114157915115356,
      "learning_rate": 1.4945808692898594e-05,
      "loss": 0.5455,
      "step": 11299
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.632258653640747,
      "learning_rate": 1.4944923534505831e-05,
      "loss": 0.4488,
      "step": 11300
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7034772634506226,
      "learning_rate": 1.4944038324826302e-05,
      "loss": 0.5275,
      "step": 11301
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.674296259880066,
      "learning_rate": 1.4943153063869189e-05,
      "loss": 0.4738,
      "step": 11302
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.612144947052002,
      "learning_rate": 1.4942267751643675e-05,
      "loss": 0.4798,
      "step": 11303
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5477463006973267,
      "learning_rate": 1.4941382388158938e-05,
      "loss": 0.4702,
      "step": 11304
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6115778684616089,
      "learning_rate": 1.494049697342417e-05,
      "loss": 0.4972,
      "step": 11305
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5846893787384033,
      "learning_rate": 1.4939611507448543e-05,
      "loss": 0.4378,
      "step": 11306
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8307664394378662,
      "learning_rate": 1.493872599024125e-05,
      "loss": 0.5101,
      "step": 11307
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6463967561721802,
      "learning_rate": 1.493784042181147e-05,
      "loss": 0.4635,
      "step": 11308
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5688890218734741,
      "learning_rate": 1.4936954802168386e-05,
      "loss": 0.4754,
      "step": 11309
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5527838468551636,
      "learning_rate": 1.4936069131321189e-05,
      "loss": 0.443,
      "step": 11310
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.570310115814209,
      "learning_rate": 1.493518340927906e-05,
      "loss": 0.5077,
      "step": 11311
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5695180892944336,
      "learning_rate": 1.493429763605119e-05,
      "loss": 0.4443,
      "step": 11312
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.636893630027771,
      "learning_rate": 1.493341181164676e-05,
      "loss": 0.5193,
      "step": 11313
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6037161350250244,
      "learning_rate": 1.4932525936074961e-05,
      "loss": 0.4683,
      "step": 11314
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6820796728134155,
      "learning_rate": 1.4931640009344983e-05,
      "loss": 0.5213,
      "step": 11315
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7411807775497437,
      "learning_rate": 1.4930754031466009e-05,
      "loss": 0.4633,
      "step": 11316
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6456395387649536,
      "learning_rate": 1.4929868002447232e-05,
      "loss": 0.5204,
      "step": 11317
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5552817583084106,
      "learning_rate": 1.4928981922297842e-05,
      "loss": 0.4534,
      "step": 11318
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7628637552261353,
      "learning_rate": 1.4928095791027027e-05,
      "loss": 0.5265,
      "step": 11319
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5737502574920654,
      "learning_rate": 1.4927209608643977e-05,
      "loss": 0.4561,
      "step": 11320
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7190113067626953,
      "learning_rate": 1.4926323375157885e-05,
      "loss": 0.6024,
      "step": 11321
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5543309450149536,
      "learning_rate": 1.4925437090577943e-05,
      "loss": 0.46,
      "step": 11322
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.204704523086548,
      "learning_rate": 1.4924550754913341e-05,
      "loss": 0.535,
      "step": 11323
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.529744029045105,
      "learning_rate": 1.4923664368173274e-05,
      "loss": 0.4486,
      "step": 11324
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5998928546905518,
      "learning_rate": 1.4922777930366935e-05,
      "loss": 0.5002,
      "step": 11325
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5614391565322876,
      "learning_rate": 1.4921891441503514e-05,
      "loss": 0.4598,
      "step": 11326
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6009358167648315,
      "learning_rate": 1.4921004901592209e-05,
      "loss": 0.5176,
      "step": 11327
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5341147184371948,
      "learning_rate": 1.4920118310642216e-05,
      "loss": 0.4347,
      "step": 11328
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5271704196929932,
      "learning_rate": 1.4919231668662726e-05,
      "loss": 0.4713,
      "step": 11329
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6422582864761353,
      "learning_rate": 1.491834497566294e-05,
      "loss": 0.4929,
      "step": 11330
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7043919563293457,
      "learning_rate": 1.4917458231652048e-05,
      "loss": 0.5157,
      "step": 11331
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6414973735809326,
      "learning_rate": 1.4916571436639253e-05,
      "loss": 0.5095,
      "step": 11332
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5892305374145508,
      "learning_rate": 1.4915684590633753e-05,
      "loss": 0.5095,
      "step": 11333
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.496949315071106,
      "learning_rate": 1.491479769364474e-05,
      "loss": 0.434,
      "step": 11334
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.625788927078247,
      "learning_rate": 1.4913910745681417e-05,
      "loss": 0.4853,
      "step": 11335
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4929801225662231,
      "learning_rate": 1.4913023746752982e-05,
      "loss": 0.4465,
      "step": 11336
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.551784873008728,
      "learning_rate": 1.4912136696868631e-05,
      "loss": 0.4875,
      "step": 11337
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7032488584518433,
      "learning_rate": 1.491124959603757e-05,
      "loss": 0.5123,
      "step": 11338
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6232846975326538,
      "learning_rate": 1.4910362444268997e-05,
      "loss": 0.4931,
      "step": 11339
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.588489294052124,
      "learning_rate": 1.4909475241572113e-05,
      "loss": 0.4388,
      "step": 11340
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6502892971038818,
      "learning_rate": 1.4908587987956122e-05,
      "loss": 0.5262,
      "step": 11341
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5344144105911255,
      "learning_rate": 1.4907700683430224e-05,
      "loss": 0.4456,
      "step": 11342
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6445220708847046,
      "learning_rate": 1.490681332800362e-05,
      "loss": 0.5662,
      "step": 11343
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7194007635116577,
      "learning_rate": 1.4905925921685518e-05,
      "loss": 0.467,
      "step": 11344
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6140422821044922,
      "learning_rate": 1.4905038464485115e-05,
      "loss": 0.5161,
      "step": 11345
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5515716075897217,
      "learning_rate": 1.4904150956411625e-05,
      "loss": 0.4296,
      "step": 11346
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.685224175453186,
      "learning_rate": 1.4903263397474243e-05,
      "loss": 0.4868,
      "step": 11347
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5877400636672974,
      "learning_rate": 1.4902375787682184e-05,
      "loss": 0.4715,
      "step": 11348
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5972670316696167,
      "learning_rate": 1.4901488127044644e-05,
      "loss": 0.5174,
      "step": 11349
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4925658702850342,
      "learning_rate": 1.4900600415570835e-05,
      "loss": 0.4545,
      "step": 11350
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8323653936386108,
      "learning_rate": 1.4899712653269965e-05,
      "loss": 0.5258,
      "step": 11351
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6031029224395752,
      "learning_rate": 1.4898824840151238e-05,
      "loss": 0.4382,
      "step": 11352
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.618072509765625,
      "learning_rate": 1.4897936976223863e-05,
      "loss": 0.4949,
      "step": 11353
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4695868492126465,
      "learning_rate": 1.4897049061497052e-05,
      "loss": 0.4351,
      "step": 11354
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.624942660331726,
      "learning_rate": 1.4896161095980008e-05,
      "loss": 0.5066,
      "step": 11355
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.674769401550293,
      "learning_rate": 1.4895273079681949e-05,
      "loss": 0.4756,
      "step": 11356
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7035937309265137,
      "learning_rate": 1.4894385012612077e-05,
      "loss": 0.5783,
      "step": 11357
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.618589997291565,
      "learning_rate": 1.4893496894779606e-05,
      "loss": 0.4477,
      "step": 11358
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6807687282562256,
      "learning_rate": 1.4892608726193749e-05,
      "loss": 0.5506,
      "step": 11359
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5417119264602661,
      "learning_rate": 1.4891720506863714e-05,
      "loss": 0.4559,
      "step": 11360
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5743951797485352,
      "learning_rate": 1.4890832236798715e-05,
      "loss": 0.5103,
      "step": 11361
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5301408767700195,
      "learning_rate": 1.4889943916007967e-05,
      "loss": 0.4468,
      "step": 11362
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.084613084793091,
      "learning_rate": 1.4889055544500676e-05,
      "loss": 0.9927,
      "step": 11363
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.530859351158142,
      "learning_rate": 1.4888167122286068e-05,
      "loss": 1.05,
      "step": 11364
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7363966703414917,
      "learning_rate": 1.4887278649373347e-05,
      "loss": 0.5274,
      "step": 11365
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6669840812683105,
      "learning_rate": 1.4886390125771733e-05,
      "loss": 0.4752,
      "step": 11366
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5934377908706665,
      "learning_rate": 1.4885501551490436e-05,
      "loss": 0.491,
      "step": 11367
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5023292303085327,
      "learning_rate": 1.4884612926538678e-05,
      "loss": 0.4452,
      "step": 11368
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5821332931518555,
      "learning_rate": 1.488372425092567e-05,
      "loss": 0.5509,
      "step": 11369
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6128677129745483,
      "learning_rate": 1.4882835524660637e-05,
      "loss": 0.4719,
      "step": 11370
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9657927751541138,
      "learning_rate": 1.488194674775279e-05,
      "loss": 0.5424,
      "step": 11371
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6307134628295898,
      "learning_rate": 1.4881057920211345e-05,
      "loss": 0.458,
      "step": 11372
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6358003616333008,
      "learning_rate": 1.4880169042045527e-05,
      "loss": 0.5225,
      "step": 11373
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5791646242141724,
      "learning_rate": 1.4879280113264553e-05,
      "loss": 0.432,
      "step": 11374
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.624942660331726,
      "learning_rate": 1.4878391133877638e-05,
      "loss": 0.5587,
      "step": 11375
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5385693311691284,
      "learning_rate": 1.4877502103894012e-05,
      "loss": 0.4416,
      "step": 11376
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6082960367202759,
      "learning_rate": 1.4876613023322886e-05,
      "loss": 0.5724,
      "step": 11377
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6093287467956543,
      "learning_rate": 1.4875723892173486e-05,
      "loss": 0.4465,
      "step": 11378
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.605672001838684,
      "learning_rate": 1.4874834710455032e-05,
      "loss": 0.529,
      "step": 11379
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.668151617050171,
      "learning_rate": 1.4873945478176748e-05,
      "loss": 0.4682,
      "step": 11380
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.814018964767456,
      "learning_rate": 1.4873056195347852e-05,
      "loss": 0.564,
      "step": 11381
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5369279384613037,
      "learning_rate": 1.4872166861977575e-05,
      "loss": 0.443,
      "step": 11382
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5465881824493408,
      "learning_rate": 1.4871277478075134e-05,
      "loss": 0.4793,
      "step": 11383
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6064485311508179,
      "learning_rate": 1.4870388043649757e-05,
      "loss": 0.4381,
      "step": 11384
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6618587970733643,
      "learning_rate": 1.4869498558710667e-05,
      "loss": 0.4876,
      "step": 11385
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6127150058746338,
      "learning_rate": 1.486860902326709e-05,
      "loss": 0.443,
      "step": 11386
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5883337259292603,
      "learning_rate": 1.4867719437328254e-05,
      "loss": 0.5246,
      "step": 11387
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5642513036727905,
      "learning_rate": 1.4866829800903385e-05,
      "loss": 0.4424,
      "step": 11388
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6719547510147095,
      "learning_rate": 1.4865940114001704e-05,
      "loss": 0.5532,
      "step": 11389
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5565004348754883,
      "learning_rate": 1.4865050376632445e-05,
      "loss": 0.4558,
      "step": 11390
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5877503156661987,
      "learning_rate": 1.4864160588804834e-05,
      "loss": 0.5042,
      "step": 11391
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.51530921459198,
      "learning_rate": 1.4863270750528102e-05,
      "loss": 0.4446,
      "step": 11392
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5887913703918457,
      "learning_rate": 1.4862380861811472e-05,
      "loss": 0.5666,
      "step": 11393
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6269848346710205,
      "learning_rate": 1.486149092266418e-05,
      "loss": 0.4778,
      "step": 11394
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4888509511947632,
      "learning_rate": 1.486060093309545e-05,
      "loss": 0.4993,
      "step": 11395
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6049941778182983,
      "learning_rate": 1.4859710893114518e-05,
      "loss": 0.4427,
      "step": 11396
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6467033624649048,
      "learning_rate": 1.485882080273061e-05,
      "loss": 0.5743,
      "step": 11397
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.553335428237915,
      "learning_rate": 1.4857930661952966e-05,
      "loss": 0.4459,
      "step": 11398
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5519505739212036,
      "learning_rate": 1.4857040470790807e-05,
      "loss": 0.4764,
      "step": 11399
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5840762853622437,
      "learning_rate": 1.4856150229253377e-05,
      "loss": 0.4452,
      "step": 11400
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.579150915145874,
      "learning_rate": 1.48552599373499e-05,
      "loss": 0.493,
      "step": 11401
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6896740198135376,
      "learning_rate": 1.4854369595089612e-05,
      "loss": 0.497,
      "step": 11402
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5805598497390747,
      "learning_rate": 1.4853479202481752e-05,
      "loss": 0.4841,
      "step": 11403
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.519152283668518,
      "learning_rate": 1.4852588759535551e-05,
      "loss": 0.464,
      "step": 11404
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.580722689628601,
      "learning_rate": 1.4851698266260247e-05,
      "loss": 0.5087,
      "step": 11405
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5357438325881958,
      "learning_rate": 1.485080772266507e-05,
      "loss": 0.4616,
      "step": 11406
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5977674722671509,
      "learning_rate": 1.4849917128759264e-05,
      "loss": 0.5079,
      "step": 11407
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6047641038894653,
      "learning_rate": 1.4849026484552058e-05,
      "loss": 0.4357,
      "step": 11408
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.648144245147705,
      "learning_rate": 1.4848135790052694e-05,
      "loss": 0.5208,
      "step": 11409
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5445712804794312,
      "learning_rate": 1.4847245045270412e-05,
      "loss": 0.4292,
      "step": 11410
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4873067140579224,
      "learning_rate": 1.4846354250214444e-05,
      "loss": 0.485,
      "step": 11411
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6288338899612427,
      "learning_rate": 1.4845463404894033e-05,
      "loss": 0.4625,
      "step": 11412
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6107150316238403,
      "learning_rate": 1.4844572509318423e-05,
      "loss": 0.5111,
      "step": 11413
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5683567523956299,
      "learning_rate": 1.4843681563496845e-05,
      "loss": 0.4431,
      "step": 11414
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8225958347320557,
      "learning_rate": 1.4842790567438546e-05,
      "loss": 1.0268,
      "step": 11415
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7308369874954224,
      "learning_rate": 1.4841899521152762e-05,
      "loss": 1.0911,
      "step": 11416
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.670342206954956,
      "learning_rate": 1.4841008424648737e-05,
      "loss": 0.4806,
      "step": 11417
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.802398681640625,
      "learning_rate": 1.484011727793572e-05,
      "loss": 0.4593,
      "step": 11418
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6637077331542969,
      "learning_rate": 1.483922608102294e-05,
      "loss": 0.5078,
      "step": 11419
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5447101593017578,
      "learning_rate": 1.483833483391965e-05,
      "loss": 0.4411,
      "step": 11420
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5252623558044434,
      "learning_rate": 1.483744353663509e-05,
      "loss": 0.4685,
      "step": 11421
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7325334548950195,
      "learning_rate": 1.4836552189178505e-05,
      "loss": 0.4435,
      "step": 11422
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.648276925086975,
      "learning_rate": 1.4835660791559142e-05,
      "loss": 0.5137,
      "step": 11423
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7112054824829102,
      "learning_rate": 1.4834769343786242e-05,
      "loss": 0.4854,
      "step": 11424
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6863468885421753,
      "learning_rate": 1.4833877845869054e-05,
      "loss": 0.581,
      "step": 11425
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.586893081665039,
      "learning_rate": 1.4832986297816823e-05,
      "loss": 0.4561,
      "step": 11426
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0435690879821777,
      "learning_rate": 1.4832094699638793e-05,
      "loss": 1.0444,
      "step": 11427
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.709529161453247,
      "learning_rate": 1.4831203051344218e-05,
      "loss": 1.0711,
      "step": 11428
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.997952938079834,
      "learning_rate": 1.4830311352942337e-05,
      "loss": 0.99,
      "step": 11429
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.571820616722107,
      "learning_rate": 1.4829419604442406e-05,
      "loss": 1.0347,
      "step": 11430
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0413925647735596,
      "learning_rate": 1.4828527805853672e-05,
      "loss": 0.4914,
      "step": 11431
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9340320825576782,
      "learning_rate": 1.4827635957185382e-05,
      "loss": 0.4766,
      "step": 11432
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8603746891021729,
      "learning_rate": 1.482674405844679e-05,
      "loss": 0.4779,
      "step": 11433
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6018998622894287,
      "learning_rate": 1.4825852109647141e-05,
      "loss": 0.4312,
      "step": 11434
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5654715299606323,
      "learning_rate": 1.4824960110795687e-05,
      "loss": 0.4936,
      "step": 11435
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7042125463485718,
      "learning_rate": 1.4824068061901686e-05,
      "loss": 0.4474,
      "step": 11436
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9522197246551514,
      "learning_rate": 1.482317596297438e-05,
      "loss": 0.5153,
      "step": 11437
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.855991005897522,
      "learning_rate": 1.482228381402303e-05,
      "loss": 0.4764,
      "step": 11438
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6322766542434692,
      "learning_rate": 1.4821391615056887e-05,
      "loss": 0.4914,
      "step": 11439
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6022908687591553,
      "learning_rate": 1.4820499366085199e-05,
      "loss": 0.4521,
      "step": 11440
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7419331073760986,
      "learning_rate": 1.4819607067117228e-05,
      "loss": 0.514,
      "step": 11441
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.734043836593628,
      "learning_rate": 1.4818714718162222e-05,
      "loss": 0.4359,
      "step": 11442
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.81996488571167,
      "learning_rate": 1.4817822319229438e-05,
      "loss": 0.516,
      "step": 11443
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6505184173583984,
      "learning_rate": 1.4816929870328137e-05,
      "loss": 0.4537,
      "step": 11444
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5732004642486572,
      "learning_rate": 1.4816037371467566e-05,
      "loss": 0.4638,
      "step": 11445
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6141499280929565,
      "learning_rate": 1.4815144822656991e-05,
      "loss": 0.4878,
      "step": 11446
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7589107751846313,
      "learning_rate": 1.4814252223905658e-05,
      "loss": 0.536,
      "step": 11447
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6031866073608398,
      "learning_rate": 1.4813359575222835e-05,
      "loss": 0.4697,
      "step": 11448
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6082141399383545,
      "learning_rate": 1.4812466876617776e-05,
      "loss": 0.4796,
      "step": 11449
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5947046279907227,
      "learning_rate": 1.481157412809974e-05,
      "loss": 0.4388,
      "step": 11450
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.3930246829986572,
      "learning_rate": 1.4810681329677988e-05,
      "loss": 0.9784,
      "step": 11451
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.160156726837158,
      "learning_rate": 1.4809788481361775e-05,
      "loss": 1.0488,
      "step": 11452
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7358202934265137,
      "learning_rate": 1.4808895583160362e-05,
      "loss": 0.5031,
      "step": 11453
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6869354248046875,
      "learning_rate": 1.4808002635083018e-05,
      "loss": 0.4747,
      "step": 11454
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6596227884292603,
      "learning_rate": 1.4807109637138995e-05,
      "loss": 0.5704,
      "step": 11455
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5957940816879272,
      "learning_rate": 1.4806216589337558e-05,
      "loss": 0.4337,
      "step": 11456
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6560165882110596,
      "learning_rate": 1.4805323491687969e-05,
      "loss": 0.5298,
      "step": 11457
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.681749939918518,
      "learning_rate": 1.4804430344199494e-05,
      "loss": 0.4576,
      "step": 11458
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6942354440689087,
      "learning_rate": 1.480353714688139e-05,
      "loss": 0.4925,
      "step": 11459
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6343287229537964,
      "learning_rate": 1.4802643899742927e-05,
      "loss": 0.4549,
      "step": 11460
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5989770889282227,
      "learning_rate": 1.4801750602793366e-05,
      "loss": 0.4947,
      "step": 11461
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.519385814666748,
      "learning_rate": 1.4800857256041976e-05,
      "loss": 0.4338,
      "step": 11462
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6897932291030884,
      "learning_rate": 1.4799963859498014e-05,
      "loss": 0.5615,
      "step": 11463
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6375608444213867,
      "learning_rate": 1.4799070413170756e-05,
      "loss": 0.459,
      "step": 11464
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6780251264572144,
      "learning_rate": 1.479817691706946e-05,
      "loss": 0.5044,
      "step": 11465
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5459985733032227,
      "learning_rate": 1.4797283371203397e-05,
      "loss": 0.4481,
      "step": 11466
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5745619535446167,
      "learning_rate": 1.4796389775581836e-05,
      "loss": 0.5134,
      "step": 11467
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.510978102684021,
      "learning_rate": 1.4795496130214045e-05,
      "loss": 0.4596,
      "step": 11468
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6993567943572998,
      "learning_rate": 1.4794602435109286e-05,
      "loss": 0.5393,
      "step": 11469
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6163901090621948,
      "learning_rate": 1.4793708690276836e-05,
      "loss": 0.4541,
      "step": 11470
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6446467638015747,
      "learning_rate": 1.4792814895725958e-05,
      "loss": 0.5058,
      "step": 11471
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.56143057346344,
      "learning_rate": 1.4791921051465933e-05,
      "loss": 0.4148,
      "step": 11472
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6104751825332642,
      "learning_rate": 1.4791027157506015e-05,
      "loss": 0.5069,
      "step": 11473
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6183472871780396,
      "learning_rate": 1.4790133213855488e-05,
      "loss": 0.4732,
      "step": 11474
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6855559349060059,
      "learning_rate": 1.478923922052362e-05,
      "loss": 0.557,
      "step": 11475
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5468343496322632,
      "learning_rate": 1.4788345177519683e-05,
      "loss": 0.4363,
      "step": 11476
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.671079397201538,
      "learning_rate": 1.4787451084852951e-05,
      "loss": 0.5136,
      "step": 11477
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6034680604934692,
      "learning_rate": 1.4786556942532693e-05,
      "loss": 0.4451,
      "step": 11478
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5905202627182007,
      "learning_rate": 1.4785662750568184e-05,
      "loss": 0.4943,
      "step": 11479
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5746113061904907,
      "learning_rate": 1.4784768508968705e-05,
      "loss": 0.4493,
      "step": 11480
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6466294527053833,
      "learning_rate": 1.478387421774352e-05,
      "loss": 0.5069,
      "step": 11481
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6785176992416382,
      "learning_rate": 1.4782979876901916e-05,
      "loss": 0.4508,
      "step": 11482
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5982029438018799,
      "learning_rate": 1.4782085486453155e-05,
      "loss": 0.4688,
      "step": 11483
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.511904239654541,
      "learning_rate": 1.4781191046406525e-05,
      "loss": 0.4234,
      "step": 11484
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8775783777236938,
      "learning_rate": 1.4780296556771296e-05,
      "loss": 0.5682,
      "step": 11485
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5215773582458496,
      "learning_rate": 1.477940201755675e-05,
      "loss": 0.4313,
      "step": 11486
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5652672052383423,
      "learning_rate": 1.4778507428772159e-05,
      "loss": 0.4539,
      "step": 11487
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6823627948760986,
      "learning_rate": 1.4777612790426808e-05,
      "loss": 0.5279,
      "step": 11488
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.588721513748169,
      "learning_rate": 1.477671810252997e-05,
      "loss": 0.5217,
      "step": 11489
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6025508642196655,
      "learning_rate": 1.477582336509093e-05,
      "loss": 0.4582,
      "step": 11490
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.355090856552124,
      "learning_rate": 1.4774928578118961e-05,
      "loss": 1.0787,
      "step": 11491
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8603005409240723,
      "learning_rate": 1.477403374162335e-05,
      "loss": 0.9734,
      "step": 11492
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6802505254745483,
      "learning_rate": 1.4773138855613376e-05,
      "loss": 0.5053,
      "step": 11493
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8338792324066162,
      "learning_rate": 1.4772243920098318e-05,
      "loss": 0.4905,
      "step": 11494
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.790251612663269,
      "learning_rate": 1.477134893508746e-05,
      "loss": 0.5533,
      "step": 11495
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5469976663589478,
      "learning_rate": 1.4770453900590083e-05,
      "loss": 0.4549,
      "step": 11496
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.550437331199646,
      "learning_rate": 1.4769558816615467e-05,
      "loss": 0.5082,
      "step": 11497
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6471457481384277,
      "learning_rate": 1.4768663683172906e-05,
      "loss": 0.4834,
      "step": 11498
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7492804527282715,
      "learning_rate": 1.476776850027167e-05,
      "loss": 0.5143,
      "step": 11499
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5999099016189575,
      "learning_rate": 1.4766873267921057e-05,
      "loss": 0.4418,
      "step": 11500
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6795494556427002,
      "learning_rate": 1.4765977986130344e-05,
      "loss": 0.5803,
      "step": 11501
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5288512706756592,
      "learning_rate": 1.4765082654908815e-05,
      "loss": 0.4372,
      "step": 11502
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.639639973640442,
      "learning_rate": 1.4764187274265763e-05,
      "loss": 0.5336,
      "step": 11503
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5280457735061646,
      "learning_rate": 1.4763291844210468e-05,
      "loss": 0.4238,
      "step": 11504
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6320239305496216,
      "learning_rate": 1.4762396364752221e-05,
      "loss": 0.5501,
      "step": 11505
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.633113145828247,
      "learning_rate": 1.476150083590031e-05,
      "loss": 0.4457,
      "step": 11506
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5978059768676758,
      "learning_rate": 1.4760605257664015e-05,
      "loss": 0.4393,
      "step": 11507
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6265358924865723,
      "learning_rate": 1.4759709630052639e-05,
      "loss": 0.4929,
      "step": 11508
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5566822290420532,
      "learning_rate": 1.4758813953075457e-05,
      "loss": 0.4745,
      "step": 11509
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6794381141662598,
      "learning_rate": 1.4757918226741768e-05,
      "loss": 0.4936,
      "step": 11510
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6303569078445435,
      "learning_rate": 1.4757022451060856e-05,
      "loss": 0.4786,
      "step": 11511
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6072083711624146,
      "learning_rate": 1.4756126626042016e-05,
      "loss": 0.433,
      "step": 11512
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5944647789001465,
      "learning_rate": 1.4755230751694538e-05,
      "loss": 0.5189,
      "step": 11513
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5526758432388306,
      "learning_rate": 1.4754334828027712e-05,
      "loss": 0.477,
      "step": 11514
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7040765285491943,
      "learning_rate": 1.475343885505083e-05,
      "loss": 0.5063,
      "step": 11515
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6600855588912964,
      "learning_rate": 1.475254283277319e-05,
      "loss": 0.4416,
      "step": 11516
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6551538705825806,
      "learning_rate": 1.4751646761204077e-05,
      "loss": 0.5,
      "step": 11517
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6166402101516724,
      "learning_rate": 1.475075064035279e-05,
      "loss": 0.4358,
      "step": 11518
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5543574094772339,
      "learning_rate": 1.4749854470228621e-05,
      "loss": 0.4953,
      "step": 11519
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6205389499664307,
      "learning_rate": 1.4748958250840868e-05,
      "loss": 0.4654,
      "step": 11520
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5209360122680664,
      "learning_rate": 1.4748061982198825e-05,
      "loss": 0.471,
      "step": 11521
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5831772089004517,
      "learning_rate": 1.4747165664311785e-05,
      "loss": 0.4763,
      "step": 11522
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5317083597183228,
      "learning_rate": 1.4746269297189045e-05,
      "loss": 0.5187,
      "step": 11523
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6131117343902588,
      "learning_rate": 1.4745372880839904e-05,
      "loss": 0.4585,
      "step": 11524
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5470467805862427,
      "learning_rate": 1.4744476415273656e-05,
      "loss": 0.5004,
      "step": 11525
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.597668170928955,
      "learning_rate": 1.4743579900499603e-05,
      "loss": 0.4555,
      "step": 11526
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.665257453918457,
      "learning_rate": 1.474268333652704e-05,
      "loss": 0.5684,
      "step": 11527
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5977168083190918,
      "learning_rate": 1.4741786723365266e-05,
      "loss": 0.5126,
      "step": 11528
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7864755392074585,
      "learning_rate": 1.4740890061023583e-05,
      "loss": 0.5056,
      "step": 11529
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5282645225524902,
      "learning_rate": 1.4739993349511288e-05,
      "loss": 0.4344,
      "step": 11530
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6879055500030518,
      "learning_rate": 1.4739096588837684e-05,
      "loss": 0.4867,
      "step": 11531
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6273138523101807,
      "learning_rate": 1.4738199779012068e-05,
      "loss": 0.4713,
      "step": 11532
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.601450800895691,
      "learning_rate": 1.4737302920043744e-05,
      "loss": 0.4719,
      "step": 11533
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4802970886230469,
      "learning_rate": 1.4736406011942014e-05,
      "loss": 0.4334,
      "step": 11534
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5804780721664429,
      "learning_rate": 1.473550905471618e-05,
      "loss": 0.5219,
      "step": 11535
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5337679386138916,
      "learning_rate": 1.4734612048375543e-05,
      "loss": 0.4368,
      "step": 11536
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6486616134643555,
      "learning_rate": 1.4733714992929408e-05,
      "loss": 0.5371,
      "step": 11537
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4875495433807373,
      "learning_rate": 1.4732817888387083e-05,
      "loss": 0.4731,
      "step": 11538
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.587148666381836,
      "learning_rate": 1.4731920734757864e-05,
      "loss": 0.4783,
      "step": 11539
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5640312433242798,
      "learning_rate": 1.4731023532051064e-05,
      "loss": 0.4499,
      "step": 11540
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.587886095046997,
      "learning_rate": 1.473012628027598e-05,
      "loss": 0.535,
      "step": 11541
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.512739896774292,
      "learning_rate": 1.4729228979441925e-05,
      "loss": 0.4456,
      "step": 11542
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8958059549331665,
      "learning_rate": 1.4728331629558201e-05,
      "loss": 0.5981,
      "step": 11543
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5142083168029785,
      "learning_rate": 1.4727434230634124e-05,
      "loss": 0.4316,
      "step": 11544
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.875876545906067,
      "learning_rate": 1.4726536782678986e-05,
      "loss": 1.0399,
      "step": 11545
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7013665437698364,
      "learning_rate": 1.4725639285702107e-05,
      "loss": 1.0563,
      "step": 11546
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7052887678146362,
      "learning_rate": 1.4724741739712794e-05,
      "loss": 0.5188,
      "step": 11547
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7132102251052856,
      "learning_rate": 1.4723844144720351e-05,
      "loss": 0.4862,
      "step": 11548
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.667039155960083,
      "learning_rate": 1.4722946500734093e-05,
      "loss": 0.5173,
      "step": 11549
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5916579961776733,
      "learning_rate": 1.4722048807763324e-05,
      "loss": 0.4851,
      "step": 11550
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.564415693283081,
      "learning_rate": 1.4721151065817358e-05,
      "loss": 0.4736,
      "step": 11551
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.697852373123169,
      "learning_rate": 1.4720253274905509e-05,
      "loss": 0.4703,
      "step": 11552
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8052483797073364,
      "learning_rate": 1.4719355435037084e-05,
      "loss": 0.5092,
      "step": 11553
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6282482147216797,
      "learning_rate": 1.4718457546221395e-05,
      "loss": 0.4686,
      "step": 11554
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6187753677368164,
      "learning_rate": 1.471755960846776e-05,
      "loss": 0.5353,
      "step": 11555
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5577380657196045,
      "learning_rate": 1.4716661621785485e-05,
      "loss": 0.4246,
      "step": 11556
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.561452031135559,
      "learning_rate": 1.4715763586183887e-05,
      "loss": 0.4786,
      "step": 11557
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.638677954673767,
      "learning_rate": 1.471486550167228e-05,
      "loss": 0.4346,
      "step": 11558
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.621045470237732,
      "learning_rate": 1.4713967368259981e-05,
      "loss": 0.4709,
      "step": 11559
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5218415260314941,
      "learning_rate": 1.4713069185956297e-05,
      "loss": 0.4334,
      "step": 11560
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6602777242660522,
      "learning_rate": 1.471217095477055e-05,
      "loss": 0.5617,
      "step": 11561
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5841293334960938,
      "learning_rate": 1.4711272674712059e-05,
      "loss": 0.4375,
      "step": 11562
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5787626504898071,
      "learning_rate": 1.4710374345790133e-05,
      "loss": 0.4954,
      "step": 11563
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5699191093444824,
      "learning_rate": 1.4709475968014096e-05,
      "loss": 0.4549,
      "step": 11564
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5713896751403809,
      "learning_rate": 1.4708577541393259e-05,
      "loss": 0.4722,
      "step": 11565
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4843896627426147,
      "learning_rate": 1.4707679065936943e-05,
      "loss": 0.4138,
      "step": 11566
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6734626293182373,
      "learning_rate": 1.4706780541654473e-05,
      "loss": 0.5097,
      "step": 11567
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5684186220169067,
      "learning_rate": 1.4705881968555155e-05,
      "loss": 0.5479,
      "step": 11568
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7019107341766357,
      "learning_rate": 1.4704983346648321e-05,
      "loss": 0.5178,
      "step": 11569
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5076849460601807,
      "learning_rate": 1.4704084675943285e-05,
      "loss": 0.4301,
      "step": 11570
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5096882581710815,
      "learning_rate": 1.470318595644937e-05,
      "loss": 0.4652,
      "step": 11571
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5422629117965698,
      "learning_rate": 1.4702287188175892e-05,
      "loss": 0.4621,
      "step": 11572
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.66697359085083,
      "learning_rate": 1.4701388371132178e-05,
      "loss": 0.5182,
      "step": 11573
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5213918685913086,
      "learning_rate": 1.4700489505327546e-05,
      "loss": 0.4376,
      "step": 11574
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6276342868804932,
      "learning_rate": 1.4699590590771328e-05,
      "loss": 0.5097,
      "step": 11575
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.583690881729126,
      "learning_rate": 1.4698691627472835e-05,
      "loss": 0.4603,
      "step": 11576
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6783766746520996,
      "learning_rate": 1.46977926154414e-05,
      "loss": 0.474,
      "step": 11577
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5338894128799438,
      "learning_rate": 1.469689355468634e-05,
      "loss": 0.4749,
      "step": 11578
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5236176252365112,
      "learning_rate": 1.4695994445216985e-05,
      "loss": 0.5117,
      "step": 11579
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5491076707839966,
      "learning_rate": 1.469509528704266e-05,
      "loss": 0.4696,
      "step": 11580
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.842036247253418,
      "learning_rate": 1.4694196080172686e-05,
      "loss": 0.5273,
      "step": 11581
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6054021120071411,
      "learning_rate": 1.4693296824616394e-05,
      "loss": 0.4303,
      "step": 11582
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6253705024719238,
      "learning_rate": 1.4692397520383108e-05,
      "loss": 0.5078,
      "step": 11583
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5151340961456299,
      "learning_rate": 1.4691498167482157e-05,
      "loss": 0.4528,
      "step": 11584
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0076005458831787,
      "learning_rate": 1.469059876592287e-05,
      "loss": 1.024,
      "step": 11585
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.712211012840271,
      "learning_rate": 1.4689699315714567e-05,
      "loss": 1.0591,
      "step": 11586
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6965614557266235,
      "learning_rate": 1.4688799816866588e-05,
      "loss": 0.5228,
      "step": 11587
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.743345022201538,
      "learning_rate": 1.4687900269388255e-05,
      "loss": 0.4441,
      "step": 11588
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7659039497375488,
      "learning_rate": 1.46870006732889e-05,
      "loss": 0.5423,
      "step": 11589
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5342522859573364,
      "learning_rate": 1.4686101028577854e-05,
      "loss": 0.421,
      "step": 11590
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7028554677963257,
      "learning_rate": 1.4685201335264447e-05,
      "loss": 0.5466,
      "step": 11591
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5625137090682983,
      "learning_rate": 1.4684301593358008e-05,
      "loss": 0.411,
      "step": 11592
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6896499395370483,
      "learning_rate": 1.4683401802867875e-05,
      "loss": 0.4993,
      "step": 11593
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6405014991760254,
      "learning_rate": 1.4682501963803373e-05,
      "loss": 0.4746,
      "step": 11594
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5921010971069336,
      "learning_rate": 1.4681602076173838e-05,
      "loss": 0.5199,
      "step": 11595
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5715947151184082,
      "learning_rate": 1.4680702139988605e-05,
      "loss": 0.4823,
      "step": 11596
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5124002695083618,
      "learning_rate": 1.4679802155257004e-05,
      "loss": 0.4716,
      "step": 11597
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5456880331039429,
      "learning_rate": 1.4678902121988374e-05,
      "loss": 0.4774,
      "step": 11598
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5380560159683228,
      "learning_rate": 1.4678002040192043e-05,
      "loss": 0.4776,
      "step": 11599
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.622328519821167,
      "learning_rate": 1.4677101909877354e-05,
      "loss": 0.4655,
      "step": 11600
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5668535232543945,
      "learning_rate": 1.4676201731053638e-05,
      "loss": 0.4838,
      "step": 11601
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6323797702789307,
      "learning_rate": 1.467530150373023e-05,
      "loss": 0.4385,
      "step": 11602
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5870568752288818,
      "learning_rate": 1.4674401227916475e-05,
      "loss": 0.4543,
      "step": 11603
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6146060228347778,
      "learning_rate": 1.4673500903621698e-05,
      "loss": 0.4881,
      "step": 11604
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5584063529968262,
      "learning_rate": 1.4672600530855246e-05,
      "loss": 0.4644,
      "step": 11605
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5998749732971191,
      "learning_rate": 1.4671700109626454e-05,
      "loss": 0.4523,
      "step": 11606
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.669456124305725,
      "learning_rate": 1.4670799639944665e-05,
      "loss": 0.5087,
      "step": 11607
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5924373865127563,
      "learning_rate": 1.4669899121819212e-05,
      "loss": 0.4549,
      "step": 11608
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6080526113510132,
      "learning_rate": 1.4668998555259437e-05,
      "loss": 0.5682,
      "step": 11609
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5154683589935303,
      "learning_rate": 1.466809794027468e-05,
      "loss": 0.4179,
      "step": 11610
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7523657083511353,
      "learning_rate": 1.4667197276874285e-05,
      "loss": 0.4938,
      "step": 11611
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6764657497406006,
      "learning_rate": 1.466629656506759e-05,
      "loss": 0.4345,
      "step": 11612
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.615587830543518,
      "learning_rate": 1.4665395804863937e-05,
      "loss": 0.4814,
      "step": 11613
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5948258638381958,
      "learning_rate": 1.4664494996272672e-05,
      "loss": 0.4908,
      "step": 11614
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6083121299743652,
      "learning_rate": 1.4663594139303132e-05,
      "loss": 0.4968,
      "step": 11615
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6634252071380615,
      "learning_rate": 1.4662693233964664e-05,
      "loss": 0.466,
      "step": 11616
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.671678066253662,
      "learning_rate": 1.4661792280266611e-05,
      "loss": 0.5525,
      "step": 11617
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5793683528900146,
      "learning_rate": 1.4660891278218318e-05,
      "loss": 0.4325,
      "step": 11618
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7410792112350464,
      "learning_rate": 1.4659990227829131e-05,
      "loss": 0.5376,
      "step": 11619
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.634775996208191,
      "learning_rate": 1.465908912910839e-05,
      "loss": 0.4869,
      "step": 11620
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6108530759811401,
      "learning_rate": 1.4658187982065448e-05,
      "loss": 0.5022,
      "step": 11621
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5489305257797241,
      "learning_rate": 1.4657286786709645e-05,
      "loss": 0.4418,
      "step": 11622
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6660776138305664,
      "learning_rate": 1.4656385543050334e-05,
      "loss": 0.5694,
      "step": 11623
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5901916027069092,
      "learning_rate": 1.4655484251096854e-05,
      "loss": 0.4432,
      "step": 11624
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.542606234550476,
      "learning_rate": 1.4654582910858563e-05,
      "loss": 0.4574,
      "step": 11625
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6715738773345947,
      "learning_rate": 1.46536815223448e-05,
      "loss": 0.4809,
      "step": 11626
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.550520896911621,
      "learning_rate": 1.4652780085564922e-05,
      "loss": 0.4826,
      "step": 11627
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6527804136276245,
      "learning_rate": 1.4651878600528271e-05,
      "loss": 0.4638,
      "step": 11628
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6630518436431885,
      "learning_rate": 1.4650977067244204e-05,
      "loss": 0.5177,
      "step": 11629
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.57411527633667,
      "learning_rate": 1.4650075485722062e-05,
      "loss": 0.4711,
      "step": 11630
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6360676288604736,
      "learning_rate": 1.4649173855971205e-05,
      "loss": 0.4873,
      "step": 11631
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5837236642837524,
      "learning_rate": 1.4648272178000979e-05,
      "loss": 0.4541,
      "step": 11632
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6203560829162598,
      "learning_rate": 1.464737045182074e-05,
      "loss": 0.4583,
      "step": 11633
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.629172921180725,
      "learning_rate": 1.4646468677439836e-05,
      "loss": 0.4464,
      "step": 11634
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5760868787765503,
      "learning_rate": 1.464556685486762e-05,
      "loss": 0.5105,
      "step": 11635
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5638983249664307,
      "learning_rate": 1.464466498411345e-05,
      "loss": 0.4474,
      "step": 11636
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5821419954299927,
      "learning_rate": 1.4643763065186677e-05,
      "loss": 0.4535,
      "step": 11637
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7843648195266724,
      "learning_rate": 1.4642861098096653e-05,
      "loss": 0.4805,
      "step": 11638
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.559914231300354,
      "learning_rate": 1.4641959082852739e-05,
      "loss": 0.4917,
      "step": 11639
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5929573774337769,
      "learning_rate": 1.4641057019464282e-05,
      "loss": 0.458,
      "step": 11640
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.591052770614624,
      "learning_rate": 1.4640154907940645e-05,
      "loss": 0.5414,
      "step": 11641
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.562421441078186,
      "learning_rate": 1.463925274829118e-05,
      "loss": 0.4491,
      "step": 11642
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.511381983757019,
      "learning_rate": 1.4638350540525248e-05,
      "loss": 0.4802,
      "step": 11643
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5935956239700317,
      "learning_rate": 1.4637448284652201e-05,
      "loss": 0.4529,
      "step": 11644
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9078497886657715,
      "learning_rate": 1.46365459806814e-05,
      "loss": 0.9987,
      "step": 11645
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7477400302886963,
      "learning_rate": 1.4635643628622202e-05,
      "loss": 1.0953,
      "step": 11646
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7182060480117798,
      "learning_rate": 1.463474122848397e-05,
      "loss": 0.4939,
      "step": 11647
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7684320211410522,
      "learning_rate": 1.4633838780276057e-05,
      "loss": 0.4596,
      "step": 11648
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6525061130523682,
      "learning_rate": 1.4632936284007828e-05,
      "loss": 0.5304,
      "step": 11649
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.606329083442688,
      "learning_rate": 1.4632033739688643e-05,
      "loss": 0.456,
      "step": 11650
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6449722051620483,
      "learning_rate": 1.463113114732786e-05,
      "loss": 0.5265,
      "step": 11651
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5916273593902588,
      "learning_rate": 1.4630228506934839e-05,
      "loss": 0.454,
      "step": 11652
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5460054874420166,
      "learning_rate": 1.4629325818518948e-05,
      "loss": 0.4875,
      "step": 11653
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6499323844909668,
      "learning_rate": 1.462842308208954e-05,
      "loss": 0.5068,
      "step": 11654
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8976373672485352,
      "learning_rate": 1.4627520297655988e-05,
      "loss": 1.0807,
      "step": 11655
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6807559728622437,
      "learning_rate": 1.4626617465227649e-05,
      "loss": 1.1084,
      "step": 11656
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7318181991577148,
      "learning_rate": 1.4625714584813891e-05,
      "loss": 0.5371,
      "step": 11657
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.586457371711731,
      "learning_rate": 1.4624811656424072e-05,
      "loss": 0.4451,
      "step": 11658
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6388797760009766,
      "learning_rate": 1.4623908680067562e-05,
      "loss": 0.5454,
      "step": 11659
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5704716444015503,
      "learning_rate": 1.4623005655753727e-05,
      "loss": 0.4606,
      "step": 11660
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5482054948806763,
      "learning_rate": 1.4622102583491928e-05,
      "loss": 0.4802,
      "step": 11661
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6917179822921753,
      "learning_rate": 1.4621199463291537e-05,
      "loss": 0.4884,
      "step": 11662
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6844737529754639,
      "learning_rate": 1.4620296295161916e-05,
      "loss": 0.5131,
      "step": 11663
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5780532360076904,
      "learning_rate": 1.4619393079112434e-05,
      "loss": 0.4477,
      "step": 11664
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5948351621627808,
      "learning_rate": 1.4618489815152459e-05,
      "loss": 0.531,
      "step": 11665
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6636539697647095,
      "learning_rate": 1.4617586503291358e-05,
      "loss": 0.4849,
      "step": 11666
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7510868310928345,
      "learning_rate": 1.4616683143538503e-05,
      "loss": 0.5039,
      "step": 11667
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5288190841674805,
      "learning_rate": 1.4615779735903264e-05,
      "loss": 0.4479,
      "step": 11668
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6303050518035889,
      "learning_rate": 1.4614876280395004e-05,
      "loss": 0.4966,
      "step": 11669
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4430549144744873,
      "learning_rate": 1.46139727770231e-05,
      "loss": 0.4181,
      "step": 11670
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5833566188812256,
      "learning_rate": 1.461306922579692e-05,
      "loss": 0.4911,
      "step": 11671
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6595039367675781,
      "learning_rate": 1.4612165626725832e-05,
      "loss": 0.4681,
      "step": 11672
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.608725905418396,
      "learning_rate": 1.4611261979819214e-05,
      "loss": 0.5351,
      "step": 11673
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5849668979644775,
      "learning_rate": 1.4610358285086435e-05,
      "loss": 0.4469,
      "step": 11674
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.685099720954895,
      "learning_rate": 1.460945454253687e-05,
      "loss": 0.5244,
      "step": 11675
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5922791957855225,
      "learning_rate": 1.4608550752179886e-05,
      "loss": 0.4296,
      "step": 11676
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.549014687538147,
      "learning_rate": 1.4607646914024864e-05,
      "loss": 0.5077,
      "step": 11677
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5834870338439941,
      "learning_rate": 1.4606743028081176e-05,
      "loss": 0.4338,
      "step": 11678
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8429312705993652,
      "learning_rate": 1.4605839094358196e-05,
      "loss": 0.5413,
      "step": 11679
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4907655715942383,
      "learning_rate": 1.4604935112865301e-05,
      "loss": 0.4406,
      "step": 11680
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7064484357833862,
      "learning_rate": 1.4604031083611863e-05,
      "loss": 0.5292,
      "step": 11681
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5199222564697266,
      "learning_rate": 1.460312700660726e-05,
      "loss": 0.409,
      "step": 11682
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7886953353881836,
      "learning_rate": 1.4602222881860873e-05,
      "loss": 0.5087,
      "step": 11683
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.697557806968689,
      "learning_rate": 1.4601318709382071e-05,
      "loss": 0.4294,
      "step": 11684
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6468627452850342,
      "learning_rate": 1.4600414489180236e-05,
      "loss": 0.5446,
      "step": 11685
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5497058629989624,
      "learning_rate": 1.4599510221264747e-05,
      "loss": 0.436,
      "step": 11686
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5445556640625,
      "learning_rate": 1.4598605905644984e-05,
      "loss": 0.5208,
      "step": 11687
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7334481477737427,
      "learning_rate": 1.4597701542330324e-05,
      "loss": 0.4994,
      "step": 11688
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5436760187149048,
      "learning_rate": 1.4596797131330145e-05,
      "loss": 0.5026,
      "step": 11689
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.691831350326538,
      "learning_rate": 1.459589267265383e-05,
      "loss": 0.4726,
      "step": 11690
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5420054197311401,
      "learning_rate": 1.4594988166310762e-05,
      "loss": 0.4758,
      "step": 11691
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.567314624786377,
      "learning_rate": 1.4594083612310312e-05,
      "loss": 0.4742,
      "step": 11692
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5701044797897339,
      "learning_rate": 1.4593179010661873e-05,
      "loss": 0.4875,
      "step": 11693
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5455361604690552,
      "learning_rate": 1.4592274361374823e-05,
      "loss": 0.4713,
      "step": 11694
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6754388809204102,
      "learning_rate": 1.4591369664458544e-05,
      "loss": 0.5346,
      "step": 11695
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5824111700057983,
      "learning_rate": 1.459046491992242e-05,
      "loss": 0.4353,
      "step": 11696
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6492903232574463,
      "learning_rate": 1.4589560127775832e-05,
      "loss": 0.4935,
      "step": 11697
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.739245891571045,
      "learning_rate": 1.458865528802817e-05,
      "loss": 0.4426,
      "step": 11698
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.587885856628418,
      "learning_rate": 1.4587750400688813e-05,
      "loss": 0.5133,
      "step": 11699
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5257046222686768,
      "learning_rate": 1.4586845465767145e-05,
      "loss": 0.428,
      "step": 11700
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5967439413070679,
      "learning_rate": 1.4585940483272562e-05,
      "loss": 0.4955,
      "step": 11701
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.562219262123108,
      "learning_rate": 1.4585035453214437e-05,
      "loss": 0.4306,
      "step": 11702
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6229411363601685,
      "learning_rate": 1.4584130375602166e-05,
      "loss": 0.5292,
      "step": 11703
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6378400325775146,
      "learning_rate": 1.458322525044513e-05,
      "loss": 0.4631,
      "step": 11704
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6773741245269775,
      "learning_rate": 1.4582320077752721e-05,
      "loss": 0.5183,
      "step": 11705
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.528940200805664,
      "learning_rate": 1.4581414857534323e-05,
      "loss": 0.4291,
      "step": 11706
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5435861349105835,
      "learning_rate": 1.458050958979933e-05,
      "loss": 0.4994,
      "step": 11707
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6873265504837036,
      "learning_rate": 1.4579604274557123e-05,
      "loss": 0.4561,
      "step": 11708
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7549874782562256,
      "learning_rate": 1.4578698911817105e-05,
      "loss": 0.5805,
      "step": 11709
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5869377851486206,
      "learning_rate": 1.4577793501588648e-05,
      "loss": 0.441,
      "step": 11710
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5382249355316162,
      "learning_rate": 1.4576888043881158e-05,
      "loss": 0.4923,
      "step": 11711
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8327564001083374,
      "learning_rate": 1.4575982538704018e-05,
      "loss": 0.497,
      "step": 11712
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5988069772720337,
      "learning_rate": 1.4575076986066622e-05,
      "loss": 0.4918,
      "step": 11713
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5547066926956177,
      "learning_rate": 1.4574171385978365e-05,
      "loss": 0.4329,
      "step": 11714
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5871323347091675,
      "learning_rate": 1.4573265738448632e-05,
      "loss": 0.5483,
      "step": 11715
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5383015871047974,
      "learning_rate": 1.4572360043486822e-05,
      "loss": 0.4075,
      "step": 11716
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5808697938919067,
      "learning_rate": 1.4571454301102329e-05,
      "loss": 0.4819,
      "step": 11717
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5666093826293945,
      "learning_rate": 1.4570548511304539e-05,
      "loss": 0.4436,
      "step": 11718
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6583439111709595,
      "learning_rate": 1.4569642674102857e-05,
      "loss": 0.4912,
      "step": 11719
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5513198375701904,
      "learning_rate": 1.4568736789506671e-05,
      "loss": 0.4359,
      "step": 11720
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6666208505630493,
      "learning_rate": 1.4567830857525378e-05,
      "loss": 0.5043,
      "step": 11721
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5649651288986206,
      "learning_rate": 1.4566924878168378e-05,
      "loss": 0.45,
      "step": 11722
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5429024696350098,
      "learning_rate": 1.4566018851445062e-05,
      "loss": 0.5661,
      "step": 11723
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4711519479751587,
      "learning_rate": 1.456511277736483e-05,
      "loss": 0.4256,
      "step": 11724
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.515035629272461,
      "learning_rate": 1.4564206655937078e-05,
      "loss": 0.4541,
      "step": 11725
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.62018620967865,
      "learning_rate": 1.4563300487171201e-05,
      "loss": 0.466,
      "step": 11726
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.559051513671875,
      "learning_rate": 1.4562394271076608e-05,
      "loss": 0.4958,
      "step": 11727
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6637248992919922,
      "learning_rate": 1.4561488007662685e-05,
      "loss": 0.4873,
      "step": 11728
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6260097026824951,
      "learning_rate": 1.456058169693884e-05,
      "loss": 0.5519,
      "step": 11729
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.647860050201416,
      "learning_rate": 1.455967533891447e-05,
      "loss": 0.4428,
      "step": 11730
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5522414445877075,
      "learning_rate": 1.4558768933598972e-05,
      "loss": 0.4613,
      "step": 11731
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5482288599014282,
      "learning_rate": 1.4557862481001755e-05,
      "loss": 0.4743,
      "step": 11732
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.524966835975647,
      "learning_rate": 1.4556955981132212e-05,
      "loss": 0.4387,
      "step": 11733
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6473276615142822,
      "learning_rate": 1.4556049433999753e-05,
      "loss": 0.4566,
      "step": 11734
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5847996473312378,
      "learning_rate": 1.4555142839613771e-05,
      "loss": 0.5242,
      "step": 11735
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6445109844207764,
      "learning_rate": 1.4554236197983676e-05,
      "loss": 0.5139,
      "step": 11736
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6143509149551392,
      "learning_rate": 1.455332950911887e-05,
      "loss": 0.4939,
      "step": 11737
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8865755796432495,
      "learning_rate": 1.4552422773028751e-05,
      "loss": 0.4721,
      "step": 11738
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.597338080406189,
      "learning_rate": 1.4551515989722732e-05,
      "loss": 0.4758,
      "step": 11739
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5230509042739868,
      "learning_rate": 1.4550609159210216e-05,
      "loss": 0.4459,
      "step": 11740
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.636534333229065,
      "learning_rate": 1.4549702281500603e-05,
      "loss": 0.4915,
      "step": 11741
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5648261308670044,
      "learning_rate": 1.4548795356603305e-05,
      "loss": 0.442,
      "step": 11742
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4835233688354492,
      "learning_rate": 1.4547888384527722e-05,
      "loss": 0.4737,
      "step": 11743
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5789000988006592,
      "learning_rate": 1.4546981365283263e-05,
      "loss": 0.4788,
      "step": 11744
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6238040924072266,
      "learning_rate": 1.454607429887934e-05,
      "loss": 0.4877,
      "step": 11745
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5088294744491577,
      "learning_rate": 1.4545167185325355e-05,
      "loss": 0.4755,
      "step": 11746
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6466913223266602,
      "learning_rate": 1.4544260024630718e-05,
      "loss": 0.4979,
      "step": 11747
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5641692876815796,
      "learning_rate": 1.454335281680484e-05,
      "loss": 0.4305,
      "step": 11748
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5889623165130615,
      "learning_rate": 1.4542445561857126e-05,
      "loss": 0.4783,
      "step": 11749
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.595598816871643,
      "learning_rate": 1.4541538259796987e-05,
      "loss": 0.4663,
      "step": 11750
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6955842971801758,
      "learning_rate": 1.4540630910633837e-05,
      "loss": 0.5073,
      "step": 11751
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6685861349105835,
      "learning_rate": 1.4539723514377082e-05,
      "loss": 0.4375,
      "step": 11752
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6126561164855957,
      "learning_rate": 1.4538816071036138e-05,
      "loss": 0.5131,
      "step": 11753
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5086736679077148,
      "learning_rate": 1.4537908580620406e-05,
      "loss": 0.4254,
      "step": 11754
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6948747634887695,
      "learning_rate": 1.4537001043139311e-05,
      "loss": 0.5,
      "step": 11755
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5094667673110962,
      "learning_rate": 1.4536093458602258e-05,
      "loss": 0.3992,
      "step": 11756
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5752909183502197,
      "learning_rate": 1.4535185827018663e-05,
      "loss": 0.5084,
      "step": 11757
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7092796564102173,
      "learning_rate": 1.453427814839794e-05,
      "loss": 0.4718,
      "step": 11758
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7170530557632446,
      "learning_rate": 1.4533370422749502e-05,
      "loss": 0.4948,
      "step": 11759
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.637392520904541,
      "learning_rate": 1.4532462650082762e-05,
      "loss": 0.4991,
      "step": 11760
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5403382778167725,
      "learning_rate": 1.4531554830407135e-05,
      "loss": 0.4817,
      "step": 11761
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5564799308776855,
      "learning_rate": 1.453064696373204e-05,
      "loss": 0.4617,
      "step": 11762
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5136771202087402,
      "learning_rate": 1.4529739050066894e-05,
      "loss": 0.4718,
      "step": 11763
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5112115144729614,
      "learning_rate": 1.4528831089421104e-05,
      "loss": 0.4261,
      "step": 11764
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5417802333831787,
      "learning_rate": 1.4527923081804099e-05,
      "loss": 0.5059,
      "step": 11765
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6166688203811646,
      "learning_rate": 1.4527015027225288e-05,
      "loss": 0.4783,
      "step": 11766
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.644561767578125,
      "learning_rate": 1.4526106925694095e-05,
      "loss": 0.544,
      "step": 11767
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5314561128616333,
      "learning_rate": 1.4525198777219935e-05,
      "loss": 0.4351,
      "step": 11768
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6141546964645386,
      "learning_rate": 1.4524290581812225e-05,
      "loss": 0.5353,
      "step": 11769
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6109086275100708,
      "learning_rate": 1.452338233948039e-05,
      "loss": 0.442,
      "step": 11770
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6997957229614258,
      "learning_rate": 1.4522474050233845e-05,
      "loss": 0.5602,
      "step": 11771
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6171008348464966,
      "learning_rate": 1.4521565714082013e-05,
      "loss": 0.4516,
      "step": 11772
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.576230525970459,
      "learning_rate": 1.4520657331034315e-05,
      "loss": 0.4775,
      "step": 11773
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.559912085533142,
      "learning_rate": 1.4519748901100169e-05,
      "loss": 0.4157,
      "step": 11774
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.57914137840271,
      "learning_rate": 1.4518840424289002e-05,
      "loss": 0.4792,
      "step": 11775
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6574037075042725,
      "learning_rate": 1.4517931900610234e-05,
      "loss": 0.4714,
      "step": 11776
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6195379495620728,
      "learning_rate": 1.4517023330073286e-05,
      "loss": 0.4999,
      "step": 11777
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.652613639831543,
      "learning_rate": 1.4516114712687584e-05,
      "loss": 0.4846,
      "step": 11778
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5911282300949097,
      "learning_rate": 1.4515206048462549e-05,
      "loss": 0.5156,
      "step": 11779
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7026973962783813,
      "learning_rate": 1.4514297337407608e-05,
      "loss": 0.5269,
      "step": 11780
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6012966632843018,
      "learning_rate": 1.4513388579532187e-05,
      "loss": 0.5171,
      "step": 11781
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6272788047790527,
      "learning_rate": 1.4512479774845704e-05,
      "loss": 0.4803,
      "step": 11782
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7571468353271484,
      "learning_rate": 1.4511570923357593e-05,
      "loss": 0.538,
      "step": 11783
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.509704828262329,
      "learning_rate": 1.451066202507728e-05,
      "loss": 0.4354,
      "step": 11784
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7037237882614136,
      "learning_rate": 1.4509753080014186e-05,
      "loss": 0.493,
      "step": 11785
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7092889547348022,
      "learning_rate": 1.4508844088177742e-05,
      "loss": 0.5154,
      "step": 11786
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5850330591201782,
      "learning_rate": 1.4507935049577375e-05,
      "loss": 0.5159,
      "step": 11787
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.715442419052124,
      "learning_rate": 1.450702596422251e-05,
      "loss": 0.4657,
      "step": 11788
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.647934913635254,
      "learning_rate": 1.4506116832122583e-05,
      "loss": 0.5054,
      "step": 11789
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5591181516647339,
      "learning_rate": 1.4505207653287014e-05,
      "loss": 0.43,
      "step": 11790
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5497238636016846,
      "learning_rate": 1.4504298427725244e-05,
      "loss": 0.4656,
      "step": 11791
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6239246129989624,
      "learning_rate": 1.450338915544669e-05,
      "loss": 0.4592,
      "step": 11792
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5696369409561157,
      "learning_rate": 1.4502479836460792e-05,
      "loss": 0.5005,
      "step": 11793
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7280259132385254,
      "learning_rate": 1.4501570470776978e-05,
      "loss": 0.4894,
      "step": 11794
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.640724778175354,
      "learning_rate": 1.4500661058404682e-05,
      "loss": 0.4901,
      "step": 11795
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5782493352890015,
      "learning_rate": 1.4499751599353331e-05,
      "loss": 0.45,
      "step": 11796
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6628209352493286,
      "learning_rate": 1.4498842093632361e-05,
      "loss": 0.4789,
      "step": 11797
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.541398525238037,
      "learning_rate": 1.4497932541251204e-05,
      "loss": 0.4711,
      "step": 11798
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4782074689865112,
      "learning_rate": 1.4497022942219297e-05,
      "loss": 0.5137,
      "step": 11799
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5482417345046997,
      "learning_rate": 1.4496113296546068e-05,
      "loss": 0.4687,
      "step": 11800
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.590348243713379,
      "learning_rate": 1.4495203604240954e-05,
      "loss": 0.4903,
      "step": 11801
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5830377340316772,
      "learning_rate": 1.4494293865313394e-05,
      "loss": 0.4465,
      "step": 11802
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5431811809539795,
      "learning_rate": 1.4493384079772815e-05,
      "loss": 0.4585,
      "step": 11803
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.505035400390625,
      "learning_rate": 1.4492474247628664e-05,
      "loss": 0.4612,
      "step": 11804
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6074059009552002,
      "learning_rate": 1.4491564368890365e-05,
      "loss": 0.5007,
      "step": 11805
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6414400339126587,
      "learning_rate": 1.4490654443567362e-05,
      "loss": 0.4557,
      "step": 11806
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8544148206710815,
      "learning_rate": 1.4489744471669094e-05,
      "loss": 0.5342,
      "step": 11807
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5768569707870483,
      "learning_rate": 1.4488834453204996e-05,
      "loss": 0.4348,
      "step": 11808
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.540314793586731,
      "learning_rate": 1.4487924388184506e-05,
      "loss": 0.4771,
      "step": 11809
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6541167497634888,
      "learning_rate": 1.4487014276617064e-05,
      "loss": 0.4845,
      "step": 11810
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.878738284111023,
      "learning_rate": 1.448610411851211e-05,
      "loss": 1.0205,
      "step": 11811
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6390163898468018,
      "learning_rate": 1.448519391387908e-05,
      "loss": 0.9942,
      "step": 11812
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7746849060058594,
      "learning_rate": 1.4484283662727419e-05,
      "loss": 0.5464,
      "step": 11813
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.819042682647705,
      "learning_rate": 1.4483373365066566e-05,
      "loss": 0.4528,
      "step": 11814
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6188198328018188,
      "learning_rate": 1.4482463020905958e-05,
      "loss": 0.4572,
      "step": 11815
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.562837839126587,
      "learning_rate": 1.4481552630255045e-05,
      "loss": 0.4647,
      "step": 11816
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7631118297576904,
      "learning_rate": 1.4480642193123264e-05,
      "loss": 0.5826,
      "step": 11817
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.62359619140625,
      "learning_rate": 1.4479731709520058e-05,
      "loss": 0.4175,
      "step": 11818
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.698798656463623,
      "learning_rate": 1.4478821179454872e-05,
      "loss": 0.5351,
      "step": 11819
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5702846050262451,
      "learning_rate": 1.4477910602937147e-05,
      "loss": 0.4277,
      "step": 11820
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6905927658081055,
      "learning_rate": 1.4476999979976325e-05,
      "loss": 0.4839,
      "step": 11821
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.55843186378479,
      "learning_rate": 1.447608931058186e-05,
      "loss": 0.4633,
      "step": 11822
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6092076301574707,
      "learning_rate": 1.4475178594763188e-05,
      "loss": 0.5392,
      "step": 11823
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5870375633239746,
      "learning_rate": 1.4474267832529759e-05,
      "loss": 0.4617,
      "step": 11824
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.036961793899536,
      "learning_rate": 1.4473357023891017e-05,
      "loss": 1.0542,
      "step": 11825
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6789337396621704,
      "learning_rate": 1.447244616885641e-05,
      "loss": 1.098,
      "step": 11826
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7656528949737549,
      "learning_rate": 1.4471535267435386e-05,
      "loss": 0.508,
      "step": 11827
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7729787826538086,
      "learning_rate": 1.447062431963739e-05,
      "loss": 0.4297,
      "step": 11828
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6431632041931152,
      "learning_rate": 1.4469713325471868e-05,
      "loss": 0.5008,
      "step": 11829
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5237243175506592,
      "learning_rate": 1.4468802284948275e-05,
      "loss": 0.4587,
      "step": 11830
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6438649892807007,
      "learning_rate": 1.4467891198076052e-05,
      "loss": 0.5429,
      "step": 11831
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6089439392089844,
      "learning_rate": 1.4466980064864659e-05,
      "loss": 0.4521,
      "step": 11832
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5508129596710205,
      "learning_rate": 1.4466068885323534e-05,
      "loss": 0.4671,
      "step": 11833
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6747817993164062,
      "learning_rate": 1.4465157659462137e-05,
      "loss": 0.4267,
      "step": 11834
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5885833501815796,
      "learning_rate": 1.4464246387289913e-05,
      "loss": 0.4798,
      "step": 11835
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.628009557723999,
      "learning_rate": 1.4463335068816316e-05,
      "loss": 0.4742,
      "step": 11836
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6455516815185547,
      "learning_rate": 1.4462423704050796e-05,
      "loss": 0.4786,
      "step": 11837
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5422601699829102,
      "learning_rate": 1.4461512293002808e-05,
      "loss": 0.4177,
      "step": 11838
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.656049132347107,
      "learning_rate": 1.44606008356818e-05,
      "loss": 0.56,
      "step": 11839
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6261796951293945,
      "learning_rate": 1.4459689332097234e-05,
      "loss": 0.5054,
      "step": 11840
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5431033372879028,
      "learning_rate": 1.4458777782258552e-05,
      "loss": 0.4811,
      "step": 11841
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.577433705329895,
      "learning_rate": 1.4457866186175217e-05,
      "loss": 0.422,
      "step": 11842
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6101739406585693,
      "learning_rate": 1.4456954543856682e-05,
      "loss": 0.571,
      "step": 11843
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5775048732757568,
      "learning_rate": 1.4456042855312403e-05,
      "loss": 0.4483,
      "step": 11844
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7353860139846802,
      "learning_rate": 1.4455131120551831e-05,
      "loss": 0.5717,
      "step": 11845
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6020978689193726,
      "learning_rate": 1.4454219339584428e-05,
      "loss": 0.4582,
      "step": 11846
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.788877010345459,
      "learning_rate": 1.4453307512419642e-05,
      "loss": 0.562,
      "step": 11847
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5778217315673828,
      "learning_rate": 1.4452395639066943e-05,
      "loss": 0.4498,
      "step": 11848
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.604660987854004,
      "learning_rate": 1.4451483719535777e-05,
      "loss": 0.4884,
      "step": 11849
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.526430368423462,
      "learning_rate": 1.4450571753835609e-05,
      "loss": 0.4123,
      "step": 11850
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5892045497894287,
      "learning_rate": 1.4449659741975892e-05,
      "loss": 0.5213,
      "step": 11851
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5685886144638062,
      "learning_rate": 1.444874768396609e-05,
      "loss": 0.4573,
      "step": 11852
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6256698369979858,
      "learning_rate": 1.4447835579815662e-05,
      "loss": 0.5283,
      "step": 11853
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6908912658691406,
      "learning_rate": 1.4446923429534065e-05,
      "loss": 0.5086,
      "step": 11854
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6547062397003174,
      "learning_rate": 1.444601123313076e-05,
      "loss": 0.5116,
      "step": 11855
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6885488033294678,
      "learning_rate": 1.4445098990615211e-05,
      "loss": 0.4815,
      "step": 11856
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6687616109848022,
      "learning_rate": 1.4444186701996876e-05,
      "loss": 0.5388,
      "step": 11857
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.53294837474823,
      "learning_rate": 1.4443274367285222e-05,
      "loss": 0.4284,
      "step": 11858
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5752028226852417,
      "learning_rate": 1.4442361986489703e-05,
      "loss": 0.5221,
      "step": 11859
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5070708990097046,
      "learning_rate": 1.4441449559619787e-05,
      "loss": 0.4304,
      "step": 11860
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6240226030349731,
      "learning_rate": 1.4440537086684938e-05,
      "loss": 0.4877,
      "step": 11861
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.623558521270752,
      "learning_rate": 1.443962456769462e-05,
      "loss": 0.4867,
      "step": 11862
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5626990795135498,
      "learning_rate": 1.4438712002658295e-05,
      "loss": 0.4236,
      "step": 11863
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.794832468032837,
      "learning_rate": 1.4437799391585427e-05,
      "loss": 0.5302,
      "step": 11864
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6395599842071533,
      "learning_rate": 1.4436886734485483e-05,
      "loss": 0.5022,
      "step": 11865
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6264278888702393,
      "learning_rate": 1.4435974031367931e-05,
      "loss": 0.4635,
      "step": 11866
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7072142362594604,
      "learning_rate": 1.4435061282242234e-05,
      "loss": 0.4898,
      "step": 11867
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5592576265335083,
      "learning_rate": 1.4434148487117861e-05,
      "loss": 0.4322,
      "step": 11868
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5702115297317505,
      "learning_rate": 1.4433235646004273e-05,
      "loss": 0.4999,
      "step": 11869
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5220675468444824,
      "learning_rate": 1.4432322758910945e-05,
      "loss": 0.4506,
      "step": 11870
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6078442335128784,
      "learning_rate": 1.4431409825847342e-05,
      "loss": 0.5062,
      "step": 11871
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6023749113082886,
      "learning_rate": 1.4430496846822935e-05,
      "loss": 0.4515,
      "step": 11872
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6148262023925781,
      "learning_rate": 1.4429583821847188e-05,
      "loss": 0.4895,
      "step": 11873
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5507153272628784,
      "learning_rate": 1.4428670750929574e-05,
      "loss": 0.4412,
      "step": 11874
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.663074016571045,
      "learning_rate": 1.4427757634079562e-05,
      "loss": 0.5083,
      "step": 11875
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6129118204116821,
      "learning_rate": 1.4426844471306627e-05,
      "loss": 0.4325,
      "step": 11876
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5705658197402954,
      "learning_rate": 1.4425931262620233e-05,
      "loss": 0.5464,
      "step": 11877
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6597710847854614,
      "learning_rate": 1.4425018008029851e-05,
      "loss": 0.4412,
      "step": 11878
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.2798585891723633,
      "learning_rate": 1.4424104707544961e-05,
      "loss": 0.9634,
      "step": 11879
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.775429129600525,
      "learning_rate": 1.4423191361175031e-05,
      "loss": 0.9886,
      "step": 11880
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6109662055969238,
      "learning_rate": 1.442227796892953e-05,
      "loss": 0.4634,
      "step": 11881
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6359769105911255,
      "learning_rate": 1.4421364530817939e-05,
      "loss": 0.4582,
      "step": 11882
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.614639401435852,
      "learning_rate": 1.4420451046849722e-05,
      "loss": 0.4776,
      "step": 11883
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5684068202972412,
      "learning_rate": 1.4419537517034365e-05,
      "loss": 0.4361,
      "step": 11884
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6301729679107666,
      "learning_rate": 1.4418623941381334e-05,
      "loss": 0.4598,
      "step": 11885
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5333229303359985,
      "learning_rate": 1.4417710319900107e-05,
      "loss": 0.4227,
      "step": 11886
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.693095326423645,
      "learning_rate": 1.4416796652600161e-05,
      "loss": 0.557,
      "step": 11887
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4824482202529907,
      "learning_rate": 1.4415882939490969e-05,
      "loss": 0.4418,
      "step": 11888
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.583755612373352,
      "learning_rate": 1.441496918058201e-05,
      "loss": 0.4761,
      "step": 11889
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5278195142745972,
      "learning_rate": 1.4414055375882763e-05,
      "loss": 0.4305,
      "step": 11890
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5916932821273804,
      "learning_rate": 1.44131415254027e-05,
      "loss": 0.498,
      "step": 11891
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5419689416885376,
      "learning_rate": 1.4412227629151306e-05,
      "loss": 0.4557,
      "step": 11892
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.606130599975586,
      "learning_rate": 1.4411313687138051e-05,
      "loss": 0.4846,
      "step": 11893
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5618842840194702,
      "learning_rate": 1.4410399699372423e-05,
      "loss": 0.4912,
      "step": 11894
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6098246574401855,
      "learning_rate": 1.4409485665863896e-05,
      "loss": 0.5228,
      "step": 11895
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6132334470748901,
      "learning_rate": 1.4408571586621951e-05,
      "loss": 0.4382,
      "step": 11896
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6291512250900269,
      "learning_rate": 1.440765746165607e-05,
      "loss": 0.4852,
      "step": 11897
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5039387941360474,
      "learning_rate": 1.4406743290975733e-05,
      "loss": 0.4307,
      "step": 11898
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5839585065841675,
      "learning_rate": 1.4405829074590422e-05,
      "loss": 0.5322,
      "step": 11899
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6177536249160767,
      "learning_rate": 1.4404914812509619e-05,
      "loss": 0.4628,
      "step": 11900
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.594740629196167,
      "learning_rate": 1.4404000504742801e-05,
      "loss": 0.5521,
      "step": 11901
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8050005435943604,
      "learning_rate": 1.440308615129946e-05,
      "loss": 0.4484,
      "step": 11902
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5795226097106934,
      "learning_rate": 1.4402171752189072e-05,
      "loss": 0.5377,
      "step": 11903
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.627880573272705,
      "learning_rate": 1.4401257307421124e-05,
      "loss": 0.4338,
      "step": 11904
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.520262598991394,
      "learning_rate": 1.4400342817005099e-05,
      "loss": 0.4652,
      "step": 11905
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6315871477127075,
      "learning_rate": 1.4399428280950484e-05,
      "loss": 0.4631,
      "step": 11906
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.700441837310791,
      "learning_rate": 1.4398513699266766e-05,
      "loss": 0.4908,
      "step": 11907
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6585582494735718,
      "learning_rate": 1.4397599071963423e-05,
      "loss": 0.4423,
      "step": 11908
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6026532649993896,
      "learning_rate": 1.4396684399049946e-05,
      "loss": 0.5592,
      "step": 11909
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.573137879371643,
      "learning_rate": 1.4395769680535823e-05,
      "loss": 0.4384,
      "step": 11910
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5845264196395874,
      "learning_rate": 1.4394854916430534e-05,
      "loss": 0.4986,
      "step": 11911
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.579405426979065,
      "learning_rate": 1.4393940106743579e-05,
      "loss": 0.4851,
      "step": 11912
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6510403156280518,
      "learning_rate": 1.4393025251484434e-05,
      "loss": 0.5137,
      "step": 11913
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5641840696334839,
      "learning_rate": 1.4392110350662594e-05,
      "loss": 0.4546,
      "step": 11914
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9449273347854614,
      "learning_rate": 1.4391195404287545e-05,
      "loss": 1.0071,
      "step": 11915
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6664483547210693,
      "learning_rate": 1.439028041236878e-05,
      "loss": 1.037,
      "step": 11916
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6766096353530884,
      "learning_rate": 1.4389365374915784e-05,
      "loss": 0.5586,
      "step": 11917
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6169310808181763,
      "learning_rate": 1.4388450291938051e-05,
      "loss": 0.4185,
      "step": 11918
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9058465957641602,
      "learning_rate": 1.4387535163445071e-05,
      "loss": 0.5926,
      "step": 11919
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6000351905822754,
      "learning_rate": 1.438661998944634e-05,
      "loss": 0.4277,
      "step": 11920
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.743438482284546,
      "learning_rate": 1.4385704769951339e-05,
      "loss": 0.587,
      "step": 11921
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.676027536392212,
      "learning_rate": 1.4384789504969567e-05,
      "loss": 0.4521,
      "step": 11922
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6569169759750366,
      "learning_rate": 1.4383874194510518e-05,
      "loss": 0.4734,
      "step": 11923
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.65146803855896,
      "learning_rate": 1.4382958838583685e-05,
      "loss": 0.4446,
      "step": 11924
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5715863704681396,
      "learning_rate": 1.4382043437198556e-05,
      "loss": 0.5055,
      "step": 11925
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.572722315788269,
      "learning_rate": 1.4381127990364634e-05,
      "loss": 0.4548,
      "step": 11926
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5813579559326172,
      "learning_rate": 1.4380212498091405e-05,
      "loss": 0.473,
      "step": 11927
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6096240282058716,
      "learning_rate": 1.4379296960388372e-05,
      "loss": 0.5192,
      "step": 11928
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5664703845977783,
      "learning_rate": 1.4378381377265021e-05,
      "loss": 0.4639,
      "step": 11929
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4710341691970825,
      "learning_rate": 1.4377465748730859e-05,
      "loss": 0.4202,
      "step": 11930
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5980384349822998,
      "learning_rate": 1.4376550074795375e-05,
      "loss": 0.5026,
      "step": 11931
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.657989740371704,
      "learning_rate": 1.4375634355468067e-05,
      "loss": 0.4966,
      "step": 11932
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6800771951675415,
      "learning_rate": 1.4374718590758436e-05,
      "loss": 0.5795,
      "step": 11933
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6375279426574707,
      "learning_rate": 1.4373802780675976e-05,
      "loss": 0.4546,
      "step": 11934
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6867597103118896,
      "learning_rate": 1.437288692523019e-05,
      "loss": 0.5246,
      "step": 11935
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.551682472229004,
      "learning_rate": 1.4371971024430571e-05,
      "loss": 0.4469,
      "step": 11936
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6455621719360352,
      "learning_rate": 1.4371055078286623e-05,
      "loss": 0.5129,
      "step": 11937
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6125166416168213,
      "learning_rate": 1.4370139086807845e-05,
      "loss": 0.4926,
      "step": 11938
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.584637999534607,
      "learning_rate": 1.4369223050003734e-05,
      "loss": 0.4889,
      "step": 11939
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.528388500213623,
      "learning_rate": 1.4368306967883794e-05,
      "loss": 0.4358,
      "step": 11940
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5659047365188599,
      "learning_rate": 1.4367390840457525e-05,
      "loss": 0.5046,
      "step": 11941
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5831398963928223,
      "learning_rate": 1.4366474667734429e-05,
      "loss": 0.4826,
      "step": 11942
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7232065200805664,
      "learning_rate": 1.436555844972401e-05,
      "loss": 0.5081,
      "step": 11943
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5450797080993652,
      "learning_rate": 1.4364642186435768e-05,
      "loss": 0.4389,
      "step": 11944
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.622866153717041,
      "learning_rate": 1.4363725877879207e-05,
      "loss": 0.5032,
      "step": 11945
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5553339719772339,
      "learning_rate": 1.4362809524063832e-05,
      "loss": 0.4299,
      "step": 11946
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7227939367294312,
      "learning_rate": 1.4361893124999144e-05,
      "loss": 0.5333,
      "step": 11947
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5459270477294922,
      "learning_rate": 1.4360976680694651e-05,
      "loss": 0.4461,
      "step": 11948
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.708039402961731,
      "learning_rate": 1.4360060191159855e-05,
      "loss": 0.5028,
      "step": 11949
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.511077642440796,
      "learning_rate": 1.4359143656404265e-05,
      "loss": 0.426,
      "step": 11950
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5900092124938965,
      "learning_rate": 1.4358227076437381e-05,
      "loss": 0.514,
      "step": 11951
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6046183109283447,
      "learning_rate": 1.4357310451268715e-05,
      "loss": 0.4349,
      "step": 11952
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6227515935897827,
      "learning_rate": 1.4356393780907771e-05,
      "loss": 0.4946,
      "step": 11953
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.545607328414917,
      "learning_rate": 1.435547706536406e-05,
      "loss": 0.4141,
      "step": 11954
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5890905857086182,
      "learning_rate": 1.4354560304647082e-05,
      "loss": 0.4735,
      "step": 11955
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6856714487075806,
      "learning_rate": 1.4353643498766356e-05,
      "loss": 0.4827,
      "step": 11956
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5763717889785767,
      "learning_rate": 1.435272664773138e-05,
      "loss": 0.4706,
      "step": 11957
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6370208263397217,
      "learning_rate": 1.4351809751551673e-05,
      "loss": 0.4547,
      "step": 11958
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5623682737350464,
      "learning_rate": 1.4350892810236736e-05,
      "loss": 0.5095,
      "step": 11959
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6027966737747192,
      "learning_rate": 1.4349975823796086e-05,
      "loss": 0.4468,
      "step": 11960
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5536816120147705,
      "learning_rate": 1.434905879223923e-05,
      "loss": 0.5204,
      "step": 11961
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5591683387756348,
      "learning_rate": 1.4348141715575678e-05,
      "loss": 0.4707,
      "step": 11962
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6058540344238281,
      "learning_rate": 1.4347224593814946e-05,
      "loss": 0.5174,
      "step": 11963
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.531356930732727,
      "learning_rate": 1.4346307426966543e-05,
      "loss": 0.4148,
      "step": 11964
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9032564163208008,
      "learning_rate": 1.4345390215039979e-05,
      "loss": 0.9689,
      "step": 11965
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6893426179885864,
      "learning_rate": 1.4344472958044775e-05,
      "loss": 1.056,
      "step": 11966
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.69287109375,
      "learning_rate": 1.4343555655990434e-05,
      "loss": 0.5082,
      "step": 11967
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6407759189605713,
      "learning_rate": 1.4342638308886478e-05,
      "loss": 0.4806,
      "step": 11968
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6215343475341797,
      "learning_rate": 1.4341720916742419e-05,
      "loss": 0.5286,
      "step": 11969
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5505733489990234,
      "learning_rate": 1.4340803479567769e-05,
      "loss": 0.496,
      "step": 11970
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6114329099655151,
      "learning_rate": 1.4339885997372048e-05,
      "loss": 0.5055,
      "step": 11971
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6960361003875732,
      "learning_rate": 1.4338968470164767e-05,
      "loss": 0.5,
      "step": 11972
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6246477365493774,
      "learning_rate": 1.4338050897955444e-05,
      "loss": 0.4969,
      "step": 11973
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.648378610610962,
      "learning_rate": 1.4337133280753601e-05,
      "loss": 0.4782,
      "step": 11974
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7399166822433472,
      "learning_rate": 1.4336215618568745e-05,
      "loss": 0.5456,
      "step": 11975
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5125271081924438,
      "learning_rate": 1.4335297911410402e-05,
      "loss": 0.4391,
      "step": 11976
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5878921747207642,
      "learning_rate": 1.4334380159288088e-05,
      "loss": 0.5078,
      "step": 11977
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.51165771484375,
      "learning_rate": 1.433346236221132e-05,
      "loss": 0.4438,
      "step": 11978
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4792745113372803,
      "learning_rate": 1.4332544520189617e-05,
      "loss": 0.4442,
      "step": 11979
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5531445741653442,
      "learning_rate": 1.43316266332325e-05,
      "loss": 0.4384,
      "step": 11980
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5587576627731323,
      "learning_rate": 1.4330708701349486e-05,
      "loss": 0.516,
      "step": 11981
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.520094633102417,
      "learning_rate": 1.4329790724550102e-05,
      "loss": 0.4304,
      "step": 11982
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8012523651123047,
      "learning_rate": 1.432887270284386e-05,
      "loss": 1.06,
      "step": 11983
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5735385417938232,
      "learning_rate": 1.4327954636240289e-05,
      "loss": 1.0332,
      "step": 11984
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6938425302505493,
      "learning_rate": 1.4327036524748903e-05,
      "loss": 0.5117,
      "step": 11985
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.651677131652832,
      "learning_rate": 1.4326118368379234e-05,
      "loss": 0.4731,
      "step": 11986
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6440011262893677,
      "learning_rate": 1.4325200167140798e-05,
      "loss": 0.5162,
      "step": 11987
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5230845212936401,
      "learning_rate": 1.432428192104312e-05,
      "loss": 0.4389,
      "step": 11988
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7178525924682617,
      "learning_rate": 1.4323363630095724e-05,
      "loss": 0.5302,
      "step": 11989
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.65596342086792,
      "learning_rate": 1.4322445294308133e-05,
      "loss": 0.4423,
      "step": 11990
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7472413778305054,
      "learning_rate": 1.432152691368987e-05,
      "loss": 0.5495,
      "step": 11991
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5033413171768188,
      "learning_rate": 1.4320608488250467e-05,
      "loss": 0.4405,
      "step": 11992
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5843626260757446,
      "learning_rate": 1.4319690017999444e-05,
      "loss": 0.5101,
      "step": 11993
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6237810850143433,
      "learning_rate": 1.4318771502946324e-05,
      "loss": 0.4833,
      "step": 11994
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.67728590965271,
      "learning_rate": 1.4317852943100643e-05,
      "loss": 0.5252,
      "step": 11995
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5642808675765991,
      "learning_rate": 1.431693433847192e-05,
      "loss": 0.4464,
      "step": 11996
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5628182888031006,
      "learning_rate": 1.4316015689069684e-05,
      "loss": 0.4985,
      "step": 11997
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6559548377990723,
      "learning_rate": 1.4315096994903463e-05,
      "loss": 0.4668,
      "step": 11998
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.894680380821228,
      "learning_rate": 1.4314178255982787e-05,
      "loss": 0.981,
      "step": 11999
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6050832271575928,
      "learning_rate": 1.4313259472317188e-05,
      "loss": 1.0469,
      "step": 12000
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8292090892791748,
      "learning_rate": 1.4312340643916183e-05,
      "loss": 0.4859,
      "step": 12001
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.784936547279358,
      "learning_rate": 1.4311421770789315e-05,
      "loss": 0.4802,
      "step": 12002
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7172199487686157,
      "learning_rate": 1.4310502852946106e-05,
      "loss": 0.5103,
      "step": 12003
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5421236753463745,
      "learning_rate": 1.4309583890396092e-05,
      "loss": 0.4278,
      "step": 12004
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6410404443740845,
      "learning_rate": 1.4308664883148801e-05,
      "loss": 0.5697,
      "step": 12005
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.626779556274414,
      "learning_rate": 1.4307745831213767e-05,
      "loss": 0.4386,
      "step": 12006
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6712844371795654,
      "learning_rate": 1.4306826734600519e-05,
      "loss": 0.4722,
      "step": 12007
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5683777332305908,
      "learning_rate": 1.4305907593318589e-05,
      "loss": 0.4887,
      "step": 12008
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6583818197250366,
      "learning_rate": 1.4304988407377511e-05,
      "loss": 0.5431,
      "step": 12009
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.57089364528656,
      "learning_rate": 1.4304069176786826e-05,
      "loss": 0.4216,
      "step": 12010
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7217801809310913,
      "learning_rate": 1.4303149901556053e-05,
      "loss": 0.5017,
      "step": 12011
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5113818645477295,
      "learning_rate": 1.4302230581694739e-05,
      "loss": 0.4328,
      "step": 12012
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9607850313186646,
      "learning_rate": 1.4301311217212412e-05,
      "loss": 1.0181,
      "step": 12013
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6632380485534668,
      "learning_rate": 1.430039180811861e-05,
      "loss": 1.0186,
      "step": 12014
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7042183876037598,
      "learning_rate": 1.429947235442287e-05,
      "loss": 0.5276,
      "step": 12015
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.657348871231079,
      "learning_rate": 1.4298552856134724e-05,
      "loss": 0.437,
      "step": 12016
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7949506044387817,
      "learning_rate": 1.4297633313263713e-05,
      "loss": 0.4863,
      "step": 12017
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6018768548965454,
      "learning_rate": 1.4296713725819373e-05,
      "loss": 0.4138,
      "step": 12018
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9833621978759766,
      "learning_rate": 1.4295794093811237e-05,
      "loss": 0.4949,
      "step": 12019
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7075539827346802,
      "learning_rate": 1.4294874417248851e-05,
      "loss": 0.4889,
      "step": 12020
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6153250932693481,
      "learning_rate": 1.4293954696141747e-05,
      "loss": 0.4961,
      "step": 12021
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6593425273895264,
      "learning_rate": 1.4293034930499467e-05,
      "loss": 0.4722,
      "step": 12022
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.604122519493103,
      "learning_rate": 1.4292115120331551e-05,
      "loss": 0.4957,
      "step": 12023
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5564842224121094,
      "learning_rate": 1.4291195265647538e-05,
      "loss": 0.4301,
      "step": 12024
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.564492106437683,
      "learning_rate": 1.4290275366456968e-05,
      "loss": 0.4889,
      "step": 12025
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8049416542053223,
      "learning_rate": 1.428935542276938e-05,
      "loss": 0.4346,
      "step": 12026
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6390467882156372,
      "learning_rate": 1.4288435434594317e-05,
      "loss": 0.5064,
      "step": 12027
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5591386556625366,
      "learning_rate": 1.4287515401941326e-05,
      "loss": 0.4531,
      "step": 12028
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5501923561096191,
      "learning_rate": 1.4286595324819938e-05,
      "loss": 0.5097,
      "step": 12029
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5813759565353394,
      "learning_rate": 1.4285675203239704e-05,
      "loss": 0.4302,
      "step": 12030
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6506139039993286,
      "learning_rate": 1.4284755037210165e-05,
      "loss": 0.4969,
      "step": 12031
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6392790079116821,
      "learning_rate": 1.4283834826740865e-05,
      "loss": 0.4444,
      "step": 12032
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6109646558761597,
      "learning_rate": 1.4282914571841348e-05,
      "loss": 0.5042,
      "step": 12033
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5325641632080078,
      "learning_rate": 1.4281994272521158e-05,
      "loss": 0.4735,
      "step": 12034
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6983345746994019,
      "learning_rate": 1.4281073928789839e-05,
      "loss": 0.4926,
      "step": 12035
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6445566415786743,
      "learning_rate": 1.428015354065694e-05,
      "loss": 0.4615,
      "step": 12036
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6840085983276367,
      "learning_rate": 1.4279233108132003e-05,
      "loss": 0.5329,
      "step": 12037
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5580095052719116,
      "learning_rate": 1.4278312631224576e-05,
      "loss": 0.431,
      "step": 12038
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5729572772979736,
      "learning_rate": 1.4277392109944204e-05,
      "loss": 0.4895,
      "step": 12039
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6237188577651978,
      "learning_rate": 1.4276471544300438e-05,
      "loss": 0.4671,
      "step": 12040
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7345082759857178,
      "learning_rate": 1.4275550934302822e-05,
      "loss": 0.4824,
      "step": 12041
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.570703148841858,
      "learning_rate": 1.4274630279960905e-05,
      "loss": 0.4361,
      "step": 12042
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.670333981513977,
      "learning_rate": 1.427370958128424e-05,
      "loss": 0.5568,
      "step": 12043
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.644278645515442,
      "learning_rate": 1.4272788838282372e-05,
      "loss": 0.4596,
      "step": 12044
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5997825860977173,
      "learning_rate": 1.427186805096485e-05,
      "loss": 0.5045,
      "step": 12045
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.678674578666687,
      "learning_rate": 1.4270947219341224e-05,
      "loss": 0.4778,
      "step": 12046
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5383687019348145,
      "learning_rate": 1.4270026343421048e-05,
      "loss": 0.4634,
      "step": 12047
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5013059377670288,
      "learning_rate": 1.4269105423213872e-05,
      "loss": 0.4415,
      "step": 12048
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5537225008010864,
      "learning_rate": 1.4268184458729244e-05,
      "loss": 0.4814,
      "step": 12049
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6438045501708984,
      "learning_rate": 1.4267263449976716e-05,
      "loss": 0.4984,
      "step": 12050
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6103070974349976,
      "learning_rate": 1.4266342396965847e-05,
      "loss": 0.4919,
      "step": 12051
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.552390217781067,
      "learning_rate": 1.4265421299706179e-05,
      "loss": 0.4487,
      "step": 12052
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5597822666168213,
      "learning_rate": 1.4264500158207274e-05,
      "loss": 0.5486,
      "step": 12053
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6012526750564575,
      "learning_rate": 1.4263578972478684e-05,
      "loss": 0.4713,
      "step": 12054
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.682310938835144,
      "learning_rate": 1.4262657742529963e-05,
      "loss": 0.6057,
      "step": 12055
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5522319078445435,
      "learning_rate": 1.4261736468370666e-05,
      "loss": 0.4419,
      "step": 12056
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6781784296035767,
      "learning_rate": 1.4260815150010345e-05,
      "loss": 0.5087,
      "step": 12057
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6682415008544922,
      "learning_rate": 1.4259893787458557e-05,
      "loss": 0.5057,
      "step": 12058
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5872520208358765,
      "learning_rate": 1.4258972380724857e-05,
      "loss": 0.5026,
      "step": 12059
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4833974838256836,
      "learning_rate": 1.4258050929818805e-05,
      "loss": 0.4137,
      "step": 12060
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5805225372314453,
      "learning_rate": 1.4257129434749959e-05,
      "loss": 0.5039,
      "step": 12061
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5531431436538696,
      "learning_rate": 1.4256207895527867e-05,
      "loss": 0.4525,
      "step": 12062
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.663619041442871,
      "learning_rate": 1.4255286312162098e-05,
      "loss": 0.4743,
      "step": 12063
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.603463053703308,
      "learning_rate": 1.4254364684662203e-05,
      "loss": 0.4581,
      "step": 12064
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5467103719711304,
      "learning_rate": 1.4253443013037744e-05,
      "loss": 0.5041,
      "step": 12065
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4738892316818237,
      "learning_rate": 1.4252521297298278e-05,
      "loss": 0.4739,
      "step": 12066
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5950506925582886,
      "learning_rate": 1.4251599537453369e-05,
      "loss": 0.5143,
      "step": 12067
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5387943983078003,
      "learning_rate": 1.425067773351257e-05,
      "loss": 0.4643,
      "step": 12068
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9015439748764038,
      "learning_rate": 1.424975588548545e-05,
      "loss": 0.462,
      "step": 12069
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5756596326828003,
      "learning_rate": 1.4248833993381564e-05,
      "loss": 0.4375,
      "step": 12070
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0781006813049316,
      "learning_rate": 1.4247912057210473e-05,
      "loss": 0.9838,
      "step": 12071
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7066270112991333,
      "learning_rate": 1.4246990076981746e-05,
      "loss": 1.0481,
      "step": 12072
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.694337248802185,
      "learning_rate": 1.4246068052704938e-05,
      "loss": 0.5439,
      "step": 12073
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7049821615219116,
      "learning_rate": 1.4245145984389614e-05,
      "loss": 0.4778,
      "step": 12074
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.586338996887207,
      "learning_rate": 1.4244223872045337e-05,
      "loss": 0.4758,
      "step": 12075
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5691496133804321,
      "learning_rate": 1.4243301715681672e-05,
      "loss": 0.4637,
      "step": 12076
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.608258843421936,
      "learning_rate": 1.4242379515308184e-05,
      "loss": 0.5689,
      "step": 12077
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5330255031585693,
      "learning_rate": 1.4241457270934435e-05,
      "loss": 0.4459,
      "step": 12078
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6522458791732788,
      "learning_rate": 1.4240534982569994e-05,
      "loss": 0.4791,
      "step": 12079
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5635299682617188,
      "learning_rate": 1.4239612650224423e-05,
      "loss": 0.424,
      "step": 12080
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6245057582855225,
      "learning_rate": 1.423869027390729e-05,
      "loss": 0.532,
      "step": 12081
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.512882113456726,
      "learning_rate": 1.423776785362816e-05,
      "loss": 0.4292,
      "step": 12082
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.651168942451477,
      "learning_rate": 1.4236845389396603e-05,
      "loss": 0.5668,
      "step": 12083
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5169963836669922,
      "learning_rate": 1.4235922881222182e-05,
      "loss": 0.4581,
      "step": 12084
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5672192573547363,
      "learning_rate": 1.4235000329114467e-05,
      "loss": 0.5057,
      "step": 12085
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4788892269134521,
      "learning_rate": 1.4234077733083025e-05,
      "loss": 0.4257,
      "step": 12086
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6514850854873657,
      "learning_rate": 1.423315509313743e-05,
      "loss": 0.5108,
      "step": 12087
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.560228943824768,
      "learning_rate": 1.4232232409287243e-05,
      "loss": 0.4422,
      "step": 12088
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5531834363937378,
      "learning_rate": 1.4231309681542042e-05,
      "loss": 0.4902,
      "step": 12089
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.733710527420044,
      "learning_rate": 1.4230386909911393e-05,
      "loss": 0.5621,
      "step": 12090
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4653500318527222,
      "learning_rate": 1.4229464094404866e-05,
      "loss": 0.4555,
      "step": 12091
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5414972305297852,
      "learning_rate": 1.4228541235032034e-05,
      "loss": 0.4909,
      "step": 12092
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7167069911956787,
      "learning_rate": 1.4227618331802467e-05,
      "loss": 0.5271,
      "step": 12093
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.485988974571228,
      "learning_rate": 1.4226695384725738e-05,
      "loss": 0.4825,
      "step": 12094
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.721742033958435,
      "learning_rate": 1.4225772393811418e-05,
      "loss": 0.5947,
      "step": 12095
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4926189184188843,
      "learning_rate": 1.4224849359069081e-05,
      "loss": 0.4318,
      "step": 12096
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.55800199508667,
      "learning_rate": 1.42239262805083e-05,
      "loss": 0.474,
      "step": 12097
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5766340494155884,
      "learning_rate": 1.4223003158138652e-05,
      "loss": 0.4533,
      "step": 12098
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6268131732940674,
      "learning_rate": 1.4222079991969707e-05,
      "loss": 0.5187,
      "step": 12099
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5661966800689697,
      "learning_rate": 1.422115678201104e-05,
      "loss": 0.4326,
      "step": 12100
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.588128685951233,
      "learning_rate": 1.4220233528272227e-05,
      "loss": 0.4918,
      "step": 12101
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6070053577423096,
      "learning_rate": 1.4219310230762844e-05,
      "loss": 0.4673,
      "step": 12102
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6569935083389282,
      "learning_rate": 1.4218386889492469e-05,
      "loss": 0.5584,
      "step": 12103
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5762348175048828,
      "learning_rate": 1.4217463504470673e-05,
      "loss": 0.4586,
      "step": 12104
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7332864999771118,
      "learning_rate": 1.4216540075707039e-05,
      "loss": 0.4999,
      "step": 12105
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5849937200546265,
      "learning_rate": 1.4215616603211139e-05,
      "loss": 0.4505,
      "step": 12106
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5744367837905884,
      "learning_rate": 1.4214693086992558e-05,
      "loss": 0.4713,
      "step": 12107
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5659754276275635,
      "learning_rate": 1.4213769527060866e-05,
      "loss": 0.4508,
      "step": 12108
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.55985689163208,
      "learning_rate": 1.421284592342565e-05,
      "loss": 0.4996,
      "step": 12109
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.56892991065979,
      "learning_rate": 1.4211922276096481e-05,
      "loss": 0.4466,
      "step": 12110
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5529899597167969,
      "learning_rate": 1.4210998585082944e-05,
      "loss": 0.4784,
      "step": 12111
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6953413486480713,
      "learning_rate": 1.4210074850394618e-05,
      "loss": 0.4483,
      "step": 12112
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4917715787887573,
      "learning_rate": 1.4209151072041086e-05,
      "loss": 0.4514,
      "step": 12113
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6669362783432007,
      "learning_rate": 1.4208227250031921e-05,
      "loss": 0.4604,
      "step": 12114
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5960774421691895,
      "learning_rate": 1.4207303384376714e-05,
      "loss": 0.4903,
      "step": 12115
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6342780590057373,
      "learning_rate": 1.4206379475085043e-05,
      "loss": 0.4678,
      "step": 12116
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8500272035598755,
      "learning_rate": 1.4205455522166491e-05,
      "loss": 0.5363,
      "step": 12117
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.691306471824646,
      "learning_rate": 1.4204531525630638e-05,
      "loss": 0.4808,
      "step": 12118
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5649662017822266,
      "learning_rate": 1.4203607485487073e-05,
      "loss": 0.5058,
      "step": 12119
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5825532674789429,
      "learning_rate": 1.4202683401745374e-05,
      "loss": 0.4325,
      "step": 12120
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5822786092758179,
      "learning_rate": 1.420175927441513e-05,
      "loss": 0.4947,
      "step": 12121
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5020488500595093,
      "learning_rate": 1.4200835103505919e-05,
      "loss": 0.4314,
      "step": 12122
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5555349588394165,
      "learning_rate": 1.4199910889027335e-05,
      "loss": 0.4616,
      "step": 12123
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7769664525985718,
      "learning_rate": 1.4198986630988955e-05,
      "loss": 0.4597,
      "step": 12124
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5856571197509766,
      "learning_rate": 1.4198062329400373e-05,
      "loss": 0.4641,
      "step": 12125
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6387027502059937,
      "learning_rate": 1.419713798427117e-05,
      "loss": 0.4514,
      "step": 12126
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.652880072593689,
      "learning_rate": 1.4196213595610934e-05,
      "loss": 0.5343,
      "step": 12127
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.553823709487915,
      "learning_rate": 1.4195289163429253e-05,
      "loss": 0.5112,
      "step": 12128
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6001524925231934,
      "learning_rate": 1.4194364687735717e-05,
      "loss": 0.5198,
      "step": 12129
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5491816997528076,
      "learning_rate": 1.4193440168539908e-05,
      "loss": 0.428,
      "step": 12130
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7403644323349,
      "learning_rate": 1.4192515605851425e-05,
      "loss": 0.4968,
      "step": 12131
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5411189794540405,
      "learning_rate": 1.4191590999679845e-05,
      "loss": 0.4429,
      "step": 12132
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5115282535552979,
      "learning_rate": 1.4190666350034764e-05,
      "loss": 0.4569,
      "step": 12133
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.471834421157837,
      "learning_rate": 1.4189741656925775e-05,
      "loss": 0.4517,
      "step": 12134
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.680203914642334,
      "learning_rate": 1.4188816920362464e-05,
      "loss": 0.5288,
      "step": 12135
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5203936100006104,
      "learning_rate": 1.4187892140354421e-05,
      "loss": 0.4675,
      "step": 12136
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6434967517852783,
      "learning_rate": 1.4186967316911244e-05,
      "loss": 0.4663,
      "step": 12137
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5873366594314575,
      "learning_rate": 1.4186042450042516e-05,
      "loss": 0.4865,
      "step": 12138
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.686933159828186,
      "learning_rate": 1.418511753975784e-05,
      "loss": 0.5456,
      "step": 12139
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4987595081329346,
      "learning_rate": 1.41841925860668e-05,
      "loss": 0.4238,
      "step": 12140
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7570420503616333,
      "learning_rate": 1.4183267588978991e-05,
      "loss": 0.5156,
      "step": 12141
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6854194402694702,
      "learning_rate": 1.4182342548504004e-05,
      "loss": 0.4833,
      "step": 12142
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.630712866783142,
      "learning_rate": 1.4181417464651444e-05,
      "loss": 0.507,
      "step": 12143
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6395223140716553,
      "learning_rate": 1.4180492337430893e-05,
      "loss": 0.4525,
      "step": 12144
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6301993131637573,
      "learning_rate": 1.4179567166851955e-05,
      "loss": 0.4813,
      "step": 12145
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6364688873291016,
      "learning_rate": 1.4178641952924221e-05,
      "loss": 0.4281,
      "step": 12146
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4824929237365723,
      "learning_rate": 1.4177716695657289e-05,
      "loss": 0.4804,
      "step": 12147
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4936915636062622,
      "learning_rate": 1.4176791395060749e-05,
      "loss": 0.4403,
      "step": 12148
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7249644994735718,
      "learning_rate": 1.4175866051144211e-05,
      "loss": 0.5378,
      "step": 12149
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5738186836242676,
      "learning_rate": 1.4174940663917258e-05,
      "loss": 0.4364,
      "step": 12150
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.603023886680603,
      "learning_rate": 1.4174015233389498e-05,
      "loss": 0.5086,
      "step": 12151
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5601890087127686,
      "learning_rate": 1.4173089759570524e-05,
      "loss": 0.4605,
      "step": 12152
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.2097673416137695,
      "learning_rate": 1.4172164242469938e-05,
      "loss": 0.5096,
      "step": 12153
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4889556169509888,
      "learning_rate": 1.4171238682097336e-05,
      "loss": 0.4307,
      "step": 12154
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.585924744606018,
      "learning_rate": 1.4170313078462318e-05,
      "loss": 0.5421,
      "step": 12155
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5009530782699585,
      "learning_rate": 1.4169387431574484e-05,
      "loss": 0.4207,
      "step": 12156
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6921292543411255,
      "learning_rate": 1.4168461741443438e-05,
      "loss": 0.4961,
      "step": 12157
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5654312372207642,
      "learning_rate": 1.4167536008078775e-05,
      "loss": 0.4537,
      "step": 12158
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5665925741195679,
      "learning_rate": 1.4166610231490103e-05,
      "loss": 0.4795,
      "step": 12159
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5360333919525146,
      "learning_rate": 1.4165684411687015e-05,
      "loss": 0.5198,
      "step": 12160
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5764164924621582,
      "learning_rate": 1.4164758548679123e-05,
      "loss": 0.4896,
      "step": 12161
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5973728895187378,
      "learning_rate": 1.4163832642476023e-05,
      "loss": 0.4803,
      "step": 12162
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5823523998260498,
      "learning_rate": 1.4162906693087322e-05,
      "loss": 0.4933,
      "step": 12163
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.617908239364624,
      "learning_rate": 1.416198070052262e-05,
      "loss": 0.4389,
      "step": 12164
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6542161703109741,
      "learning_rate": 1.4161054664791523e-05,
      "loss": 0.5337,
      "step": 12165
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6257224082946777,
      "learning_rate": 1.4160128585903632e-05,
      "loss": 0.4909,
      "step": 12166
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5735647678375244,
      "learning_rate": 1.4159202463868563e-05,
      "loss": 0.4988,
      "step": 12167
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.575914740562439,
      "learning_rate": 1.4158276298695905e-05,
      "loss": 0.4836,
      "step": 12168
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6525108814239502,
      "learning_rate": 1.4157350090395276e-05,
      "loss": 0.5259,
      "step": 12169
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4906439781188965,
      "learning_rate": 1.4156423838976282e-05,
      "loss": 0.4375,
      "step": 12170
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5353449583053589,
      "learning_rate": 1.4155497544448522e-05,
      "loss": 0.4828,
      "step": 12171
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.604164958000183,
      "learning_rate": 1.4154571206821611e-05,
      "loss": 0.4691,
      "step": 12172
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6392765045166016,
      "learning_rate": 1.415364482610515e-05,
      "loss": 0.4806,
      "step": 12173
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.546566367149353,
      "learning_rate": 1.4152718402308749e-05,
      "loss": 0.4376,
      "step": 12174
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6444261074066162,
      "learning_rate": 1.4151791935442022e-05,
      "loss": 0.4988,
      "step": 12175
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5094419717788696,
      "learning_rate": 1.4150865425514568e-05,
      "loss": 0.436,
      "step": 12176
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6000783443450928,
      "learning_rate": 1.4149938872536009e-05,
      "loss": 0.5593,
      "step": 12177
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5417858362197876,
      "learning_rate": 1.4149012276515943e-05,
      "loss": 0.4579,
      "step": 12178
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6561459302902222,
      "learning_rate": 1.4148085637463986e-05,
      "loss": 0.504,
      "step": 12179
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5889253616333008,
      "learning_rate": 1.4147158955389748e-05,
      "loss": 0.4418,
      "step": 12180
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5683681964874268,
      "learning_rate": 1.4146232230302842e-05,
      "loss": 0.5036,
      "step": 12181
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5326794385910034,
      "learning_rate": 1.4145305462212874e-05,
      "loss": 0.4389,
      "step": 12182
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.502219319343567,
      "learning_rate": 1.4144378651129462e-05,
      "loss": 0.5111,
      "step": 12183
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5833991765975952,
      "learning_rate": 1.4143451797062213e-05,
      "loss": 0.4457,
      "step": 12184
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5570909976959229,
      "learning_rate": 1.414252490002075e-05,
      "loss": 0.4872,
      "step": 12185
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.539817452430725,
      "learning_rate": 1.4141597960014673e-05,
      "loss": 0.448,
      "step": 12186
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7852089405059814,
      "learning_rate": 1.4140670977053604e-05,
      "loss": 1.0769,
      "step": 12187
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5821973085403442,
      "learning_rate": 1.4139743951147156e-05,
      "loss": 1.0272,
      "step": 12188
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6136540174484253,
      "learning_rate": 1.4138816882304944e-05,
      "loss": 0.4825,
      "step": 12189
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.705281376838684,
      "learning_rate": 1.413788977053658e-05,
      "loss": 0.487,
      "step": 12190
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6780281066894531,
      "learning_rate": 1.4136962615851686e-05,
      "loss": 0.4933,
      "step": 12191
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.54945969581604,
      "learning_rate": 1.4136035418259872e-05,
      "loss": 0.4665,
      "step": 12192
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5315320491790771,
      "learning_rate": 1.413510817777076e-05,
      "loss": 0.5234,
      "step": 12193
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6151903867721558,
      "learning_rate": 1.413418089439396e-05,
      "loss": 0.4582,
      "step": 12194
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7235445976257324,
      "learning_rate": 1.4133253568139096e-05,
      "loss": 1.0098,
      "step": 12195
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5209287405014038,
      "learning_rate": 1.413232619901578e-05,
      "loss": 1.0176,
      "step": 12196
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.685251235961914,
      "learning_rate": 1.4131398787033637e-05,
      "loss": 0.5222,
      "step": 12197
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5931507349014282,
      "learning_rate": 1.413047133220228e-05,
      "loss": 0.4432,
      "step": 12198
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6323456764221191,
      "learning_rate": 1.4129543834531332e-05,
      "loss": 0.4728,
      "step": 12199
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5640308856964111,
      "learning_rate": 1.4128616294030409e-05,
      "loss": 0.4478,
      "step": 12200
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6342071294784546,
      "learning_rate": 1.4127688710709136e-05,
      "loss": 0.4941,
      "step": 12201
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.590556263923645,
      "learning_rate": 1.4126761084577126e-05,
      "loss": 0.4345,
      "step": 12202
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5723226070404053,
      "learning_rate": 1.4125833415644009e-05,
      "loss": 0.4845,
      "step": 12203
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5360753536224365,
      "learning_rate": 1.4124905703919397e-05,
      "loss": 0.4491,
      "step": 12204
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.499110460281372,
      "learning_rate": 1.412397794941292e-05,
      "loss": 0.4683,
      "step": 12205
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5087034702301025,
      "learning_rate": 1.4123050152134197e-05,
      "loss": 0.4254,
      "step": 12206
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7342932224273682,
      "learning_rate": 1.412212231209285e-05,
      "loss": 0.4971,
      "step": 12207
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.512728214263916,
      "learning_rate": 1.4121194429298503e-05,
      "loss": 0.4261,
      "step": 12208
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6496425867080688,
      "learning_rate": 1.412026650376078e-05,
      "loss": 0.5123,
      "step": 12209
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.680670976638794,
      "learning_rate": 1.4119338535489304e-05,
      "loss": 0.4356,
      "step": 12210
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6738144159317017,
      "learning_rate": 1.4118410524493705e-05,
      "loss": 0.518,
      "step": 12211
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5292394161224365,
      "learning_rate": 1.4117482470783596e-05,
      "loss": 0.4359,
      "step": 12212
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6038256883621216,
      "learning_rate": 1.4116554374368614e-05,
      "loss": 0.5254,
      "step": 12213
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6188520193099976,
      "learning_rate": 1.411562623525838e-05,
      "loss": 0.4315,
      "step": 12214
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6096657514572144,
      "learning_rate": 1.4114698053462519e-05,
      "loss": 0.4731,
      "step": 12215
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.563894510269165,
      "learning_rate": 1.411376982899066e-05,
      "loss": 0.4331,
      "step": 12216
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.546667218208313,
      "learning_rate": 1.4112841561852431e-05,
      "loss": 0.4785,
      "step": 12217
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.548150897026062,
      "learning_rate": 1.4111913252057459e-05,
      "loss": 0.4446,
      "step": 12218
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.897313117980957,
      "learning_rate": 1.4110984899615367e-05,
      "loss": 0.9809,
      "step": 12219
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6886320114135742,
      "learning_rate": 1.411005650453579e-05,
      "loss": 1.0506,
      "step": 12220
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7207460403442383,
      "learning_rate": 1.4109128066828359e-05,
      "loss": 0.5119,
      "step": 12221
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6550326347351074,
      "learning_rate": 1.4108199586502691e-05,
      "loss": 0.4351,
      "step": 12222
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6684943437576294,
      "learning_rate": 1.410727106356843e-05,
      "loss": 0.5059,
      "step": 12223
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5699958801269531,
      "learning_rate": 1.4106342498035199e-05,
      "loss": 0.4676,
      "step": 12224
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.686508297920227,
      "learning_rate": 1.410541388991263e-05,
      "loss": 0.5535,
      "step": 12225
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6110873222351074,
      "learning_rate": 1.4104485239210352e-05,
      "loss": 0.447,
      "step": 12226
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6431105136871338,
      "learning_rate": 1.4103556545938003e-05,
      "loss": 0.4855,
      "step": 12227
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5878334045410156,
      "learning_rate": 1.4102627810105203e-05,
      "loss": 0.495,
      "step": 12228
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7465990781784058,
      "learning_rate": 1.41016990317216e-05,
      "loss": 0.4934,
      "step": 12229
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6115785837173462,
      "learning_rate": 1.4100770210796814e-05,
      "loss": 0.443,
      "step": 12230
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5662665367126465,
      "learning_rate": 1.4099841347340485e-05,
      "loss": 0.4873,
      "step": 12231
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4836379289627075,
      "learning_rate": 1.4098912441362248e-05,
      "loss": 0.431,
      "step": 12232
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.765793800354004,
      "learning_rate": 1.4097983492871733e-05,
      "loss": 0.547,
      "step": 12233
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5459542274475098,
      "learning_rate": 1.4097054501878575e-05,
      "loss": 0.4501,
      "step": 12234
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7152310609817505,
      "learning_rate": 1.409612546839241e-05,
      "loss": 0.5999,
      "step": 12235
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5536645650863647,
      "learning_rate": 1.4095196392422876e-05,
      "loss": 0.4509,
      "step": 12236
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5683720111846924,
      "learning_rate": 1.4094267273979605e-05,
      "loss": 0.4897,
      "step": 12237
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.677383542060852,
      "learning_rate": 1.4093338113072234e-05,
      "loss": 0.4468,
      "step": 12238
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6926745176315308,
      "learning_rate": 1.4092408909710405e-05,
      "loss": 0.5684,
      "step": 12239
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5807424783706665,
      "learning_rate": 1.409147966390375e-05,
      "loss": 0.4223,
      "step": 12240
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.616301417350769,
      "learning_rate": 1.4090550375661907e-05,
      "loss": 0.4764,
      "step": 12241
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.713400959968567,
      "learning_rate": 1.4089621044994514e-05,
      "loss": 0.4576,
      "step": 12242
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.588667869567871,
      "learning_rate": 1.4088691671911215e-05,
      "loss": 0.5065,
      "step": 12243
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6222060918807983,
      "learning_rate": 1.4087762256421644e-05,
      "loss": 0.4658,
      "step": 12244
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5980308055877686,
      "learning_rate": 1.4086832798535442e-05,
      "loss": 0.503,
      "step": 12245
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5622608661651611,
      "learning_rate": 1.4085903298262244e-05,
      "loss": 0.4419,
      "step": 12246
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7865979671478271,
      "learning_rate": 1.4084973755611703e-05,
      "loss": 0.5198,
      "step": 12247
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5331748723983765,
      "learning_rate": 1.4084044170593445e-05,
      "loss": 0.4092,
      "step": 12248
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6056255102157593,
      "learning_rate": 1.4083114543217122e-05,
      "loss": 0.4806,
      "step": 12249
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4749614000320435,
      "learning_rate": 1.4082184873492373e-05,
      "loss": 0.4136,
      "step": 12250
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.731872797012329,
      "learning_rate": 1.4081255161428837e-05,
      "loss": 0.467,
      "step": 12251
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.665414810180664,
      "learning_rate": 1.408032540703616e-05,
      "loss": 0.4953,
      "step": 12252
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5137852430343628,
      "learning_rate": 1.4079395610323982e-05,
      "loss": 0.4671,
      "step": 12253
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5339329242706299,
      "learning_rate": 1.407846577130195e-05,
      "loss": 0.46,
      "step": 12254
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.546990156173706,
      "learning_rate": 1.4077535889979706e-05,
      "loss": 0.473,
      "step": 12255
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5934882164001465,
      "learning_rate": 1.407660596636689e-05,
      "loss": 0.4453,
      "step": 12256
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5884100198745728,
      "learning_rate": 1.4075676000473157e-05,
      "loss": 0.4911,
      "step": 12257
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5789605379104614,
      "learning_rate": 1.4074745992308143e-05,
      "loss": 0.4739,
      "step": 12258
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.669044852256775,
      "learning_rate": 1.40738159418815e-05,
      "loss": 0.4839,
      "step": 12259
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6033222675323486,
      "learning_rate": 1.4072885849202869e-05,
      "loss": 0.442,
      "step": 12260
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5242009162902832,
      "learning_rate": 1.4071955714281902e-05,
      "loss": 0.522,
      "step": 12261
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5256820917129517,
      "learning_rate": 1.407102553712824e-05,
      "loss": 0.4409,
      "step": 12262
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.521349310874939,
      "learning_rate": 1.4070095317751533e-05,
      "loss": 0.5011,
      "step": 12263
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5461467504501343,
      "learning_rate": 1.4069165056161432e-05,
      "loss": 0.4524,
      "step": 12264
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6794121265411377,
      "learning_rate": 1.4068234752367583e-05,
      "loss": 0.5022,
      "step": 12265
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5511704683303833,
      "learning_rate": 1.4067304406379628e-05,
      "loss": 0.4618,
      "step": 12266
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6128565073013306,
      "learning_rate": 1.4066374018207228e-05,
      "loss": 0.5081,
      "step": 12267
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6283835172653198,
      "learning_rate": 1.4065443587860024e-05,
      "loss": 0.4521,
      "step": 12268
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5641560554504395,
      "learning_rate": 1.406451311534767e-05,
      "loss": 0.4917,
      "step": 12269
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5953636169433594,
      "learning_rate": 1.4063582600679818e-05,
      "loss": 0.4599,
      "step": 12270
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6540080308914185,
      "learning_rate": 1.4062652043866112e-05,
      "loss": 0.5106,
      "step": 12271
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4978840351104736,
      "learning_rate": 1.4061721444916207e-05,
      "loss": 0.4349,
      "step": 12272
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5463981628417969,
      "learning_rate": 1.4060790803839763e-05,
      "loss": 0.4909,
      "step": 12273
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5936648845672607,
      "learning_rate": 1.4059860120646417e-05,
      "loss": 0.4523,
      "step": 12274
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7197535037994385,
      "learning_rate": 1.4058929395345838e-05,
      "loss": 0.4776,
      "step": 12275
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.675539255142212,
      "learning_rate": 1.4057998627947661e-05,
      "loss": 0.4578,
      "step": 12276
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6510242223739624,
      "learning_rate": 1.4057067818461555e-05,
      "loss": 0.492,
      "step": 12277
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5283254384994507,
      "learning_rate": 1.4056136966897167e-05,
      "loss": 0.4474,
      "step": 12278
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6373860836029053,
      "learning_rate": 1.4055206073264152e-05,
      "loss": 0.5417,
      "step": 12279
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.494288444519043,
      "learning_rate": 1.4054275137572166e-05,
      "loss": 0.417,
      "step": 12280
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6827577352523804,
      "learning_rate": 1.4053344159830863e-05,
      "loss": 0.5545,
      "step": 12281
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.555241346359253,
      "learning_rate": 1.4052413140049898e-05,
      "loss": 0.4436,
      "step": 12282
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.652172565460205,
      "learning_rate": 1.4051482078238932e-05,
      "loss": 0.505,
      "step": 12283
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5486719608306885,
      "learning_rate": 1.4050550974407618e-05,
      "loss": 0.4315,
      "step": 12284
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6741843223571777,
      "learning_rate": 1.4049619828565614e-05,
      "loss": 1.0379,
      "step": 12285
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6080982685089111,
      "learning_rate": 1.4048688640722573e-05,
      "loss": 1.0573,
      "step": 12286
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6421555280685425,
      "learning_rate": 1.4047757410888156e-05,
      "loss": 0.51,
      "step": 12287
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6319859027862549,
      "learning_rate": 1.4046826139072027e-05,
      "loss": 0.448,
      "step": 12288
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6434602737426758,
      "learning_rate": 1.4045894825283834e-05,
      "loss": 0.5276,
      "step": 12289
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5070643424987793,
      "learning_rate": 1.4044963469533247e-05,
      "loss": 0.4316,
      "step": 12290
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5021454095840454,
      "learning_rate": 1.4044032071829916e-05,
      "loss": 0.4587,
      "step": 12291
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5596219301223755,
      "learning_rate": 1.404310063218351e-05,
      "loss": 0.465,
      "step": 12292
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5329737663269043,
      "learning_rate": 1.4042169150603684e-05,
      "loss": 0.4847,
      "step": 12293
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5945608615875244,
      "learning_rate": 1.4041237627100099e-05,
      "loss": 0.4708,
      "step": 12294
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5714541673660278,
      "learning_rate": 1.4040306061682418e-05,
      "loss": 0.469,
      "step": 12295
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6177924871444702,
      "learning_rate": 1.4039374454360305e-05,
      "loss": 0.4552,
      "step": 12296
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6222891807556152,
      "learning_rate": 1.4038442805143413e-05,
      "loss": 0.5099,
      "step": 12297
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5854216814041138,
      "learning_rate": 1.4037511114041418e-05,
      "loss": 0.4603,
      "step": 12298
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6684801578521729,
      "learning_rate": 1.4036579381063973e-05,
      "loss": 0.535,
      "step": 12299
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5932790040969849,
      "learning_rate": 1.4035647606220745e-05,
      "loss": 0.4469,
      "step": 12300
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5202114582061768,
      "learning_rate": 1.4034715789521398e-05,
      "loss": 0.4668,
      "step": 12301
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5369935035705566,
      "learning_rate": 1.4033783930975599e-05,
      "loss": 0.4496,
      "step": 12302
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.563247561454773,
      "learning_rate": 1.403285203059301e-05,
      "loss": 0.4858,
      "step": 12303
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6210558414459229,
      "learning_rate": 1.4031920088383295e-05,
      "loss": 0.4563,
      "step": 12304
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6082102060317993,
      "learning_rate": 1.4030988104356118e-05,
      "loss": 0.4775,
      "step": 12305
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6055198907852173,
      "learning_rate": 1.4030056078521155e-05,
      "loss": 0.4391,
      "step": 12306
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6140518188476562,
      "learning_rate": 1.402912401088806e-05,
      "loss": 0.5067,
      "step": 12307
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5709717273712158,
      "learning_rate": 1.4028191901466511e-05,
      "loss": 0.4701,
      "step": 12308
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.673902988433838,
      "learning_rate": 1.4027259750266168e-05,
      "loss": 0.6122,
      "step": 12309
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5271130800247192,
      "learning_rate": 1.4026327557296702e-05,
      "loss": 0.4599,
      "step": 12310
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5897196531295776,
      "learning_rate": 1.4025395322567782e-05,
      "loss": 0.4431,
      "step": 12311
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5018800497055054,
      "learning_rate": 1.4024463046089073e-05,
      "loss": 0.4315,
      "step": 12312
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.512052059173584,
      "learning_rate": 1.402353072787025e-05,
      "loss": 0.5197,
      "step": 12313
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.532159686088562,
      "learning_rate": 1.4022598367920978e-05,
      "loss": 0.4366,
      "step": 12314
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7533957958221436,
      "learning_rate": 1.4021665966250928e-05,
      "loss": 0.5954,
      "step": 12315
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6094951629638672,
      "learning_rate": 1.4020733522869772e-05,
      "loss": 0.4681,
      "step": 12316
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.518660306930542,
      "learning_rate": 1.4019801037787177e-05,
      "loss": 0.5156,
      "step": 12317
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6817741394042969,
      "learning_rate": 1.401886851101282e-05,
      "loss": 0.4843,
      "step": 12318
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5911074876785278,
      "learning_rate": 1.4017935942556369e-05,
      "loss": 0.5433,
      "step": 12319
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6028624773025513,
      "learning_rate": 1.4017003332427499e-05,
      "loss": 0.4597,
      "step": 12320
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6823861598968506,
      "learning_rate": 1.4016070680635879e-05,
      "loss": 0.4881,
      "step": 12321
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6022650003433228,
      "learning_rate": 1.4015137987191186e-05,
      "loss": 0.456,
      "step": 12322
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5736513137817383,
      "learning_rate": 1.4014205252103086e-05,
      "loss": 0.53,
      "step": 12323
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5835286378860474,
      "learning_rate": 1.4013272475381265e-05,
      "loss": 0.4627,
      "step": 12324
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6445366144180298,
      "learning_rate": 1.4012339657035388e-05,
      "loss": 0.5069,
      "step": 12325
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.524844765663147,
      "learning_rate": 1.4011406797075133e-05,
      "loss": 0.4153,
      "step": 12326
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5985066890716553,
      "learning_rate": 1.4010473895510175e-05,
      "loss": 0.5271,
      "step": 12327
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5895649194717407,
      "learning_rate": 1.4009540952350191e-05,
      "loss": 0.4479,
      "step": 12328
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5999526977539062,
      "learning_rate": 1.4008607967604854e-05,
      "loss": 0.5137,
      "step": 12329
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.654034972190857,
      "learning_rate": 1.4007674941283842e-05,
      "loss": 0.4615,
      "step": 12330
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6021631956100464,
      "learning_rate": 1.4006741873396832e-05,
      "loss": 0.5029,
      "step": 12331
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6125783920288086,
      "learning_rate": 1.4005808763953505e-05,
      "loss": 0.4783,
      "step": 12332
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.563184380531311,
      "learning_rate": 1.4004875612963531e-05,
      "loss": 0.483,
      "step": 12333
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4896469116210938,
      "learning_rate": 1.4003942420436599e-05,
      "loss": 0.4323,
      "step": 12334
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.593173623085022,
      "learning_rate": 1.4003009186382374e-05,
      "loss": 0.4864,
      "step": 12335
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6145877838134766,
      "learning_rate": 1.4002075910810546e-05,
      "loss": 0.4833,
      "step": 12336
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5252087116241455,
      "learning_rate": 1.4001142593730792e-05,
      "loss": 0.4815,
      "step": 12337
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4665817022323608,
      "learning_rate": 1.400020923515279e-05,
      "loss": 0.4271,
      "step": 12338
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6066327095031738,
      "learning_rate": 1.3999275835086223e-05,
      "loss": 0.5054,
      "step": 12339
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5681930780410767,
      "learning_rate": 1.399834239354077e-05,
      "loss": 0.462,
      "step": 12340
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6467132568359375,
      "learning_rate": 1.399740891052611e-05,
      "loss": 0.4674,
      "step": 12341
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5370080471038818,
      "learning_rate": 1.399647538605193e-05,
      "loss": 0.4157,
      "step": 12342
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.705670714378357,
      "learning_rate": 1.3995541820127908e-05,
      "loss": 0.5041,
      "step": 12343
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5848203897476196,
      "learning_rate": 1.399460821276373e-05,
      "loss": 0.4372,
      "step": 12344
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5018737316131592,
      "learning_rate": 1.3993674563969078e-05,
      "loss": 0.4743,
      "step": 12345
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5231797695159912,
      "learning_rate": 1.3992740873753632e-05,
      "loss": 0.4427,
      "step": 12346
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5948349237442017,
      "learning_rate": 1.3991807142127082e-05,
      "loss": 0.5245,
      "step": 12347
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5545381307601929,
      "learning_rate": 1.3990873369099104e-05,
      "loss": 0.4481,
      "step": 12348
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5014121532440186,
      "learning_rate": 1.3989939554679388e-05,
      "loss": 0.5347,
      "step": 12349
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4275949001312256,
      "learning_rate": 1.3989005698877623e-05,
      "loss": 0.412,
      "step": 12350
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.568793535232544,
      "learning_rate": 1.3988071801703484e-05,
      "loss": 0.4953,
      "step": 12351
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5774015188217163,
      "learning_rate": 1.3987137863166668e-05,
      "loss": 0.4802,
      "step": 12352
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5748028755187988,
      "learning_rate": 1.3986203883276855e-05,
      "loss": 0.4766,
      "step": 12353
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.609178900718689,
      "learning_rate": 1.3985269862043733e-05,
      "loss": 0.4817,
      "step": 12354
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5480616092681885,
      "learning_rate": 1.3984335799476989e-05,
      "loss": 0.4624,
      "step": 12355
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5831170082092285,
      "learning_rate": 1.3983401695586314e-05,
      "loss": 0.4432,
      "step": 12356
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6214375495910645,
      "learning_rate": 1.398246755038139e-05,
      "loss": 0.5281,
      "step": 12357
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5636171102523804,
      "learning_rate": 1.3981533363871914e-05,
      "loss": 0.4162,
      "step": 12358
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6261495351791382,
      "learning_rate": 1.3980599136067566e-05,
      "loss": 0.4957,
      "step": 12359
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5209989547729492,
      "learning_rate": 1.3979664866978045e-05,
      "loss": 0.4582,
      "step": 12360
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6087522506713867,
      "learning_rate": 1.397873055661303e-05,
      "loss": 0.5087,
      "step": 12361
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.676353931427002,
      "learning_rate": 1.3977796204982216e-05,
      "loss": 0.4476,
      "step": 12362
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5565944910049438,
      "learning_rate": 1.39768618120953e-05,
      "loss": 0.4712,
      "step": 12363
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.595917820930481,
      "learning_rate": 1.3975927377961964e-05,
      "loss": 0.4827,
      "step": 12364
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5150973796844482,
      "learning_rate": 1.3974992902591904e-05,
      "loss": 0.4847,
      "step": 12365
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7449572086334229,
      "learning_rate": 1.3974058385994814e-05,
      "loss": 0.4485,
      "step": 12366
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.506508708000183,
      "learning_rate": 1.3973123828180377e-05,
      "loss": 0.475,
      "step": 12367
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5287123918533325,
      "learning_rate": 1.3972189229158299e-05,
      "loss": 0.4186,
      "step": 12368
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6035003662109375,
      "learning_rate": 1.3971254588938263e-05,
      "loss": 0.4763,
      "step": 12369
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5446040630340576,
      "learning_rate": 1.3970319907529973e-05,
      "loss": 0.4704,
      "step": 12370
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.606540560722351,
      "learning_rate": 1.3969385184943108e-05,
      "loss": 0.4723,
      "step": 12371
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.681877613067627,
      "learning_rate": 1.3968450421187375e-05,
      "loss": 0.4505,
      "step": 12372
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5731858015060425,
      "learning_rate": 1.3967515616272467e-05,
      "loss": 0.5176,
      "step": 12373
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8417950868606567,
      "learning_rate": 1.3966580770208075e-05,
      "loss": 0.465,
      "step": 12374
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6572340726852417,
      "learning_rate": 1.39656458830039e-05,
      "loss": 0.496,
      "step": 12375
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5597853660583496,
      "learning_rate": 1.3964710954669633e-05,
      "loss": 0.4227,
      "step": 12376
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.629324197769165,
      "learning_rate": 1.3963775985214973e-05,
      "loss": 0.5356,
      "step": 12377
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.548472285270691,
      "learning_rate": 1.3962840974649624e-05,
      "loss": 0.4302,
      "step": 12378
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6301909685134888,
      "learning_rate": 1.3961905922983269e-05,
      "loss": 0.5302,
      "step": 12379
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5956460237503052,
      "learning_rate": 1.396097083022562e-05,
      "loss": 0.4907,
      "step": 12380
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5848945379257202,
      "learning_rate": 1.3960035696386367e-05,
      "loss": 0.5372,
      "step": 12381
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.516325831413269,
      "learning_rate": 1.395910052147521e-05,
      "loss": 0.4427,
      "step": 12382
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5497075319290161,
      "learning_rate": 1.3958165305501852e-05,
      "loss": 0.4819,
      "step": 12383
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5747754573822021,
      "learning_rate": 1.3957230048475992e-05,
      "loss": 0.4431,
      "step": 12384
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.669982671737671,
      "learning_rate": 1.3956294750407322e-05,
      "loss": 0.5115,
      "step": 12385
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6658817529678345,
      "learning_rate": 1.3955359411305555e-05,
      "loss": 0.4385,
      "step": 12386
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.632232427597046,
      "learning_rate": 1.3954424031180384e-05,
      "loss": 0.5493,
      "step": 12387
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.512006163597107,
      "learning_rate": 1.3953488610041514e-05,
      "loss": 0.4439,
      "step": 12388
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5795519351959229,
      "learning_rate": 1.3952553147898645e-05,
      "loss": 0.5173,
      "step": 12389
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5499297380447388,
      "learning_rate": 1.3951617644761476e-05,
      "loss": 0.4394,
      "step": 12390
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6389960050582886,
      "learning_rate": 1.3950682100639716e-05,
      "loss": 0.4821,
      "step": 12391
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5080127716064453,
      "learning_rate": 1.3949746515543065e-05,
      "loss": 0.4286,
      "step": 12392
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6011067628860474,
      "learning_rate": 1.3948810889481226e-05,
      "loss": 0.5548,
      "step": 12393
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.529288411140442,
      "learning_rate": 1.3947875222463904e-05,
      "loss": 0.4113,
      "step": 12394
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6461284160614014,
      "learning_rate": 1.39469395145008e-05,
      "loss": 0.4614,
      "step": 12395
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6365513801574707,
      "learning_rate": 1.3946003765601627e-05,
      "loss": 0.4862,
      "step": 12396
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6275609731674194,
      "learning_rate": 1.3945067975776081e-05,
      "loss": 0.487,
      "step": 12397
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5160021781921387,
      "learning_rate": 1.3944132145033872e-05,
      "loss": 0.4284,
      "step": 12398
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7164877653121948,
      "learning_rate": 1.394319627338471e-05,
      "loss": 0.4829,
      "step": 12399
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6198536157608032,
      "learning_rate": 1.3942260360838293e-05,
      "loss": 0.454,
      "step": 12400
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6315357685089111,
      "learning_rate": 1.3941324407404334e-05,
      "loss": 0.5108,
      "step": 12401
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.519229531288147,
      "learning_rate": 1.3940388413092536e-05,
      "loss": 0.4325,
      "step": 12402
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6266660690307617,
      "learning_rate": 1.393945237791261e-05,
      "loss": 0.5113,
      "step": 12403
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.56618070602417,
      "learning_rate": 1.3938516301874266e-05,
      "loss": 0.4455,
      "step": 12404
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6393910646438599,
      "learning_rate": 1.3937580184987205e-05,
      "loss": 0.5608,
      "step": 12405
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5892714262008667,
      "learning_rate": 1.3936644027261144e-05,
      "loss": 0.4321,
      "step": 12406
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5816327333450317,
      "learning_rate": 1.3935707828705789e-05,
      "loss": 0.4877,
      "step": 12407
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8328555822372437,
      "learning_rate": 1.393477158933085e-05,
      "loss": 0.485,
      "step": 12408
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.738529086112976,
      "learning_rate": 1.3933835309146039e-05,
      "loss": 0.5122,
      "step": 12409
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5433028936386108,
      "learning_rate": 1.3932898988161065e-05,
      "loss": 0.431,
      "step": 12410
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5606977939605713,
      "learning_rate": 1.393196262638564e-05,
      "loss": 0.5019,
      "step": 12411
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5580799579620361,
      "learning_rate": 1.3931026223829473e-05,
      "loss": 0.4769,
      "step": 12412
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6680644750595093,
      "learning_rate": 1.3930089780502276e-05,
      "loss": 0.514,
      "step": 12413
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.511857509613037,
      "learning_rate": 1.392915329641377e-05,
      "loss": 0.4336,
      "step": 12414
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5302575826644897,
      "learning_rate": 1.3928216771573655e-05,
      "loss": 0.498,
      "step": 12415
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6019132137298584,
      "learning_rate": 1.3927280205991655e-05,
      "loss": 0.5125,
      "step": 12416
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5983607769012451,
      "learning_rate": 1.3926343599677477e-05,
      "loss": 0.5018,
      "step": 12417
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6314016580581665,
      "learning_rate": 1.392540695264084e-05,
      "loss": 0.4792,
      "step": 12418
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7439128160476685,
      "learning_rate": 1.3924470264891451e-05,
      "loss": 0.5068,
      "step": 12419
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5638760328292847,
      "learning_rate": 1.3923533536439034e-05,
      "loss": 0.4651,
      "step": 12420
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5562812089920044,
      "learning_rate": 1.3922596767293296e-05,
      "loss": 0.4913,
      "step": 12421
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5631403923034668,
      "learning_rate": 1.3921659957463962e-05,
      "loss": 0.4206,
      "step": 12422
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5221349000930786,
      "learning_rate": 1.3920723106960739e-05,
      "loss": 0.4805,
      "step": 12423
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5487070083618164,
      "learning_rate": 1.391978621579335e-05,
      "loss": 0.4124,
      "step": 12424
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.624772548675537,
      "learning_rate": 1.391884928397151e-05,
      "loss": 0.5102,
      "step": 12425
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5411632061004639,
      "learning_rate": 1.3917912311504935e-05,
      "loss": 0.4728,
      "step": 12426
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6010197401046753,
      "learning_rate": 1.3916975298403345e-05,
      "loss": 0.5041,
      "step": 12427
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5560617446899414,
      "learning_rate": 1.3916038244676457e-05,
      "loss": 0.4945,
      "step": 12428
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5883548259735107,
      "learning_rate": 1.3915101150333992e-05,
      "loss": 0.5257,
      "step": 12429
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4685044288635254,
      "learning_rate": 1.3914164015385667e-05,
      "loss": 0.438,
      "step": 12430
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8936662673950195,
      "learning_rate": 1.39132268398412e-05,
      "loss": 1.1135,
      "step": 12431
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6396235227584839,
      "learning_rate": 1.3912289623710318e-05,
      "loss": 1.0232,
      "step": 12432
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.716149926185608,
      "learning_rate": 1.3911352367002731e-05,
      "loss": 0.4911,
      "step": 12433
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5620477199554443,
      "learning_rate": 1.3910415069728166e-05,
      "loss": 0.4731,
      "step": 12434
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7320884466171265,
      "learning_rate": 1.3909477731896348e-05,
      "loss": 0.509,
      "step": 12435
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5552656650543213,
      "learning_rate": 1.390854035351699e-05,
      "loss": 0.4414,
      "step": 12436
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.548174262046814,
      "learning_rate": 1.3907602934599822e-05,
      "loss": 0.506,
      "step": 12437
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6267200708389282,
      "learning_rate": 1.3906665475154561e-05,
      "loss": 0.4765,
      "step": 12438
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5468462705612183,
      "learning_rate": 1.390572797519093e-05,
      "loss": 0.4759,
      "step": 12439
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.537168264389038,
      "learning_rate": 1.3904790434718658e-05,
      "loss": 0.4453,
      "step": 12440
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5788681507110596,
      "learning_rate": 1.3903852853747462e-05,
      "loss": 0.5413,
      "step": 12441
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5444538593292236,
      "learning_rate": 1.3902915232287071e-05,
      "loss": 0.4436,
      "step": 12442
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6765718460083008,
      "learning_rate": 1.390197757034721e-05,
      "loss": 0.4862,
      "step": 12443
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6121342182159424,
      "learning_rate": 1.3901039867937598e-05,
      "loss": 0.4554,
      "step": 12444
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5343364477157593,
      "learning_rate": 1.3900102125067967e-05,
      "loss": 0.4865,
      "step": 12445
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5116389989852905,
      "learning_rate": 1.3899164341748038e-05,
      "loss": 0.4309,
      "step": 12446
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6419767141342163,
      "learning_rate": 1.3898226517987544e-05,
      "loss": 0.5368,
      "step": 12447
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.531062126159668,
      "learning_rate": 1.3897288653796204e-05,
      "loss": 0.4484,
      "step": 12448
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5830789804458618,
      "learning_rate": 1.389635074918375e-05,
      "loss": 0.5399,
      "step": 12449
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5500811338424683,
      "learning_rate": 1.3895412804159909e-05,
      "loss": 0.4594,
      "step": 12450
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5243390798568726,
      "learning_rate": 1.3894474818734404e-05,
      "loss": 0.4717,
      "step": 12451
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4917041063308716,
      "learning_rate": 1.3893536792916971e-05,
      "loss": 0.4306,
      "step": 12452
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6432342529296875,
      "learning_rate": 1.3892598726717334e-05,
      "loss": 0.5004,
      "step": 12453
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.573836088180542,
      "learning_rate": 1.3891660620145225e-05,
      "loss": 0.4302,
      "step": 12454
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5456091165542603,
      "learning_rate": 1.3890722473210371e-05,
      "loss": 0.4949,
      "step": 12455
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5565351247787476,
      "learning_rate": 1.3889784285922505e-05,
      "loss": 0.4798,
      "step": 12456
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5185914039611816,
      "learning_rate": 1.3888846058291351e-05,
      "loss": 0.5079,
      "step": 12457
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5429435968399048,
      "learning_rate": 1.3887907790326652e-05,
      "loss": 0.4513,
      "step": 12458
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.63723623752594,
      "learning_rate": 1.3886969482038127e-05,
      "loss": 0.521,
      "step": 12459
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6262990236282349,
      "learning_rate": 1.3886031133435513e-05,
      "loss": 0.4901,
      "step": 12460
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5375182628631592,
      "learning_rate": 1.3885092744528545e-05,
      "loss": 0.4878,
      "step": 12461
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5614343881607056,
      "learning_rate": 1.3884154315326952e-05,
      "loss": 0.4604,
      "step": 12462
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.588869333267212,
      "learning_rate": 1.3883215845840468e-05,
      "loss": 0.5344,
      "step": 12463
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4806013107299805,
      "learning_rate": 1.3882277336078824e-05,
      "loss": 0.4339,
      "step": 12464
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5421342849731445,
      "learning_rate": 1.3881338786051757e-05,
      "loss": 0.4598,
      "step": 12465
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5100725889205933,
      "learning_rate": 1.3880400195769001e-05,
      "loss": 0.4353,
      "step": 12466
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5517871379852295,
      "learning_rate": 1.3879461565240288e-05,
      "loss": 0.5483,
      "step": 12467
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5621024370193481,
      "learning_rate": 1.3878522894475359e-05,
      "loss": 0.4439,
      "step": 12468
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6963001489639282,
      "learning_rate": 1.387758418348394e-05,
      "loss": 0.5647,
      "step": 12469
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.552888035774231,
      "learning_rate": 1.3876645432275775e-05,
      "loss": 0.4425,
      "step": 12470
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5790518522262573,
      "learning_rate": 1.3875706640860598e-05,
      "loss": 0.5054,
      "step": 12471
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.596396565437317,
      "learning_rate": 1.3874767809248147e-05,
      "loss": 0.4519,
      "step": 12472
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6719880104064941,
      "learning_rate": 1.3873828937448155e-05,
      "loss": 0.4723,
      "step": 12473
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.577377438545227,
      "learning_rate": 1.3872890025470365e-05,
      "loss": 0.4542,
      "step": 12474
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5794681310653687,
      "learning_rate": 1.3871951073324508e-05,
      "loss": 0.4501,
      "step": 12475
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5306508541107178,
      "learning_rate": 1.3871012081020332e-05,
      "loss": 0.4311,
      "step": 12476
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.532080888748169,
      "learning_rate": 1.3870073048567567e-05,
      "loss": 0.4783,
      "step": 12477
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6404860019683838,
      "learning_rate": 1.3869133975975957e-05,
      "loss": 0.4453,
      "step": 12478
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.591270923614502,
      "learning_rate": 1.3868194863255242e-05,
      "loss": 0.4954,
      "step": 12479
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5352256298065186,
      "learning_rate": 1.386725571041516e-05,
      "loss": 0.4533,
      "step": 12480
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6025394201278687,
      "learning_rate": 1.3866316517465452e-05,
      "loss": 0.491,
      "step": 12481
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.584324836730957,
      "learning_rate": 1.3865377284415859e-05,
      "loss": 0.4362,
      "step": 12482
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5339438915252686,
      "learning_rate": 1.386443801127612e-05,
      "loss": 0.4948,
      "step": 12483
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5968996286392212,
      "learning_rate": 1.3863498698055984e-05,
      "loss": 0.4659,
      "step": 12484
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6779296398162842,
      "learning_rate": 1.3862559344765186e-05,
      "loss": 0.5307,
      "step": 12485
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.707871437072754,
      "learning_rate": 1.3861619951413471e-05,
      "loss": 0.4558,
      "step": 12486
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.664189338684082,
      "learning_rate": 1.3860680518010583e-05,
      "loss": 1.0022,
      "step": 12487
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5753169059753418,
      "learning_rate": 1.3859741044566263e-05,
      "loss": 1.038,
      "step": 12488
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.658463478088379,
      "learning_rate": 1.3858801531090257e-05,
      "loss": 0.5309,
      "step": 12489
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.70717191696167,
      "learning_rate": 1.3857861977592308e-05,
      "loss": 0.4918,
      "step": 12490
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6168323755264282,
      "learning_rate": 1.3856922384082165e-05,
      "loss": 0.5177,
      "step": 12491
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.680673599243164,
      "learning_rate": 1.3855982750569566e-05,
      "loss": 0.4702,
      "step": 12492
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5817115306854248,
      "learning_rate": 1.3855043077064258e-05,
      "loss": 0.4967,
      "step": 12493
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.593544840812683,
      "learning_rate": 1.3854103363575994e-05,
      "loss": 0.4339,
      "step": 12494
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6346163749694824,
      "learning_rate": 1.385316361011451e-05,
      "loss": 0.5102,
      "step": 12495
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4732571840286255,
      "learning_rate": 1.3852223816689558e-05,
      "loss": 0.4238,
      "step": 12496
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6860077381134033,
      "learning_rate": 1.385128398331089e-05,
      "loss": 0.5082,
      "step": 12497
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6059848070144653,
      "learning_rate": 1.3850344109988245e-05,
      "loss": 0.4781,
      "step": 12498
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5443625450134277,
      "learning_rate": 1.3849404196731375e-05,
      "loss": 0.5028,
      "step": 12499
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7026498317718506,
      "learning_rate": 1.3848464243550029e-05,
      "loss": 0.4407,
      "step": 12500
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6143250465393066,
      "learning_rate": 1.3847524250453953e-05,
      "loss": 0.4748,
      "step": 12501
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5770983695983887,
      "learning_rate": 1.3846584217452899e-05,
      "loss": 0.4258,
      "step": 12502
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.001875877380371,
      "learning_rate": 1.3845644144556613e-05,
      "loss": 0.9829,
      "step": 12503
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8834733963012695,
      "learning_rate": 1.3844704031774854e-05,
      "loss": 1.0961,
      "step": 12504
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7045873403549194,
      "learning_rate": 1.384376387911736e-05,
      "loss": 0.484,
      "step": 12505
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7531611919403076,
      "learning_rate": 1.384282368659389e-05,
      "loss": 0.4524,
      "step": 12506
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.690773367881775,
      "learning_rate": 1.3841883454214195e-05,
      "loss": 0.5303,
      "step": 12507
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5377177000045776,
      "learning_rate": 1.3840943181988025e-05,
      "loss": 0.4313,
      "step": 12508
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6091539859771729,
      "learning_rate": 1.384000286992513e-05,
      "loss": 0.5032,
      "step": 12509
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6226043701171875,
      "learning_rate": 1.3839062518035266e-05,
      "loss": 0.4414,
      "step": 12510
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6031017303466797,
      "learning_rate": 1.3838122126328183e-05,
      "loss": 0.4737,
      "step": 12511
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6126097440719604,
      "learning_rate": 1.3837181694813642e-05,
      "loss": 0.4599,
      "step": 12512
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5359447002410889,
      "learning_rate": 1.3836241223501383e-05,
      "loss": 0.4815,
      "step": 12513
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6116647720336914,
      "learning_rate": 1.3835300712401175e-05,
      "loss": 0.4752,
      "step": 12514
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5564074516296387,
      "learning_rate": 1.3834360161522759e-05,
      "loss": 0.4892,
      "step": 12515
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4939230680465698,
      "learning_rate": 1.38334195708759e-05,
      "loss": 0.4296,
      "step": 12516
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6276381015777588,
      "learning_rate": 1.3832478940470351e-05,
      "loss": 0.5223,
      "step": 12517
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.605474591255188,
      "learning_rate": 1.3831538270315868e-05,
      "loss": 0.4628,
      "step": 12518
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.628320336341858,
      "learning_rate": 1.3830597560422203e-05,
      "loss": 0.4915,
      "step": 12519
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6213306188583374,
      "learning_rate": 1.3829656810799118e-05,
      "loss": 0.4383,
      "step": 12520
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6617857217788696,
      "learning_rate": 1.3828716021456365e-05,
      "loss": 0.4939,
      "step": 12521
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.516392469406128,
      "learning_rate": 1.382777519240371e-05,
      "loss": 0.4487,
      "step": 12522
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.648777961730957,
      "learning_rate": 1.3826834323650899e-05,
      "loss": 0.5403,
      "step": 12523
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6485308408737183,
      "learning_rate": 1.3825893415207699e-05,
      "loss": 0.4512,
      "step": 12524
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6310157775878906,
      "learning_rate": 1.3824952467083867e-05,
      "loss": 0.4602,
      "step": 12525
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5030025243759155,
      "learning_rate": 1.3824011479289162e-05,
      "loss": 0.4267,
      "step": 12526
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6011847257614136,
      "learning_rate": 1.3823070451833342e-05,
      "loss": 0.4871,
      "step": 12527
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5926002264022827,
      "learning_rate": 1.3822129384726168e-05,
      "loss": 0.4294,
      "step": 12528
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6594444513320923,
      "learning_rate": 1.38211882779774e-05,
      "loss": 0.5419,
      "step": 12529
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5861555337905884,
      "learning_rate": 1.3820247131596803e-05,
      "loss": 0.4345,
      "step": 12530
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6245454549789429,
      "learning_rate": 1.381930594559413e-05,
      "loss": 0.494,
      "step": 12531
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.544677495956421,
      "learning_rate": 1.381836471997915e-05,
      "loss": 0.4516,
      "step": 12532
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.713578462600708,
      "learning_rate": 1.3817423454761619e-05,
      "loss": 0.5556,
      "step": 12533
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6425641775131226,
      "learning_rate": 1.3816482149951303e-05,
      "loss": 0.4354,
      "step": 12534
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6109728813171387,
      "learning_rate": 1.3815540805557965e-05,
      "loss": 0.501,
      "step": 12535
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5879474878311157,
      "learning_rate": 1.3814599421591366e-05,
      "loss": 0.4368,
      "step": 12536
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5811268091201782,
      "learning_rate": 1.3813657998061273e-05,
      "loss": 0.5159,
      "step": 12537
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5313645601272583,
      "learning_rate": 1.3812716534977449e-05,
      "loss": 0.4584,
      "step": 12538
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5921727418899536,
      "learning_rate": 1.3811775032349654e-05,
      "loss": 0.5052,
      "step": 12539
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5043176412582397,
      "learning_rate": 1.381083349018766e-05,
      "loss": 0.4168,
      "step": 12540
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6197571754455566,
      "learning_rate": 1.3809891908501229e-05,
      "loss": 0.5186,
      "step": 12541
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6007863283157349,
      "learning_rate": 1.3808950287300124e-05,
      "loss": 0.4625,
      "step": 12542
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5692298412322998,
      "learning_rate": 1.3808008626594118e-05,
      "loss": 0.5041,
      "step": 12543
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5770037174224854,
      "learning_rate": 1.3807066926392967e-05,
      "loss": 0.464,
      "step": 12544
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5660213232040405,
      "learning_rate": 1.3806125186706451e-05,
      "loss": 0.4598,
      "step": 12545
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.635656476020813,
      "learning_rate": 1.3805183407544325e-05,
      "loss": 0.4946,
      "step": 12546
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5672059059143066,
      "learning_rate": 1.3804241588916365e-05,
      "loss": 0.5061,
      "step": 12547
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5956212282180786,
      "learning_rate": 1.3803299730832335e-05,
      "loss": 0.5038,
      "step": 12548
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5876624584197998,
      "learning_rate": 1.3802357833302007e-05,
      "loss": 0.5927,
      "step": 12549
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5637993812561035,
      "learning_rate": 1.3801415896335146e-05,
      "loss": 0.4372,
      "step": 12550
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6798782348632812,
      "learning_rate": 1.3800473919941526e-05,
      "loss": 0.6339,
      "step": 12551
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5444839000701904,
      "learning_rate": 1.379953190413091e-05,
      "loss": 0.4292,
      "step": 12552
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7987314462661743,
      "learning_rate": 1.3798589848913078e-05,
      "loss": 0.4907,
      "step": 12553
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5575886964797974,
      "learning_rate": 1.379764775429779e-05,
      "loss": 0.4315,
      "step": 12554
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6749882698059082,
      "learning_rate": 1.3796705620294824e-05,
      "loss": 0.4842,
      "step": 12555
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6197057962417603,
      "learning_rate": 1.3795763446913948e-05,
      "loss": 0.4725,
      "step": 12556
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5875544548034668,
      "learning_rate": 1.3794821234164936e-05,
      "loss": 0.5445,
      "step": 12557
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5580586194992065,
      "learning_rate": 1.3793878982057563e-05,
      "loss": 0.424,
      "step": 12558
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6298798322677612,
      "learning_rate": 1.3792936690601595e-05,
      "loss": 0.4683,
      "step": 12559
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6077693700790405,
      "learning_rate": 1.3791994359806805e-05,
      "loss": 0.4618,
      "step": 12560
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.633677363395691,
      "learning_rate": 1.3791051989682977e-05,
      "loss": 0.5221,
      "step": 12561
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5729585886001587,
      "learning_rate": 1.3790109580239873e-05,
      "loss": 0.4758,
      "step": 12562
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5548062324523926,
      "learning_rate": 1.3789167131487276e-05,
      "loss": 0.4791,
      "step": 12563
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6128504276275635,
      "learning_rate": 1.3788224643434952e-05,
      "loss": 0.465,
      "step": 12564
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7570589780807495,
      "learning_rate": 1.3787282116092682e-05,
      "loss": 0.5323,
      "step": 12565
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6247665882110596,
      "learning_rate": 1.378633954947024e-05,
      "loss": 0.4367,
      "step": 12566
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4789280891418457,
      "learning_rate": 1.3785396943577404e-05,
      "loss": 0.4862,
      "step": 12567
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6332457065582275,
      "learning_rate": 1.3784454298423948e-05,
      "loss": 0.4773,
      "step": 12568
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6421914100646973,
      "learning_rate": 1.378351161401965e-05,
      "loss": 0.5174,
      "step": 12569
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.562845230102539,
      "learning_rate": 1.3782568890374281e-05,
      "loss": 0.4412,
      "step": 12570
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.685412883758545,
      "learning_rate": 1.3781626127497632e-05,
      "loss": 0.4844,
      "step": 12571
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5293962955474854,
      "learning_rate": 1.3780683325399467e-05,
      "loss": 0.4782,
      "step": 12572
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5397106409072876,
      "learning_rate": 1.3779740484089574e-05,
      "loss": 0.5465,
      "step": 12573
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5916661024093628,
      "learning_rate": 1.3778797603577727e-05,
      "loss": 0.4596,
      "step": 12574
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6561648845672607,
      "learning_rate": 1.3777854683873707e-05,
      "loss": 0.5745,
      "step": 12575
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5031934976577759,
      "learning_rate": 1.377691172498729e-05,
      "loss": 0.4165,
      "step": 12576
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6334017515182495,
      "learning_rate": 1.3775968726928262e-05,
      "loss": 0.4747,
      "step": 12577
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5098539590835571,
      "learning_rate": 1.3775025689706396e-05,
      "loss": 0.3971,
      "step": 12578
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6032642126083374,
      "learning_rate": 1.3774082613331482e-05,
      "loss": 0.5128,
      "step": 12579
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5258865356445312,
      "learning_rate": 1.3773139497813293e-05,
      "loss": 0.4578,
      "step": 12580
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6393564939498901,
      "learning_rate": 1.3772196343161616e-05,
      "loss": 0.5416,
      "step": 12581
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.614806056022644,
      "learning_rate": 1.3771253149386228e-05,
      "loss": 0.4336,
      "step": 12582
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.612534523010254,
      "learning_rate": 1.3770309916496915e-05,
      "loss": 0.502,
      "step": 12583
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5463510751724243,
      "learning_rate": 1.376936664450346e-05,
      "loss": 0.435,
      "step": 12584
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6172138452529907,
      "learning_rate": 1.3768423333415648e-05,
      "loss": 0.532,
      "step": 12585
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5512428283691406,
      "learning_rate": 1.3767479983243258e-05,
      "loss": 0.4379,
      "step": 12586
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6405975818634033,
      "learning_rate": 1.3766536593996075e-05,
      "loss": 0.5314,
      "step": 12587
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5442155599594116,
      "learning_rate": 1.3765593165683883e-05,
      "loss": 0.4311,
      "step": 12588
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6566611528396606,
      "learning_rate": 1.3764649698316472e-05,
      "loss": 0.5288,
      "step": 12589
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5990002155303955,
      "learning_rate": 1.376370619190362e-05,
      "loss": 0.4469,
      "step": 12590
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7301011085510254,
      "learning_rate": 1.3762762646455121e-05,
      "loss": 0.498,
      "step": 12591
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5608471632003784,
      "learning_rate": 1.3761819061980753e-05,
      "loss": 0.4629,
      "step": 12592
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5642520189285278,
      "learning_rate": 1.3760875438490306e-05,
      "loss": 0.5158,
      "step": 12593
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6550452709197998,
      "learning_rate": 1.375993177599357e-05,
      "loss": 0.4521,
      "step": 12594
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6419318914413452,
      "learning_rate": 1.3758988074500326e-05,
      "loss": 0.4957,
      "step": 12595
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.527600646018982,
      "learning_rate": 1.3758044334020361e-05,
      "loss": 0.4163,
      "step": 12596
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5729745626449585,
      "learning_rate": 1.3757100554563476e-05,
      "loss": 0.4974,
      "step": 12597
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5932363271713257,
      "learning_rate": 1.3756156736139443e-05,
      "loss": 0.444,
      "step": 12598
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6354917287826538,
      "learning_rate": 1.375521287875806e-05,
      "loss": 0.5238,
      "step": 12599
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.514987826347351,
      "learning_rate": 1.3754268982429118e-05,
      "loss": 0.4244,
      "step": 12600
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6276510953903198,
      "learning_rate": 1.3753325047162397e-05,
      "loss": 0.4985,
      "step": 12601
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6040042638778687,
      "learning_rate": 1.37523810729677e-05,
      "loss": 0.4534,
      "step": 12602
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5659587383270264,
      "learning_rate": 1.3751437059854809e-05,
      "loss": 0.4797,
      "step": 12603
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6351170539855957,
      "learning_rate": 1.3750493007833514e-05,
      "loss": 0.4599,
      "step": 12604
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5419238805770874,
      "learning_rate": 1.3749548916913612e-05,
      "loss": 0.5539,
      "step": 12605
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.467668890953064,
      "learning_rate": 1.3748604787104889e-05,
      "loss": 0.4069,
      "step": 12606
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6309233903884888,
      "learning_rate": 1.3747660618417143e-05,
      "loss": 0.5,
      "step": 12607
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5913556814193726,
      "learning_rate": 1.3746716410860162e-05,
      "loss": 0.4316,
      "step": 12608
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7272496223449707,
      "learning_rate": 1.374577216444374e-05,
      "loss": 0.5246,
      "step": 12609
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7464748620986938,
      "learning_rate": 1.3744827879177671e-05,
      "loss": 0.4805,
      "step": 12610
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7063449621200562,
      "learning_rate": 1.374388355507175e-05,
      "loss": 0.4945,
      "step": 12611
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.608577847480774,
      "learning_rate": 1.3742939192135768e-05,
      "loss": 0.4395,
      "step": 12612
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5302820205688477,
      "learning_rate": 1.3741994790379525e-05,
      "loss": 0.4484,
      "step": 12613
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7031997442245483,
      "learning_rate": 1.3741050349812808e-05,
      "loss": 0.5308,
      "step": 12614
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.583875060081482,
      "learning_rate": 1.3740105870445418e-05,
      "loss": 0.4484,
      "step": 12615
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.516624093055725,
      "learning_rate": 1.373916135228715e-05,
      "loss": 0.4705,
      "step": 12616
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5448189973831177,
      "learning_rate": 1.3738216795347799e-05,
      "loss": 0.4466,
      "step": 12617
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.596504807472229,
      "learning_rate": 1.3737272199637165e-05,
      "loss": 0.469,
      "step": 12618
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.560299277305603,
      "learning_rate": 1.3736327565165038e-05,
      "loss": 0.5062,
      "step": 12619
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.530017614364624,
      "learning_rate": 1.3735382891941221e-05,
      "loss": 0.4424,
      "step": 12620
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0695338249206543,
      "learning_rate": 1.3734438179975512e-05,
      "loss": 1.0413,
      "step": 12621
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6891010999679565,
      "learning_rate": 1.3733493429277704e-05,
      "loss": 1.1141,
      "step": 12622
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6351178884506226,
      "learning_rate": 1.3732548639857602e-05,
      "loss": 0.5271,
      "step": 12623
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6777585744857788,
      "learning_rate": 1.3731603811724998e-05,
      "loss": 0.4487,
      "step": 12624
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6416093111038208,
      "learning_rate": 1.3730658944889699e-05,
      "loss": 0.4955,
      "step": 12625
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5747013092041016,
      "learning_rate": 1.3729714039361499e-05,
      "loss": 0.4241,
      "step": 12626
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4964311122894287,
      "learning_rate": 1.3728769095150201e-05,
      "loss": 0.4966,
      "step": 12627
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5167768001556396,
      "learning_rate": 1.3727824112265604e-05,
      "loss": 0.4713,
      "step": 12628
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.563107967376709,
      "learning_rate": 1.3726879090717513e-05,
      "loss": 0.494,
      "step": 12629
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6180031299591064,
      "learning_rate": 1.3725934030515724e-05,
      "loss": 0.4995,
      "step": 12630
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5418541431427002,
      "learning_rate": 1.3724988931670042e-05,
      "loss": 0.4698,
      "step": 12631
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6820200681686401,
      "learning_rate": 1.3724043794190265e-05,
      "loss": 0.4903,
      "step": 12632
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7269362211227417,
      "learning_rate": 1.3723098618086203e-05,
      "loss": 0.4767,
      "step": 12633
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6476577520370483,
      "learning_rate": 1.3722153403367652e-05,
      "loss": 0.4384,
      "step": 12634
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.584396481513977,
      "learning_rate": 1.3721208150044421e-05,
      "loss": 0.4623,
      "step": 12635
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5412966012954712,
      "learning_rate": 1.3720262858126308e-05,
      "loss": 0.4286,
      "step": 12636
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5330718755722046,
      "learning_rate": 1.3719317527623122e-05,
      "loss": 0.4599,
      "step": 12637
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6917296648025513,
      "learning_rate": 1.3718372158544667e-05,
      "loss": 0.4691,
      "step": 12638
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.664098858833313,
      "learning_rate": 1.3717426750900745e-05,
      "loss": 0.5053,
      "step": 12639
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5468717813491821,
      "learning_rate": 1.3716481304701161e-05,
      "loss": 0.4548,
      "step": 12640
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.564314365386963,
      "learning_rate": 1.3715535819955725e-05,
      "loss": 0.4617,
      "step": 12641
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5917681455612183,
      "learning_rate": 1.371459029667424e-05,
      "loss": 0.446,
      "step": 12642
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6081733703613281,
      "learning_rate": 1.3713644734866516e-05,
      "loss": 0.5153,
      "step": 12643
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.663735032081604,
      "learning_rate": 1.3712699134542355e-05,
      "loss": 0.4499,
      "step": 12644
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.559032678604126,
      "learning_rate": 1.3711753495711568e-05,
      "loss": 0.4888,
      "step": 12645
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5439350605010986,
      "learning_rate": 1.3710807818383964e-05,
      "loss": 0.4448,
      "step": 12646
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5026706457138062,
      "learning_rate": 1.3709862102569347e-05,
      "loss": 0.5211,
      "step": 12647
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.548449993133545,
      "learning_rate": 1.3708916348277528e-05,
      "loss": 0.4301,
      "step": 12648
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6085963249206543,
      "learning_rate": 1.3707970555518318e-05,
      "loss": 0.5373,
      "step": 12649
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5430735349655151,
      "learning_rate": 1.3707024724301519e-05,
      "loss": 0.4269,
      "step": 12650
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8162846565246582,
      "learning_rate": 1.370607885463695e-05,
      "loss": 0.5701,
      "step": 12651
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5154212713241577,
      "learning_rate": 1.3705132946534412e-05,
      "loss": 0.4286,
      "step": 12652
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5232245922088623,
      "learning_rate": 1.3704187000003726e-05,
      "loss": 0.4799,
      "step": 12653
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5322561264038086,
      "learning_rate": 1.3703241015054696e-05,
      "loss": 0.4413,
      "step": 12654
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.591704249382019,
      "learning_rate": 1.3702294991697133e-05,
      "loss": 0.4926,
      "step": 12655
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6662839651107788,
      "learning_rate": 1.3701348929940852e-05,
      "loss": 0.4684,
      "step": 12656
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5746715068817139,
      "learning_rate": 1.3700402829795665e-05,
      "loss": 0.519,
      "step": 12657
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.701470971107483,
      "learning_rate": 1.369945669127138e-05,
      "loss": 0.5026,
      "step": 12658
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4788436889648438,
      "learning_rate": 1.3698510514377818e-05,
      "loss": 0.4913,
      "step": 12659
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.717178225517273,
      "learning_rate": 1.3697564299124783e-05,
      "loss": 0.4554,
      "step": 12660
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.715080738067627,
      "learning_rate": 1.3696618045522098e-05,
      "loss": 0.5115,
      "step": 12661
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5346399545669556,
      "learning_rate": 1.3695671753579569e-05,
      "loss": 0.4477,
      "step": 12662
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5686322450637817,
      "learning_rate": 1.3694725423307015e-05,
      "loss": 0.5088,
      "step": 12663
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5235726833343506,
      "learning_rate": 1.3693779054714251e-05,
      "loss": 0.4462,
      "step": 12664
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5609750747680664,
      "learning_rate": 1.3692832647811094e-05,
      "loss": 0.4915,
      "step": 12665
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6284734010696411,
      "learning_rate": 1.3691886202607354e-05,
      "loss": 0.4784,
      "step": 12666
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6662936210632324,
      "learning_rate": 1.3690939719112852e-05,
      "loss": 0.5433,
      "step": 12667
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4852310419082642,
      "learning_rate": 1.3689993197337401e-05,
      "loss": 0.4162,
      "step": 12668
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.630335807800293,
      "learning_rate": 1.3689046637290821e-05,
      "loss": 0.5601,
      "step": 12669
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5824371576309204,
      "learning_rate": 1.3688100038982928e-05,
      "loss": 0.4218,
      "step": 12670
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6485153436660767,
      "learning_rate": 1.368715340242354e-05,
      "loss": 0.5247,
      "step": 12671
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5390523672103882,
      "learning_rate": 1.3686206727622477e-05,
      "loss": 0.4204,
      "step": 12672
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.566161036491394,
      "learning_rate": 1.3685260014589554e-05,
      "loss": 0.4883,
      "step": 12673
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5636074542999268,
      "learning_rate": 1.3684313263334592e-05,
      "loss": 0.4367,
      "step": 12674
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.55141019821167,
      "learning_rate": 1.3683366473867409e-05,
      "loss": 0.4746,
      "step": 12675
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9163682460784912,
      "learning_rate": 1.3682419646197827e-05,
      "loss": 0.4378,
      "step": 12676
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6039992570877075,
      "learning_rate": 1.3681472780335665e-05,
      "loss": 0.5061,
      "step": 12677
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6909617185592651,
      "learning_rate": 1.3680525876290742e-05,
      "loss": 0.4575,
      "step": 12678
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6389235258102417,
      "learning_rate": 1.3679578934072884e-05,
      "loss": 0.4935,
      "step": 12679
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4773306846618652,
      "learning_rate": 1.3678631953691903e-05,
      "loss": 0.4232,
      "step": 12680
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5664678812026978,
      "learning_rate": 1.3677684935157629e-05,
      "loss": 0.4738,
      "step": 12681
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5706695318222046,
      "learning_rate": 1.3676737878479879e-05,
      "loss": 0.4336,
      "step": 12682
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5219472646713257,
      "learning_rate": 1.367579078366848e-05,
      "loss": 0.4887,
      "step": 12683
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6344999074935913,
      "learning_rate": 1.3674843650733254e-05,
      "loss": 0.4345,
      "step": 12684
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5664377212524414,
      "learning_rate": 1.3673896479684022e-05,
      "loss": 0.4881,
      "step": 12685
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5345009565353394,
      "learning_rate": 1.3672949270530604e-05,
      "loss": 0.4422,
      "step": 12686
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5650215148925781,
      "learning_rate": 1.3672002023282836e-05,
      "loss": 0.4657,
      "step": 12687
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5546718835830688,
      "learning_rate": 1.3671054737950531e-05,
      "loss": 0.4321,
      "step": 12688
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5863102674484253,
      "learning_rate": 1.3670107414543519e-05,
      "loss": 0.5231,
      "step": 12689
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5605276823043823,
      "learning_rate": 1.3669160053071624e-05,
      "loss": 0.4169,
      "step": 12690
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.96864652633667,
      "learning_rate": 1.3668212653544673e-05,
      "loss": 0.4964,
      "step": 12691
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6092841625213623,
      "learning_rate": 1.366726521597249e-05,
      "loss": 0.4621,
      "step": 12692
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6141353845596313,
      "learning_rate": 1.3666317740364902e-05,
      "loss": 0.5246,
      "step": 12693
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5841809511184692,
      "learning_rate": 1.3665370226731737e-05,
      "loss": 0.4258,
      "step": 12694
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.572257161140442,
      "learning_rate": 1.3664422675082824e-05,
      "loss": 0.4603,
      "step": 12695
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.667895793914795,
      "learning_rate": 1.3663475085427984e-05,
      "loss": 0.4469,
      "step": 12696
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7052727937698364,
      "learning_rate": 1.3662527457777053e-05,
      "loss": 0.5942,
      "step": 12697
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.574059009552002,
      "learning_rate": 1.366157979213985e-05,
      "loss": 0.4244,
      "step": 12698
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6710506677627563,
      "learning_rate": 1.3660632088526214e-05,
      "loss": 0.5311,
      "step": 12699
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6737523078918457,
      "learning_rate": 1.365968434694597e-05,
      "loss": 0.4794,
      "step": 12700
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6827465295791626,
      "learning_rate": 1.3658736567408944e-05,
      "loss": 0.5323,
      "step": 12701
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5065536499023438,
      "learning_rate": 1.3657788749924972e-05,
      "loss": 0.4485,
      "step": 12702
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6340537071228027,
      "learning_rate": 1.365684089450388e-05,
      "loss": 0.4719,
      "step": 12703
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5089713335037231,
      "learning_rate": 1.3655893001155499e-05,
      "loss": 0.4373,
      "step": 12704
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6721831560134888,
      "learning_rate": 1.3654945069889664e-05,
      "loss": 0.495,
      "step": 12705
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6641408205032349,
      "learning_rate": 1.3653997100716204e-05,
      "loss": 0.4443,
      "step": 12706
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.628327488899231,
      "learning_rate": 1.365304909364495e-05,
      "loss": 0.4626,
      "step": 12707
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4432814121246338,
      "learning_rate": 1.3652101048685735e-05,
      "loss": 0.428,
      "step": 12708
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.448726773262024,
      "learning_rate": 1.3651152965848393e-05,
      "loss": 0.4838,
      "step": 12709
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.450860857963562,
      "learning_rate": 1.3650204845142755e-05,
      "loss": 0.4468,
      "step": 12710
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7152339220046997,
      "learning_rate": 1.3649256686578659e-05,
      "loss": 0.5096,
      "step": 12711
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.706367015838623,
      "learning_rate": 1.3648308490165932e-05,
      "loss": 0.5062,
      "step": 12712
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.567986249923706,
      "learning_rate": 1.3647360255914415e-05,
      "loss": 0.5009,
      "step": 12713
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5483148097991943,
      "learning_rate": 1.3646411983833936e-05,
      "loss": 0.4725,
      "step": 12714
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6144858598709106,
      "learning_rate": 1.3645463673934338e-05,
      "loss": 0.4609,
      "step": 12715
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.586319923400879,
      "learning_rate": 1.364451532622545e-05,
      "loss": 0.438,
      "step": 12716
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5690447092056274,
      "learning_rate": 1.364356694071711e-05,
      "loss": 0.4803,
      "step": 12717
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6755446195602417,
      "learning_rate": 1.3642618517419156e-05,
      "loss": 0.4302,
      "step": 12718
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6386213302612305,
      "learning_rate": 1.3641670056341423e-05,
      "loss": 0.4754,
      "step": 12719
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6237502098083496,
      "learning_rate": 1.3640721557493746e-05,
      "loss": 0.4444,
      "step": 12720
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5501110553741455,
      "learning_rate": 1.3639773020885966e-05,
      "loss": 0.4954,
      "step": 12721
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5629264116287231,
      "learning_rate": 1.3638824446527919e-05,
      "loss": 0.4387,
      "step": 12722
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6095050573349,
      "learning_rate": 1.3637875834429446e-05,
      "loss": 0.4761,
      "step": 12723
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5882234573364258,
      "learning_rate": 1.3636927184600379e-05,
      "loss": 0.4342,
      "step": 12724
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8032339811325073,
      "learning_rate": 1.3635978497050566e-05,
      "loss": 0.5557,
      "step": 12725
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2758963108062744,
      "learning_rate": 1.3635029771789838e-05,
      "loss": 0.4472,
      "step": 12726
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4692260026931763,
      "learning_rate": 1.3634081008828042e-05,
      "loss": 0.4976,
      "step": 12727
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5805195569992065,
      "learning_rate": 1.3633132208175012e-05,
      "loss": 0.4886,
      "step": 12728
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5567318201065063,
      "learning_rate": 1.3632183369840592e-05,
      "loss": 0.4795,
      "step": 12729
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5065654516220093,
      "learning_rate": 1.3631234493834622e-05,
      "loss": 0.4397,
      "step": 12730
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.814478874206543,
      "learning_rate": 1.3630285580166946e-05,
      "loss": 0.5007,
      "step": 12731
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.521004557609558,
      "learning_rate": 1.3629336628847401e-05,
      "loss": 0.4274,
      "step": 12732
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5494835376739502,
      "learning_rate": 1.3628387639885832e-05,
      "loss": 0.4779,
      "step": 12733
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.561206579208374,
      "learning_rate": 1.3627438613292081e-05,
      "loss": 0.4398,
      "step": 12734
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5802083015441895,
      "learning_rate": 1.3626489549075992e-05,
      "loss": 0.4734,
      "step": 12735
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.542497992515564,
      "learning_rate": 1.3625540447247407e-05,
      "loss": 0.4526,
      "step": 12736
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5548934936523438,
      "learning_rate": 1.3624591307816171e-05,
      "loss": 0.4841,
      "step": 12737
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4439401626586914,
      "learning_rate": 1.3623642130792125e-05,
      "loss": 0.41,
      "step": 12738
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.512532353401184,
      "learning_rate": 1.3622692916185116e-05,
      "loss": 0.473,
      "step": 12739
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6085110902786255,
      "learning_rate": 1.3621743664004987e-05,
      "loss": 0.4482,
      "step": 12740
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.615470290184021,
      "learning_rate": 1.362079437426159e-05,
      "loss": 0.5044,
      "step": 12741
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5377222299575806,
      "learning_rate": 1.361984504696476e-05,
      "loss": 0.4377,
      "step": 12742
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5899415016174316,
      "learning_rate": 1.361889568212435e-05,
      "loss": 0.5058,
      "step": 12743
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.596092700958252,
      "learning_rate": 1.3617946279750206e-05,
      "loss": 0.4653,
      "step": 12744
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.57953679561615,
      "learning_rate": 1.3616996839852172e-05,
      "loss": 0.5012,
      "step": 12745
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.562514066696167,
      "learning_rate": 1.3616047362440099e-05,
      "loss": 0.4557,
      "step": 12746
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6407746076583862,
      "learning_rate": 1.3615097847523833e-05,
      "loss": 0.4698,
      "step": 12747
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5596641302108765,
      "learning_rate": 1.3614148295113216e-05,
      "loss": 0.4589,
      "step": 12748
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.532755732536316,
      "learning_rate": 1.3613198705218109e-05,
      "loss": 0.4922,
      "step": 12749
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5151355266571045,
      "learning_rate": 1.361224907784835e-05,
      "loss": 0.4183,
      "step": 12750
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6469324827194214,
      "learning_rate": 1.3611299413013792e-05,
      "loss": 0.5395,
      "step": 12751
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5574430227279663,
      "learning_rate": 1.3610349710724284e-05,
      "loss": 0.4568,
      "step": 12752
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.497707724571228,
      "learning_rate": 1.3609399970989676e-05,
      "loss": 0.4958,
      "step": 12753
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5767743587493896,
      "learning_rate": 1.360845019381982e-05,
      "loss": 0.4709,
      "step": 12754
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5904022455215454,
      "learning_rate": 1.3607500379224565e-05,
      "loss": 0.4552,
      "step": 12755
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5368541479110718,
      "learning_rate": 1.3606550527213762e-05,
      "loss": 0.439,
      "step": 12756
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7093104124069214,
      "learning_rate": 1.3605600637797265e-05,
      "loss": 0.5275,
      "step": 12757
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.53717041015625,
      "learning_rate": 1.360465071098492e-05,
      "loss": 0.4271,
      "step": 12758
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5768072605133057,
      "learning_rate": 1.3603700746786586e-05,
      "loss": 0.5153,
      "step": 12759
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5494015216827393,
      "learning_rate": 1.3602750745212107e-05,
      "loss": 0.4298,
      "step": 12760
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4675487279891968,
      "learning_rate": 1.360180070627135e-05,
      "loss": 0.4755,
      "step": 12761
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.512077808380127,
      "learning_rate": 1.3600850629974153e-05,
      "loss": 0.4451,
      "step": 12762
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.883010745048523,
      "learning_rate": 1.3599900516330382e-05,
      "loss": 0.5177,
      "step": 12763
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5480154752731323,
      "learning_rate": 1.3598950365349884e-05,
      "loss": 0.4394,
      "step": 12764
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6019599437713623,
      "learning_rate": 1.3598000177042514e-05,
      "loss": 0.481,
      "step": 12765
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5001276731491089,
      "learning_rate": 1.3597049951418131e-05,
      "loss": 0.4547,
      "step": 12766
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.84618079662323,
      "learning_rate": 1.3596099688486586e-05,
      "loss": 1.0352,
      "step": 12767
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.549109697341919,
      "learning_rate": 1.359514938825774e-05,
      "loss": 1.0633,
      "step": 12768
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7542943954467773,
      "learning_rate": 1.3594199050741443e-05,
      "loss": 0.5203,
      "step": 12769
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.609934687614441,
      "learning_rate": 1.3593248675947557e-05,
      "loss": 0.476,
      "step": 12770
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.544739007949829,
      "learning_rate": 1.3592298263885932e-05,
      "loss": 0.4944,
      "step": 12771
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6462862491607666,
      "learning_rate": 1.3591347814566435e-05,
      "loss": 0.4511,
      "step": 12772
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5807840824127197,
      "learning_rate": 1.3590397327998914e-05,
      "loss": 0.5429,
      "step": 12773
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6649664640426636,
      "learning_rate": 1.3589446804193236e-05,
      "loss": 0.4656,
      "step": 12774
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.565334439277649,
      "learning_rate": 1.358849624315925e-05,
      "loss": 0.5368,
      "step": 12775
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5739283561706543,
      "learning_rate": 1.3587545644906823e-05,
      "loss": 0.4527,
      "step": 12776
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6106632947921753,
      "learning_rate": 1.358659500944581e-05,
      "loss": 0.5063,
      "step": 12777
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5049632787704468,
      "learning_rate": 1.3585644336786071e-05,
      "loss": 0.4529,
      "step": 12778
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.719408392906189,
      "learning_rate": 1.3584693626937467e-05,
      "loss": 0.5044,
      "step": 12779
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6869615316390991,
      "learning_rate": 1.3583742879909859e-05,
      "loss": 0.415,
      "step": 12780
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5844908952713013,
      "learning_rate": 1.3582792095713101e-05,
      "loss": 0.4405,
      "step": 12781
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6362804174423218,
      "learning_rate": 1.3581841274357066e-05,
      "loss": 0.4388,
      "step": 12782
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5975019931793213,
      "learning_rate": 1.3580890415851606e-05,
      "loss": 0.5037,
      "step": 12783
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6805520057678223,
      "learning_rate": 1.357993952020659e-05,
      "loss": 0.4815,
      "step": 12784
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6340126991271973,
      "learning_rate": 1.3578988587431871e-05,
      "loss": 0.4681,
      "step": 12785
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5291950702667236,
      "learning_rate": 1.3578037617537323e-05,
      "loss": 0.4512,
      "step": 12786
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5870743989944458,
      "learning_rate": 1.35770866105328e-05,
      "loss": 0.5145,
      "step": 12787
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6302416324615479,
      "learning_rate": 1.357613556642817e-05,
      "loss": 0.4413,
      "step": 12788
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6569337844848633,
      "learning_rate": 1.3575184485233293e-05,
      "loss": 0.4952,
      "step": 12789
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6676695346832275,
      "learning_rate": 1.357423336695804e-05,
      "loss": 0.4828,
      "step": 12790
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.716784119606018,
      "learning_rate": 1.3573282211612266e-05,
      "loss": 0.5807,
      "step": 12791
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5648518800735474,
      "learning_rate": 1.3572331019205847e-05,
      "loss": 0.4337,
      "step": 12792
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4741686582565308,
      "learning_rate": 1.357137978974864e-05,
      "loss": 0.4576,
      "step": 12793
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.678781747817993,
      "learning_rate": 1.3570428523250517e-05,
      "loss": 0.4402,
      "step": 12794
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6116175651550293,
      "learning_rate": 1.3569477219721336e-05,
      "loss": 0.48,
      "step": 12795
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6570351123809814,
      "learning_rate": 1.3568525879170972e-05,
      "loss": 0.4861,
      "step": 12796
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.606266736984253,
      "learning_rate": 1.3567574501609287e-05,
      "loss": 0.544,
      "step": 12797
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.474583387374878,
      "learning_rate": 1.3566623087046148e-05,
      "loss": 0.4402,
      "step": 12798
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4861259460449219,
      "learning_rate": 1.3565671635491426e-05,
      "loss": 0.4565,
      "step": 12799
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5926111936569214,
      "learning_rate": 1.3564720146954989e-05,
      "loss": 0.458,
      "step": 12800
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6571450233459473,
      "learning_rate": 1.35637686214467e-05,
      "loss": 0.5266,
      "step": 12801
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.608004093170166,
      "learning_rate": 1.3562817058976434e-05,
      "loss": 0.4774,
      "step": 12802
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5605744123458862,
      "learning_rate": 1.3561865459554059e-05,
      "loss": 0.4977,
      "step": 12803
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5174728631973267,
      "learning_rate": 1.356091382318944e-05,
      "loss": 0.4193,
      "step": 12804
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6614817380905151,
      "learning_rate": 1.3559962149892455e-05,
      "loss": 0.5444,
      "step": 12805
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0788414478302,
      "learning_rate": 1.3559010439672966e-05,
      "loss": 0.458,
      "step": 12806
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6225382089614868,
      "learning_rate": 1.3558058692540846e-05,
      "loss": 0.4902,
      "step": 12807
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.562843918800354,
      "learning_rate": 1.3557106908505975e-05,
      "loss": 0.428,
      "step": 12808
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.620675802230835,
      "learning_rate": 1.3556155087578209e-05,
      "loss": 0.4817,
      "step": 12809
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6735198497772217,
      "learning_rate": 1.3555203229767433e-05,
      "loss": 0.4398,
      "step": 12810
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4994332790374756,
      "learning_rate": 1.3554251335083513e-05,
      "loss": 0.502,
      "step": 12811
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.534262776374817,
      "learning_rate": 1.3553299403536323e-05,
      "loss": 0.454,
      "step": 12812
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8803471326828003,
      "learning_rate": 1.3552347435135737e-05,
      "loss": 0.504,
      "step": 12813
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4791274070739746,
      "learning_rate": 1.355139542989163e-05,
      "loss": 0.4558,
      "step": 12814
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.776124119758606,
      "learning_rate": 1.355044338781387e-05,
      "loss": 0.5384,
      "step": 12815
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.481261968612671,
      "learning_rate": 1.3549491308912335e-05,
      "loss": 0.4251,
      "step": 12816
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6914395093917847,
      "learning_rate": 1.3548539193196896e-05,
      "loss": 0.5595,
      "step": 12817
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4966394901275635,
      "learning_rate": 1.354758704067744e-05,
      "loss": 0.4158,
      "step": 12818
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5780364274978638,
      "learning_rate": 1.3546634851363825e-05,
      "loss": 0.489,
      "step": 12819
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5526889562606812,
      "learning_rate": 1.354568262526594e-05,
      "loss": 0.4424,
      "step": 12820
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.685240626335144,
      "learning_rate": 1.3544730362393654e-05,
      "loss": 0.5135,
      "step": 12821
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4992681741714478,
      "learning_rate": 1.3543778062756848e-05,
      "loss": 0.4354,
      "step": 12822
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.579484224319458,
      "learning_rate": 1.3542825726365394e-05,
      "loss": 0.5516,
      "step": 12823
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6271497011184692,
      "learning_rate": 1.3541873353229174e-05,
      "loss": 0.4175,
      "step": 12824
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5539066791534424,
      "learning_rate": 1.354092094335806e-05,
      "loss": 0.4875,
      "step": 12825
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5460766553878784,
      "learning_rate": 1.353996849676194e-05,
      "loss": 0.4363,
      "step": 12826
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5490589141845703,
      "learning_rate": 1.353901601345068e-05,
      "loss": 0.471,
      "step": 12827
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4908828735351562,
      "learning_rate": 1.3538063493434167e-05,
      "loss": 0.4163,
      "step": 12828
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.633663535118103,
      "learning_rate": 1.353711093672228e-05,
      "loss": 0.5104,
      "step": 12829
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5087153911590576,
      "learning_rate": 1.3536158343324894e-05,
      "loss": 0.4604,
      "step": 12830
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5054329633712769,
      "learning_rate": 1.3535205713251892e-05,
      "loss": 0.5536,
      "step": 12831
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.508875846862793,
      "learning_rate": 1.3534253046513152e-05,
      "loss": 0.4245,
      "step": 12832
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5627539157867432,
      "learning_rate": 1.3533300343118559e-05,
      "loss": 0.4737,
      "step": 12833
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4709933996200562,
      "learning_rate": 1.3532347603077991e-05,
      "loss": 0.3995,
      "step": 12834
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6189841032028198,
      "learning_rate": 1.3531394826401328e-05,
      "loss": 0.537,
      "step": 12835
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5427196025848389,
      "learning_rate": 1.3530442013098456e-05,
      "loss": 0.4421,
      "step": 12836
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5380218029022217,
      "learning_rate": 1.3529489163179252e-05,
      "loss": 0.5316,
      "step": 12837
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6346862316131592,
      "learning_rate": 1.3528536276653603e-05,
      "loss": 0.4941,
      "step": 12838
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5352226495742798,
      "learning_rate": 1.3527583353531392e-05,
      "loss": 0.4918,
      "step": 12839
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6195669174194336,
      "learning_rate": 1.35266303938225e-05,
      "loss": 0.4714,
      "step": 12840
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.564679741859436,
      "learning_rate": 1.3525677397536811e-05,
      "loss": 0.4985,
      "step": 12841
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5438354015350342,
      "learning_rate": 1.352472436468421e-05,
      "loss": 0.429,
      "step": 12842
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5735317468643188,
      "learning_rate": 1.3523771295274577e-05,
      "loss": 0.4907,
      "step": 12843
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5877881050109863,
      "learning_rate": 1.3522818189317808e-05,
      "loss": 0.4182,
      "step": 12844
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7079169750213623,
      "learning_rate": 1.3521865046823774e-05,
      "loss": 0.5531,
      "step": 12845
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5531387329101562,
      "learning_rate": 1.3520911867802373e-05,
      "loss": 0.458,
      "step": 12846
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5026426315307617,
      "learning_rate": 1.3519958652263484e-05,
      "loss": 0.5122,
      "step": 12847
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5059465169906616,
      "learning_rate": 1.3519005400216994e-05,
      "loss": 0.4432,
      "step": 12848
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.620564579963684,
      "learning_rate": 1.3518052111672792e-05,
      "loss": 0.458,
      "step": 12849
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5598350763320923,
      "learning_rate": 1.3517098786640763e-05,
      "loss": 0.4753,
      "step": 12850
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5583080053329468,
      "learning_rate": 1.3516145425130795e-05,
      "loss": 0.4968,
      "step": 12851
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5194159746170044,
      "learning_rate": 1.351519202715278e-05,
      "loss": 0.4351,
      "step": 12852
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6713496446609497,
      "learning_rate": 1.3514238592716598e-05,
      "loss": 0.5066,
      "step": 12853
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5733529329299927,
      "learning_rate": 1.3513285121832148e-05,
      "loss": 0.4755,
      "step": 12854
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.548206090927124,
      "learning_rate": 1.3512331614509307e-05,
      "loss": 0.4625,
      "step": 12855
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5223766565322876,
      "learning_rate": 1.3511378070757972e-05,
      "loss": 0.4348,
      "step": 12856
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5599439144134521,
      "learning_rate": 1.3510424490588031e-05,
      "loss": 0.4647,
      "step": 12857
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5381511449813843,
      "learning_rate": 1.3509470874009378e-05,
      "loss": 0.4382,
      "step": 12858
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6654102802276611,
      "learning_rate": 1.3508517221031898e-05,
      "loss": 0.5184,
      "step": 12859
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4709951877593994,
      "learning_rate": 1.3507563531665486e-05,
      "loss": 0.4148,
      "step": 12860
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6302361488342285,
      "learning_rate": 1.3506609805920028e-05,
      "loss": 0.4921,
      "step": 12861
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4991549253463745,
      "learning_rate": 1.3505656043805422e-05,
      "loss": 0.439,
      "step": 12862
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6037877798080444,
      "learning_rate": 1.3504702245331553e-05,
      "loss": 0.5391,
      "step": 12863
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5058708190917969,
      "learning_rate": 1.3503748410508321e-05,
      "loss": 0.4379,
      "step": 12864
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6222745180130005,
      "learning_rate": 1.3502794539345614e-05,
      "loss": 0.5352,
      "step": 12865
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5876705646514893,
      "learning_rate": 1.3501840631853327e-05,
      "loss": 0.4568,
      "step": 12866
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.555027723312378,
      "learning_rate": 1.3500886688041353e-05,
      "loss": 0.493,
      "step": 12867
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6251835823059082,
      "learning_rate": 1.3499932707919585e-05,
      "loss": 0.4541,
      "step": 12868
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5834163427352905,
      "learning_rate": 1.3498978691497917e-05,
      "loss": 0.4989,
      "step": 12869
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6304773092269897,
      "learning_rate": 1.3498024638786249e-05,
      "loss": 0.4302,
      "step": 12870
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.533751130104065,
      "learning_rate": 1.3497070549794466e-05,
      "loss": 0.4684,
      "step": 12871
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5185493230819702,
      "learning_rate": 1.3496116424532474e-05,
      "loss": 0.4205,
      "step": 12872
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5823413133621216,
      "learning_rate": 1.3495162263010163e-05,
      "loss": 0.4935,
      "step": 12873
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7981053590774536,
      "learning_rate": 1.3494208065237428e-05,
      "loss": 0.4737,
      "step": 12874
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5505925416946411,
      "learning_rate": 1.3493253831224171e-05,
      "loss": 0.4636,
      "step": 12875
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5534026622772217,
      "learning_rate": 1.3492299560980284e-05,
      "loss": 0.4562,
      "step": 12876
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5160133838653564,
      "learning_rate": 1.3491345254515666e-05,
      "loss": 0.506,
      "step": 12877
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.49113929271698,
      "learning_rate": 1.3490390911840217e-05,
      "loss": 0.445,
      "step": 12878
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5936896800994873,
      "learning_rate": 1.3489436532963829e-05,
      "loss": 0.4869,
      "step": 12879
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5732945203781128,
      "learning_rate": 1.3488482117896409e-05,
      "loss": 0.4546,
      "step": 12880
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6362663507461548,
      "learning_rate": 1.3487527666647847e-05,
      "loss": 0.5085,
      "step": 12881
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5252362489700317,
      "learning_rate": 1.3486573179228048e-05,
      "loss": 0.4707,
      "step": 12882
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5791269540786743,
      "learning_rate": 1.3485618655646908e-05,
      "loss": 0.4698,
      "step": 12883
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5988552570343018,
      "learning_rate": 1.3484664095914333e-05,
      "loss": 0.4386,
      "step": 12884
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6679290533065796,
      "learning_rate": 1.3483709500040216e-05,
      "loss": 0.5045,
      "step": 12885
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5255544185638428,
      "learning_rate": 1.3482754868034462e-05,
      "loss": 0.4428,
      "step": 12886
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5358420610427856,
      "learning_rate": 1.3481800199906967e-05,
      "loss": 0.496,
      "step": 12887
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.526578664779663,
      "learning_rate": 1.3480845495667643e-05,
      "loss": 0.4288,
      "step": 12888
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5648205280303955,
      "learning_rate": 1.3479890755326376e-05,
      "loss": 0.4779,
      "step": 12889
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.667335033416748,
      "learning_rate": 1.3478935978893087e-05,
      "loss": 0.4653,
      "step": 12890
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6346489191055298,
      "learning_rate": 1.3477981166377662e-05,
      "loss": 0.4549,
      "step": 12891
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4802480936050415,
      "learning_rate": 1.3477026317790012e-05,
      "loss": 0.4177,
      "step": 12892
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5465285778045654,
      "learning_rate": 1.347607143314004e-05,
      "loss": 0.4639,
      "step": 12893
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5656987428665161,
      "learning_rate": 1.3475116512437651e-05,
      "loss": 0.4284,
      "step": 12894
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5885887145996094,
      "learning_rate": 1.3474161555692743e-05,
      "loss": 0.4817,
      "step": 12895
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.635185718536377,
      "learning_rate": 1.3473206562915227e-05,
      "loss": 0.4539,
      "step": 12896
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6283767223358154,
      "learning_rate": 1.3472251534115001e-05,
      "loss": 0.4887,
      "step": 12897
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.609582543373108,
      "learning_rate": 1.347129646930198e-05,
      "loss": 0.4698,
      "step": 12898
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6031134128570557,
      "learning_rate": 1.347034136848606e-05,
      "loss": 0.4541,
      "step": 12899
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.525444507598877,
      "learning_rate": 1.346938623167715e-05,
      "loss": 0.432,
      "step": 12900
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5568468570709229,
      "learning_rate": 1.3468431058885159e-05,
      "loss": 0.4972,
      "step": 12901
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5223312377929688,
      "learning_rate": 1.346747585011999e-05,
      "loss": 0.4586,
      "step": 12902
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6866649389266968,
      "learning_rate": 1.3466520605391553e-05,
      "loss": 0.4945,
      "step": 12903
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.639535903930664,
      "learning_rate": 1.3465565324709755e-05,
      "loss": 0.4274,
      "step": 12904
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.506028652191162,
      "learning_rate": 1.34646100080845e-05,
      "loss": 0.4579,
      "step": 12905
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.590638518333435,
      "learning_rate": 1.3463654655525702e-05,
      "loss": 0.4324,
      "step": 12906
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6406229734420776,
      "learning_rate": 1.3462699267043262e-05,
      "loss": 0.4852,
      "step": 12907
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5758399963378906,
      "learning_rate": 1.34617438426471e-05,
      "loss": 0.4324,
      "step": 12908
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5254690647125244,
      "learning_rate": 1.3460788382347114e-05,
      "loss": 0.456,
      "step": 12909
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5994856357574463,
      "learning_rate": 1.3459832886153218e-05,
      "loss": 0.4649,
      "step": 12910
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.504323959350586,
      "learning_rate": 1.3458877354075326e-05,
      "loss": 0.4854,
      "step": 12911
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5147532224655151,
      "learning_rate": 1.3457921786123343e-05,
      "loss": 0.4668,
      "step": 12912
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6326419115066528,
      "learning_rate": 1.3456966182307181e-05,
      "loss": 0.5174,
      "step": 12913
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.540953278541565,
      "learning_rate": 1.3456010542636752e-05,
      "loss": 0.4368,
      "step": 12914
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6478098630905151,
      "learning_rate": 1.3455054867121967e-05,
      "loss": 0.5221,
      "step": 12915
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5336792469024658,
      "learning_rate": 1.345409915577274e-05,
      "loss": 0.4509,
      "step": 12916
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5038049221038818,
      "learning_rate": 1.345314340859898e-05,
      "loss": 0.4362,
      "step": 12917
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5465794801712036,
      "learning_rate": 1.3452187625610597e-05,
      "loss": 0.4398,
      "step": 12918
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9438968896865845,
      "learning_rate": 1.3451231806817513e-05,
      "loss": 1.0174,
      "step": 12919
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5841974020004272,
      "learning_rate": 1.3450275952229637e-05,
      "loss": 1.009,
      "step": 12920
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6274712085723877,
      "learning_rate": 1.3449320061856881e-05,
      "loss": 0.4811,
      "step": 12921
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7265483140945435,
      "learning_rate": 1.3448364135709157e-05,
      "loss": 0.4826,
      "step": 12922
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6284878253936768,
      "learning_rate": 1.3447408173796386e-05,
      "loss": 0.5006,
      "step": 12923
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.590120553970337,
      "learning_rate": 1.344645217612848e-05,
      "loss": 0.464,
      "step": 12924
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7382659912109375,
      "learning_rate": 1.344549614271535e-05,
      "loss": 0.5338,
      "step": 12925
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.49978768825531,
      "learning_rate": 1.3444540073566918e-05,
      "loss": 0.4373,
      "step": 12926
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.64150071144104,
      "learning_rate": 1.3443583968693099e-05,
      "loss": 0.5052,
      "step": 12927
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5258461236953735,
      "learning_rate": 1.3442627828103805e-05,
      "loss": 0.4421,
      "step": 12928
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5482045412063599,
      "learning_rate": 1.3441671651808956e-05,
      "loss": 0.4763,
      "step": 12929
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.525469183921814,
      "learning_rate": 1.344071543981847e-05,
      "loss": 0.4312,
      "step": 12930
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5442116260528564,
      "learning_rate": 1.343975919214226e-05,
      "loss": 0.4726,
      "step": 12931
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6190330982208252,
      "learning_rate": 1.343880290879025e-05,
      "loss": 0.4945,
      "step": 12932
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.776124119758606,
      "learning_rate": 1.3437846589772352e-05,
      "loss": 0.5562,
      "step": 12933
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.538698673248291,
      "learning_rate": 1.3436890235098492e-05,
      "loss": 0.4457,
      "step": 12934
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5004026889801025,
      "learning_rate": 1.3435933844778579e-05,
      "loss": 0.4767,
      "step": 12935
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5084474086761475,
      "learning_rate": 1.3434977418822539e-05,
      "loss": 0.4538,
      "step": 12936
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.536490797996521,
      "learning_rate": 1.3434020957240292e-05,
      "loss": 0.4458,
      "step": 12937
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.596665859222412,
      "learning_rate": 1.3433064460041756e-05,
      "loss": 0.4574,
      "step": 12938
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5243875980377197,
      "learning_rate": 1.3432107927236851e-05,
      "loss": 0.4534,
      "step": 12939
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4681161642074585,
      "learning_rate": 1.34311513588355e-05,
      "loss": 0.441,
      "step": 12940
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5882738828659058,
      "learning_rate": 1.3430194754847618e-05,
      "loss": 0.4947,
      "step": 12941
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5276144742965698,
      "learning_rate": 1.3429238115283138e-05,
      "loss": 0.4353,
      "step": 12942
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5883121490478516,
      "learning_rate": 1.3428281440151972e-05,
      "loss": 0.5024,
      "step": 12943
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5068663358688354,
      "learning_rate": 1.3427324729464046e-05,
      "loss": 0.4249,
      "step": 12944
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5816725492477417,
      "learning_rate": 1.3426367983229281e-05,
      "loss": 0.5029,
      "step": 12945
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.557869553565979,
      "learning_rate": 1.3425411201457602e-05,
      "loss": 0.44,
      "step": 12946
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5996177196502686,
      "learning_rate": 1.3424454384158931e-05,
      "loss": 0.4851,
      "step": 12947
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5081342458724976,
      "learning_rate": 1.342349753134319e-05,
      "loss": 0.4326,
      "step": 12948
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.45913827419281,
      "learning_rate": 1.342254064302031e-05,
      "loss": 0.4982,
      "step": 12949
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.622771143913269,
      "learning_rate": 1.3421583719200206e-05,
      "loss": 0.4464,
      "step": 12950
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5777055025100708,
      "learning_rate": 1.3420626759892808e-05,
      "loss": 0.4808,
      "step": 12951
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6506328582763672,
      "learning_rate": 1.3419669765108042e-05,
      "loss": 0.475,
      "step": 12952
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6031780242919922,
      "learning_rate": 1.3418712734855833e-05,
      "loss": 0.5183,
      "step": 12953
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.645952820777893,
      "learning_rate": 1.3417755669146104e-05,
      "loss": 0.4411,
      "step": 12954
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6365368366241455,
      "learning_rate": 1.3416798567988784e-05,
      "loss": 0.4832,
      "step": 12955
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6415914297103882,
      "learning_rate": 1.34158414313938e-05,
      "loss": 0.4411,
      "step": 12956
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.615032434463501,
      "learning_rate": 1.3414884259371079e-05,
      "loss": 0.5339,
      "step": 12957
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4621007442474365,
      "learning_rate": 1.3413927051930546e-05,
      "loss": 0.4106,
      "step": 12958
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4801867008209229,
      "learning_rate": 1.3412969809082128e-05,
      "loss": 0.4305,
      "step": 12959
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5877605676651,
      "learning_rate": 1.341201253083576e-05,
      "loss": 0.449,
      "step": 12960
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6286821365356445,
      "learning_rate": 1.3411055217201363e-05,
      "loss": 0.4939,
      "step": 12961
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8442966938018799,
      "learning_rate": 1.341009786818887e-05,
      "loss": 0.4296,
      "step": 12962
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5745635032653809,
      "learning_rate": 1.3409140483808209e-05,
      "loss": 0.5021,
      "step": 12963
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4919822216033936,
      "learning_rate": 1.3408183064069311e-05,
      "loss": 0.4255,
      "step": 12964
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5901432037353516,
      "learning_rate": 1.3407225608982102e-05,
      "loss": 0.4871,
      "step": 12965
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6375151872634888,
      "learning_rate": 1.3406268118556516e-05,
      "loss": 0.457,
      "step": 12966
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6357654333114624,
      "learning_rate": 1.3405310592802485e-05,
      "loss": 0.5349,
      "step": 12967
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5355795621871948,
      "learning_rate": 1.3404353031729935e-05,
      "loss": 0.4466,
      "step": 12968
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6669723987579346,
      "learning_rate": 1.3403395435348801e-05,
      "loss": 0.5306,
      "step": 12969
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.566673755645752,
      "learning_rate": 1.3402437803669014e-05,
      "loss": 0.4463,
      "step": 12970
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6307631731033325,
      "learning_rate": 1.3401480136700507e-05,
      "loss": 0.4917,
      "step": 12971
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.542712926864624,
      "learning_rate": 1.3400522434453209e-05,
      "loss": 0.4514,
      "step": 12972
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.670824408531189,
      "learning_rate": 1.3399564696937057e-05,
      "loss": 0.4773,
      "step": 12973
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5221303701400757,
      "learning_rate": 1.3398606924161986e-05,
      "loss": 0.4101,
      "step": 12974
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.644362211227417,
      "learning_rate": 1.3397649116137923e-05,
      "loss": 0.4424,
      "step": 12975
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4923923015594482,
      "learning_rate": 1.3396691272874807e-05,
      "loss": 0.4267,
      "step": 12976
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.551306128501892,
      "learning_rate": 1.3395733394382569e-05,
      "loss": 0.471,
      "step": 12977
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5586718320846558,
      "learning_rate": 1.3394775480671149e-05,
      "loss": 0.4589,
      "step": 12978
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5438768863677979,
      "learning_rate": 1.3393817531750478e-05,
      "loss": 0.4382,
      "step": 12979
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5372068881988525,
      "learning_rate": 1.339285954763049e-05,
      "loss": 0.4569,
      "step": 12980
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.486480474472046,
      "learning_rate": 1.3391901528321123e-05,
      "loss": 0.4839,
      "step": 12981
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5398293733596802,
      "learning_rate": 1.3390943473832315e-05,
      "loss": 0.4321,
      "step": 12982
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.052823305130005,
      "learning_rate": 1.3389985384174e-05,
      "loss": 1.0284,
      "step": 12983
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.670407772064209,
      "learning_rate": 1.3389027259356117e-05,
      "loss": 1.0741,
      "step": 12984
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6071815490722656,
      "learning_rate": 1.3388069099388598e-05,
      "loss": 0.5008,
      "step": 12985
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6448923349380493,
      "learning_rate": 1.338711090428139e-05,
      "loss": 0.4416,
      "step": 12986
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.622794508934021,
      "learning_rate": 1.3386152674044421e-05,
      "loss": 0.4464,
      "step": 12987
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5882614850997925,
      "learning_rate": 1.3385194408687641e-05,
      "loss": 0.4615,
      "step": 12988
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5117868185043335,
      "learning_rate": 1.3384236108220973e-05,
      "loss": 0.4774,
      "step": 12989
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5687390565872192,
      "learning_rate": 1.338327777265437e-05,
      "loss": 0.4462,
      "step": 12990
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.554671287536621,
      "learning_rate": 1.3382319401997765e-05,
      "loss": 0.486,
      "step": 12991
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6201159954071045,
      "learning_rate": 1.3381360996261102e-05,
      "loss": 0.4623,
      "step": 12992
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6188472509384155,
      "learning_rate": 1.3380402555454315e-05,
      "loss": 0.46,
      "step": 12993
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4942967891693115,
      "learning_rate": 1.3379444079587348e-05,
      "loss": 0.4642,
      "step": 12994
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5244003534317017,
      "learning_rate": 1.3378485568670141e-05,
      "loss": 0.5193,
      "step": 12995
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5345430374145508,
      "learning_rate": 1.337752702271264e-05,
      "loss": 0.4476,
      "step": 12996
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5442801713943481,
      "learning_rate": 1.3376568441724779e-05,
      "loss": 0.5053,
      "step": 12997
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5650088787078857,
      "learning_rate": 1.3375609825716507e-05,
      "loss": 0.435,
      "step": 12998
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.720318078994751,
      "learning_rate": 1.3374651174697762e-05,
      "loss": 0.5509,
      "step": 12999
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5363821983337402,
      "learning_rate": 1.3373692488678486e-05,
      "loss": 0.4261,
      "step": 13000
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.575124979019165,
      "learning_rate": 1.3372733767668628e-05,
      "loss": 0.4836,
      "step": 13001
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5915007591247559,
      "learning_rate": 1.3371775011678126e-05,
      "loss": 0.4399,
      "step": 13002
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5984086990356445,
      "learning_rate": 1.3370816220716923e-05,
      "loss": 0.5507,
      "step": 13003
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4926074743270874,
      "learning_rate": 1.336985739479497e-05,
      "loss": 0.4188,
      "step": 13004
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5169317722320557,
      "learning_rate": 1.3368898533922202e-05,
      "loss": 0.5125,
      "step": 13005
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6325734853744507,
      "learning_rate": 1.336793963810857e-05,
      "loss": 0.5304,
      "step": 13006
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6435484886169434,
      "learning_rate": 1.336698070736402e-05,
      "loss": 0.5295,
      "step": 13007
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7100485563278198,
      "learning_rate": 1.3366021741698498e-05,
      "loss": 0.4268,
      "step": 13008
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.52047598361969,
      "learning_rate": 1.3365062741121946e-05,
      "loss": 0.4677,
      "step": 13009
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7036648988723755,
      "learning_rate": 1.3364103705644312e-05,
      "loss": 0.4632,
      "step": 13010
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.2424192428588867,
      "learning_rate": 1.3363144635275545e-05,
      "loss": 0.4523,
      "step": 13011
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6011161804199219,
      "learning_rate": 1.3362185530025586e-05,
      "loss": 0.4167,
      "step": 13012
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6351141929626465,
      "learning_rate": 1.336122638990439e-05,
      "loss": 0.4982,
      "step": 13013
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5047180652618408,
      "learning_rate": 1.3360267214921901e-05,
      "loss": 0.4373,
      "step": 13014
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5769480466842651,
      "learning_rate": 1.3359308005088067e-05,
      "loss": 0.4528,
      "step": 13015
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4662535190582275,
      "learning_rate": 1.3358348760412837e-05,
      "loss": 0.4387,
      "step": 13016
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7507679462432861,
      "learning_rate": 1.3357389480906163e-05,
      "loss": 0.9967,
      "step": 13017
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.632095456123352,
      "learning_rate": 1.3356430166577986e-05,
      "loss": 1.0336,
      "step": 13018
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7654056549072266,
      "learning_rate": 1.3355470817438266e-05,
      "loss": 0.4895,
      "step": 13019
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.677349328994751,
      "learning_rate": 1.3354511433496943e-05,
      "loss": 0.443,
      "step": 13020
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.648747205734253,
      "learning_rate": 1.3353552014763978e-05,
      "loss": 0.5445,
      "step": 13021
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5567381381988525,
      "learning_rate": 1.3352592561249311e-05,
      "loss": 0.4133,
      "step": 13022
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4873378276824951,
      "learning_rate": 1.3351633072962901e-05,
      "loss": 0.4581,
      "step": 13023
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.496719479560852,
      "learning_rate": 1.3350673549914696e-05,
      "loss": 0.4546,
      "step": 13024
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5766098499298096,
      "learning_rate": 1.3349713992114649e-05,
      "loss": 0.433,
      "step": 13025
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5449819564819336,
      "learning_rate": 1.334875439957271e-05,
      "loss": 0.4304,
      "step": 13026
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.735019564628601,
      "learning_rate": 1.3347794772298834e-05,
      "loss": 0.5857,
      "step": 13027
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5846768617630005,
      "learning_rate": 1.334683511030297e-05,
      "loss": 0.4765,
      "step": 13028
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5655521154403687,
      "learning_rate": 1.3345875413595079e-05,
      "loss": 0.4819,
      "step": 13029
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5346519947052002,
      "learning_rate": 1.3344915682185104e-05,
      "loss": 0.4403,
      "step": 13030
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5700360536575317,
      "learning_rate": 1.3343955916083009e-05,
      "loss": 0.4673,
      "step": 13031
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5834275484085083,
      "learning_rate": 1.3342996115298743e-05,
      "loss": 0.4076,
      "step": 13032
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6605873107910156,
      "learning_rate": 1.3342036279842261e-05,
      "loss": 0.606,
      "step": 13033
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5891108512878418,
      "learning_rate": 1.3341076409723519e-05,
      "loss": 0.445,
      "step": 13034
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6028625965118408,
      "learning_rate": 1.3340116504952471e-05,
      "loss": 0.462,
      "step": 13035
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5538134574890137,
      "learning_rate": 1.3339156565539073e-05,
      "loss": 0.4458,
      "step": 13036
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5538382530212402,
      "learning_rate": 1.3338196591493285e-05,
      "loss": 0.4903,
      "step": 13037
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5116904973983765,
      "learning_rate": 1.3337236582825057e-05,
      "loss": 0.418,
      "step": 13038
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5539058446884155,
      "learning_rate": 1.333627653954435e-05,
      "loss": 0.4711,
      "step": 13039
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6775152683258057,
      "learning_rate": 1.3335316461661121e-05,
      "loss": 0.411,
      "step": 13040
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5414361953735352,
      "learning_rate": 1.3334356349185326e-05,
      "loss": 0.478,
      "step": 13041
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5725131034851074,
      "learning_rate": 1.3333396202126924e-05,
      "loss": 0.4241,
      "step": 13042
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6233549118041992,
      "learning_rate": 1.3332436020495876e-05,
      "loss": 0.4639,
      "step": 13043
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.638499140739441,
      "learning_rate": 1.3331475804302132e-05,
      "loss": 0.457,
      "step": 13044
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.573391079902649,
      "learning_rate": 1.3330515553555663e-05,
      "loss": 0.5203,
      "step": 13045
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.543565273284912,
      "learning_rate": 1.3329555268266416e-05,
      "loss": 0.4595,
      "step": 13046
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6001944541931152,
      "learning_rate": 1.3328594948444362e-05,
      "loss": 0.5305,
      "step": 13047
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6005369424819946,
      "learning_rate": 1.3327634594099452e-05,
      "loss": 0.437,
      "step": 13048
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.666853904724121,
      "learning_rate": 1.3326674205241653e-05,
      "loss": 0.5022,
      "step": 13049
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5670989751815796,
      "learning_rate": 1.3325713781880919e-05,
      "loss": 0.428,
      "step": 13050
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6131846904754639,
      "learning_rate": 1.3324753324027217e-05,
      "loss": 0.5104,
      "step": 13051
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9609556198120117,
      "learning_rate": 1.3323792831690508e-05,
      "loss": 0.4413,
      "step": 13052
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.538802146911621,
      "learning_rate": 1.3322832304880749e-05,
      "loss": 0.4953,
      "step": 13053
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5453152656555176,
      "learning_rate": 1.3321871743607906e-05,
      "loss": 0.4356,
      "step": 13054
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6343728303909302,
      "learning_rate": 1.3320911147881946e-05,
      "loss": 0.5327,
      "step": 13055
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4828952550888062,
      "learning_rate": 1.331995051771282e-05,
      "loss": 0.4317,
      "step": 13056
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6509065628051758,
      "learning_rate": 1.3318989853110502e-05,
      "loss": 0.5061,
      "step": 13057
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5987125635147095,
      "learning_rate": 1.3318029154084952e-05,
      "loss": 0.4527,
      "step": 13058
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6001527309417725,
      "learning_rate": 1.3317068420646134e-05,
      "loss": 0.4503,
      "step": 13059
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6456565856933594,
      "learning_rate": 1.3316107652804012e-05,
      "loss": 0.4674,
      "step": 13060
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.615936040878296,
      "learning_rate": 1.331514685056855e-05,
      "loss": 0.4937,
      "step": 13061
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.587212085723877,
      "learning_rate": 1.3314186013949713e-05,
      "loss": 0.5073,
      "step": 13062
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5883878469467163,
      "learning_rate": 1.331322514295747e-05,
      "loss": 0.5009,
      "step": 13063
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5519976615905762,
      "learning_rate": 1.3312264237601782e-05,
      "loss": 0.4411,
      "step": 13064
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5014979839324951,
      "learning_rate": 1.3311303297892621e-05,
      "loss": 0.4208,
      "step": 13065
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5050185918807983,
      "learning_rate": 1.3310342323839944e-05,
      "loss": 0.4327,
      "step": 13066
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6464332342147827,
      "learning_rate": 1.3309381315453726e-05,
      "loss": 0.4924,
      "step": 13067
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.720297932624817,
      "learning_rate": 1.3308420272743934e-05,
      "loss": 0.4882,
      "step": 13068
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6513733863830566,
      "learning_rate": 1.330745919572053e-05,
      "loss": 0.5573,
      "step": 13069
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5309652090072632,
      "learning_rate": 1.3306498084393489e-05,
      "loss": 0.4286,
      "step": 13070
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5711487531661987,
      "learning_rate": 1.3305536938772773e-05,
      "loss": 0.4948,
      "step": 13071
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4916259050369263,
      "learning_rate": 1.3304575758868352e-05,
      "loss": 0.4365,
      "step": 13072
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.558259129524231,
      "learning_rate": 1.3303614544690202e-05,
      "loss": 0.4847,
      "step": 13073
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5495262145996094,
      "learning_rate": 1.3302653296248278e-05,
      "loss": 0.4427,
      "step": 13074
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.663832187652588,
      "learning_rate": 1.3301692013552563e-05,
      "loss": 0.4681,
      "step": 13075
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5957828760147095,
      "learning_rate": 1.3300730696613023e-05,
      "loss": 0.4471,
      "step": 13076
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.546481966972351,
      "learning_rate": 1.3299769345439626e-05,
      "loss": 0.4737,
      "step": 13077
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6010419130325317,
      "learning_rate": 1.3298807960042343e-05,
      "loss": 0.4415,
      "step": 13078
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.589888334274292,
      "learning_rate": 1.329784654043115e-05,
      "loss": 0.5406,
      "step": 13079
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.468091607093811,
      "learning_rate": 1.3296885086616009e-05,
      "loss": 0.4216,
      "step": 13080
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6562211513519287,
      "learning_rate": 1.3295923598606904e-05,
      "loss": 0.5075,
      "step": 13081
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6037358045578003,
      "learning_rate": 1.3294962076413795e-05,
      "loss": 0.4927,
      "step": 13082
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8844856023788452,
      "learning_rate": 1.3294000520046666e-05,
      "loss": 0.537,
      "step": 13083
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.611510157585144,
      "learning_rate": 1.3293038929515481e-05,
      "loss": 0.4838,
      "step": 13084
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.608201265335083,
      "learning_rate": 1.3292077304830215e-05,
      "loss": 0.484,
      "step": 13085
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6620333194732666,
      "learning_rate": 1.3291115646000848e-05,
      "loss": 0.4471,
      "step": 13086
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5481278896331787,
      "learning_rate": 1.3290153953037345e-05,
      "loss": 0.4761,
      "step": 13087
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4833980798721313,
      "learning_rate": 1.3289192225949686e-05,
      "loss": 0.4192,
      "step": 13088
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6141114234924316,
      "learning_rate": 1.3288230464747845e-05,
      "loss": 0.4764,
      "step": 13089
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5185840129852295,
      "learning_rate": 1.328726866944179e-05,
      "loss": 0.4285,
      "step": 13090
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5125082731246948,
      "learning_rate": 1.328630684004151e-05,
      "loss": 0.4742,
      "step": 13091
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6664177179336548,
      "learning_rate": 1.3285344976556967e-05,
      "loss": 0.4634,
      "step": 13092
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.572871446609497,
      "learning_rate": 1.3284383078998146e-05,
      "loss": 0.5018,
      "step": 13093
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5379691123962402,
      "learning_rate": 1.328342114737502e-05,
      "loss": 0.4363,
      "step": 13094
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6353262662887573,
      "learning_rate": 1.3282459181697568e-05,
      "loss": 0.4639,
      "step": 13095
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5479310750961304,
      "learning_rate": 1.3281497181975762e-05,
      "loss": 0.4422,
      "step": 13096
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.56040358543396,
      "learning_rate": 1.3280535148219587e-05,
      "loss": 0.4575,
      "step": 13097
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6194614171981812,
      "learning_rate": 1.3279573080439013e-05,
      "loss": 0.4686,
      "step": 13098
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.00748872756958,
      "learning_rate": 1.3278610978644025e-05,
      "loss": 0.5356,
      "step": 13099
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.584917664527893,
      "learning_rate": 1.3277648842844592e-05,
      "loss": 0.4312,
      "step": 13100
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5642215013504028,
      "learning_rate": 1.3276686673050707e-05,
      "loss": 0.4952,
      "step": 13101
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5271040201187134,
      "learning_rate": 1.3275724469272336e-05,
      "loss": 0.4295,
      "step": 13102
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5749832391738892,
      "learning_rate": 1.3274762231519466e-05,
      "loss": 0.5007,
      "step": 13103
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.571824073791504,
      "learning_rate": 1.3273799959802076e-05,
      "loss": 0.4309,
      "step": 13104
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5531278848648071,
      "learning_rate": 1.3272837654130144e-05,
      "loss": 0.4873,
      "step": 13105
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5589689016342163,
      "learning_rate": 1.3271875314513654e-05,
      "loss": 0.4408,
      "step": 13106
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6099462509155273,
      "learning_rate": 1.3270912940962584e-05,
      "loss": 0.4892,
      "step": 13107
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.488638162612915,
      "learning_rate": 1.3269950533486915e-05,
      "loss": 0.4212,
      "step": 13108
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6969773769378662,
      "learning_rate": 1.3268988092096634e-05,
      "loss": 0.5206,
      "step": 13109
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6314677000045776,
      "learning_rate": 1.3268025616801712e-05,
      "loss": 0.4281,
      "step": 13110
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.721528172492981,
      "learning_rate": 1.3267063107612144e-05,
      "loss": 0.5109,
      "step": 13111
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6118052005767822,
      "learning_rate": 1.3266100564537906e-05,
      "loss": 0.4331,
      "step": 13112
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5843653678894043,
      "learning_rate": 1.3265137987588984e-05,
      "loss": 0.5217,
      "step": 13113
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4748666286468506,
      "learning_rate": 1.3264175376775357e-05,
      "loss": 0.4897,
      "step": 13114
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6182489395141602,
      "learning_rate": 1.3263212732107014e-05,
      "loss": 0.4528,
      "step": 13115
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.644124150276184,
      "learning_rate": 1.3262250053593933e-05,
      "loss": 0.4687,
      "step": 13116
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4924893379211426,
      "learning_rate": 1.3261287341246107e-05,
      "loss": 0.5269,
      "step": 13117
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5875056982040405,
      "learning_rate": 1.3260324595073513e-05,
      "loss": 0.4372,
      "step": 13118
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6654138565063477,
      "learning_rate": 1.325936181508614e-05,
      "loss": 0.5011,
      "step": 13119
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5193156003952026,
      "learning_rate": 1.3258399001293976e-05,
      "loss": 0.4292,
      "step": 13120
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.515589952468872,
      "learning_rate": 1.3257436153707e-05,
      "loss": 0.447,
      "step": 13121
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6084024906158447,
      "learning_rate": 1.3256473272335204e-05,
      "loss": 0.4609,
      "step": 13122
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5381911993026733,
      "learning_rate": 1.3255510357188572e-05,
      "loss": 0.4829,
      "step": 13123
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.443131446838379,
      "learning_rate": 1.3254547408277094e-05,
      "loss": 0.3954,
      "step": 13124
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5616708993911743,
      "learning_rate": 1.3253584425610755e-05,
      "loss": 0.4747,
      "step": 13125
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5682181119918823,
      "learning_rate": 1.3252621409199538e-05,
      "loss": 0.4337,
      "step": 13126
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5623016357421875,
      "learning_rate": 1.325165835905344e-05,
      "loss": 0.4844,
      "step": 13127
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4867349863052368,
      "learning_rate": 1.3250695275182444e-05,
      "loss": 0.4248,
      "step": 13128
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5941510200500488,
      "learning_rate": 1.324973215759654e-05,
      "loss": 0.5202,
      "step": 13129
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5161222219467163,
      "learning_rate": 1.3248769006305718e-05,
      "loss": 0.4534,
      "step": 13130
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5645748376846313,
      "learning_rate": 1.3247805821319965e-05,
      "loss": 0.4762,
      "step": 13131
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7044427394866943,
      "learning_rate": 1.3246842602649273e-05,
      "loss": 0.4862,
      "step": 13132
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5667710304260254,
      "learning_rate": 1.324587935030363e-05,
      "loss": 0.4726,
      "step": 13133
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6235988140106201,
      "learning_rate": 1.3244916064293028e-05,
      "loss": 0.4932,
      "step": 13134
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7247244119644165,
      "learning_rate": 1.324395274462746e-05,
      "loss": 0.4709,
      "step": 13135
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5137879848480225,
      "learning_rate": 1.324298939131691e-05,
      "loss": 0.4228,
      "step": 13136
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6688662767410278,
      "learning_rate": 1.3242026004371375e-05,
      "loss": 0.5445,
      "step": 13137
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.501556158065796,
      "learning_rate": 1.324106258380085e-05,
      "loss": 0.4224,
      "step": 13138
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6889156103134155,
      "learning_rate": 1.324009912961532e-05,
      "loss": 1.0212,
      "step": 13139
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.638381838798523,
      "learning_rate": 1.3239135641824783e-05,
      "loss": 1.1291,
      "step": 13140
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6694711446762085,
      "learning_rate": 1.323817212043923e-05,
      "loss": 0.5091,
      "step": 13141
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5336204767227173,
      "learning_rate": 1.3237208565468653e-05,
      "loss": 0.416,
      "step": 13142
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5848971605300903,
      "learning_rate": 1.3236244976923045e-05,
      "loss": 0.4978,
      "step": 13143
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5977799892425537,
      "learning_rate": 1.3235281354812401e-05,
      "loss": 0.4884,
      "step": 13144
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.61062753200531,
      "learning_rate": 1.3234317699146722e-05,
      "loss": 0.5072,
      "step": 13145
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.650847315788269,
      "learning_rate": 1.323335400993599e-05,
      "loss": 0.4721,
      "step": 13146
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6542953252792358,
      "learning_rate": 1.3232390287190209e-05,
      "loss": 0.4815,
      "step": 13147
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.649743914604187,
      "learning_rate": 1.3231426530919373e-05,
      "loss": 0.4668,
      "step": 13148
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6262389421463013,
      "learning_rate": 1.3230462741133474e-05,
      "loss": 0.4825,
      "step": 13149
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4230759143829346,
      "learning_rate": 1.3229498917842514e-05,
      "loss": 0.3925,
      "step": 13150
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9405421018600464,
      "learning_rate": 1.3228535061056482e-05,
      "loss": 0.4606,
      "step": 13151
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6278361082077026,
      "learning_rate": 1.322757117078538e-05,
      "loss": 0.4271,
      "step": 13152
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6174125671386719,
      "learning_rate": 1.3226607247039208e-05,
      "loss": 0.4886,
      "step": 13153
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6098653078079224,
      "learning_rate": 1.3225643289827953e-05,
      "loss": 0.4586,
      "step": 13154
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5993105173110962,
      "learning_rate": 1.322467929916162e-05,
      "loss": 0.4824,
      "step": 13155
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7425572872161865,
      "learning_rate": 1.322371527505021e-05,
      "loss": 0.4928,
      "step": 13156
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8342047929763794,
      "learning_rate": 1.3222751217503715e-05,
      "loss": 0.4863,
      "step": 13157
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5743956565856934,
      "learning_rate": 1.3221787126532137e-05,
      "loss": 0.4298,
      "step": 13158
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6684467792510986,
      "learning_rate": 1.3220823002145475e-05,
      "loss": 0.5141,
      "step": 13159
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5417386293411255,
      "learning_rate": 1.3219858844353728e-05,
      "loss": 0.4413,
      "step": 13160
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.323989152908325,
      "learning_rate": 1.3218894653166896e-05,
      "loss": 0.4765,
      "step": 13161
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.538262128829956,
      "learning_rate": 1.3217930428594976e-05,
      "loss": 0.4278,
      "step": 13162
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6918208599090576,
      "learning_rate": 1.3216966170647978e-05,
      "loss": 0.5332,
      "step": 13163
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5354169607162476,
      "learning_rate": 1.3216001879335892e-05,
      "loss": 0.435,
      "step": 13164
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5041037797927856,
      "learning_rate": 1.3215037554668723e-05,
      "loss": 0.4546,
      "step": 13165
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5443298816680908,
      "learning_rate": 1.3214073196656477e-05,
      "loss": 0.4343,
      "step": 13166
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5837490558624268,
      "learning_rate": 1.3213108805309149e-05,
      "loss": 0.4523,
      "step": 13167
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5480924844741821,
      "learning_rate": 1.3212144380636749e-05,
      "loss": 0.4194,
      "step": 13168
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5992095470428467,
      "learning_rate": 1.3211179922649273e-05,
      "loss": 0.5169,
      "step": 13169
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5559024810791016,
      "learning_rate": 1.3210215431356723e-05,
      "loss": 0.4814,
      "step": 13170
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5531423091888428,
      "learning_rate": 1.3209250906769111e-05,
      "loss": 0.5061,
      "step": 13171
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5046277046203613,
      "learning_rate": 1.320828634889643e-05,
      "loss": 0.4431,
      "step": 13172
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5079163312911987,
      "learning_rate": 1.3207321757748694e-05,
      "loss": 0.4776,
      "step": 13173
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5639276504516602,
      "learning_rate": 1.3206357133335902e-05,
      "loss": 0.4435,
      "step": 13174
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7027671337127686,
      "learning_rate": 1.3205392475668057e-05,
      "loss": 0.4863,
      "step": 13175
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6507863998413086,
      "learning_rate": 1.3204427784755167e-05,
      "loss": 0.4757,
      "step": 13176
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4916670322418213,
      "learning_rate": 1.3203463060607237e-05,
      "loss": 0.4549,
      "step": 13177
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6804007291793823,
      "learning_rate": 1.3202498303234272e-05,
      "loss": 0.4495,
      "step": 13178
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6113407611846924,
      "learning_rate": 1.320153351264628e-05,
      "loss": 0.4945,
      "step": 13179
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4672718048095703,
      "learning_rate": 1.3200568688853266e-05,
      "loss": 0.4155,
      "step": 13180
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6156878471374512,
      "learning_rate": 1.3199603831865235e-05,
      "loss": 0.4897,
      "step": 13181
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6273208856582642,
      "learning_rate": 1.3198638941692196e-05,
      "loss": 0.4472,
      "step": 13182
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.540889024734497,
      "learning_rate": 1.3197674018344158e-05,
      "loss": 0.4943,
      "step": 13183
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6258068084716797,
      "learning_rate": 1.3196709061831125e-05,
      "loss": 0.4475,
      "step": 13184
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6281579732894897,
      "learning_rate": 1.3195744072163108e-05,
      "loss": 0.5375,
      "step": 13185
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5416723489761353,
      "learning_rate": 1.3194779049350117e-05,
      "loss": 0.4413,
      "step": 13186
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6247117519378662,
      "learning_rate": 1.3193813993402156e-05,
      "loss": 0.4781,
      "step": 13187
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5703182220458984,
      "learning_rate": 1.3192848904329235e-05,
      "loss": 0.4763,
      "step": 13188
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5533013343811035,
      "learning_rate": 1.319188378214137e-05,
      "loss": 0.5055,
      "step": 13189
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5410523414611816,
      "learning_rate": 1.3190918626848564e-05,
      "loss": 0.4741,
      "step": 13190
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.2583138942718506,
      "learning_rate": 1.3189953438460827e-05,
      "loss": 0.4646,
      "step": 13191
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6054906845092773,
      "learning_rate": 1.3188988216988175e-05,
      "loss": 0.4427,
      "step": 13192
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.549869418144226,
      "learning_rate": 1.3188022962440614e-05,
      "loss": 0.4978,
      "step": 13193
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6307833194732666,
      "learning_rate": 1.3187057674828156e-05,
      "loss": 0.4551,
      "step": 13194
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.988847017288208,
      "learning_rate": 1.3186092354160818e-05,
      "loss": 0.5034,
      "step": 13195
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4933253526687622,
      "learning_rate": 1.3185127000448601e-05,
      "loss": 0.4026,
      "step": 13196
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6862537860870361,
      "learning_rate": 1.318416161370153e-05,
      "loss": 0.5341,
      "step": 13197
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5983552932739258,
      "learning_rate": 1.3183196193929606e-05,
      "loss": 0.4621,
      "step": 13198
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5328359603881836,
      "learning_rate": 1.3182230741142852e-05,
      "loss": 0.4415,
      "step": 13199
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4485647678375244,
      "learning_rate": 1.3181265255351275e-05,
      "loss": 0.4076,
      "step": 13200
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7851873636245728,
      "learning_rate": 1.3180299736564888e-05,
      "loss": 0.5327,
      "step": 13201
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4672214984893799,
      "learning_rate": 1.3179334184793708e-05,
      "loss": 0.4215,
      "step": 13202
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5697524547576904,
      "learning_rate": 1.317836860004775e-05,
      "loss": 0.4844,
      "step": 13203
    },
    {
      "epoch": 0.42,
      "grad_norm": 12.089326858520508,
      "learning_rate": 1.3177402982337027e-05,
      "loss": 0.4151,
      "step": 13204
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.608860969543457,
      "learning_rate": 1.3176437331671554e-05,
      "loss": 0.4888,
      "step": 13205
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6192100048065186,
      "learning_rate": 1.3175471648061344e-05,
      "loss": 0.4417,
      "step": 13206
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5927608013153076,
      "learning_rate": 1.317450593151642e-05,
      "loss": 0.4857,
      "step": 13207
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5891873836517334,
      "learning_rate": 1.3173540182046787e-05,
      "loss": 0.4379,
      "step": 13208
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6258827447891235,
      "learning_rate": 1.3172574399662473e-05,
      "loss": 0.5003,
      "step": 13209
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6203243732452393,
      "learning_rate": 1.3171608584373488e-05,
      "loss": 0.4867,
      "step": 13210
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6279568672180176,
      "learning_rate": 1.3170642736189849e-05,
      "loss": 0.4645,
      "step": 13211
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6123614311218262,
      "learning_rate": 1.3169676855121575e-05,
      "loss": 0.4695,
      "step": 13212
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5901685953140259,
      "learning_rate": 1.3168710941178685e-05,
      "loss": 0.5011,
      "step": 13213
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6374883651733398,
      "learning_rate": 1.3167744994371192e-05,
      "loss": 0.4871,
      "step": 13214
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5814920663833618,
      "learning_rate": 1.3166779014709122e-05,
      "loss": 0.4843,
      "step": 13215
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.624175786972046,
      "learning_rate": 1.3165813002202485e-05,
      "loss": 0.4587,
      "step": 13216
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5598806142807007,
      "learning_rate": 1.316484695686131e-05,
      "loss": 0.479,
      "step": 13217
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5274267196655273,
      "learning_rate": 1.3163880878695608e-05,
      "loss": 0.4216,
      "step": 13218
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6476677656173706,
      "learning_rate": 1.3162914767715402e-05,
      "loss": 0.5655,
      "step": 13219
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5618007183074951,
      "learning_rate": 1.3161948623930713e-05,
      "loss": 0.4132,
      "step": 13220
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.548274278640747,
      "learning_rate": 1.316098244735156e-05,
      "loss": 0.4771,
      "step": 13221
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5377397537231445,
      "learning_rate": 1.3160016237987967e-05,
      "loss": 0.4249,
      "step": 13222
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5507917404174805,
      "learning_rate": 1.3159049995849951e-05,
      "loss": 0.5036,
      "step": 13223
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5260462760925293,
      "learning_rate": 1.3158083720947531e-05,
      "loss": 0.4311,
      "step": 13224
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6383439302444458,
      "learning_rate": 1.3157117413290739e-05,
      "loss": 0.4734,
      "step": 13225
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.560107707977295,
      "learning_rate": 1.3156151072889584e-05,
      "loss": 0.4292,
      "step": 13226
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6039618253707886,
      "learning_rate": 1.3155184699754103e-05,
      "loss": 0.499,
      "step": 13227
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5640395879745483,
      "learning_rate": 1.3154218293894308e-05,
      "loss": 0.4212,
      "step": 13228
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6306264400482178,
      "learning_rate": 1.3153251855320224e-05,
      "loss": 0.4977,
      "step": 13229
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5191597938537598,
      "learning_rate": 1.3152285384041879e-05,
      "loss": 0.4316,
      "step": 13230
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5486372709274292,
      "learning_rate": 1.315131888006929e-05,
      "loss": 0.4914,
      "step": 13231
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7797495126724243,
      "learning_rate": 1.3150352343412486e-05,
      "loss": 0.4306,
      "step": 13232
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5665310621261597,
      "learning_rate": 1.3149385774081493e-05,
      "loss": 0.4887,
      "step": 13233
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6221470832824707,
      "learning_rate": 1.314841917208633e-05,
      "loss": 0.4177,
      "step": 13234
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6162490844726562,
      "learning_rate": 1.314745253743703e-05,
      "loss": 0.5517,
      "step": 13235
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6409815549850464,
      "learning_rate": 1.314648587014361e-05,
      "loss": 0.4467,
      "step": 13236
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5072227716445923,
      "learning_rate": 1.31455191702161e-05,
      "loss": 0.4817,
      "step": 13237
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6493686437606812,
      "learning_rate": 1.3144552437664531e-05,
      "loss": 0.4525,
      "step": 13238
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5876562595367432,
      "learning_rate": 1.314358567249892e-05,
      "loss": 0.4527,
      "step": 13239
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.629960060119629,
      "learning_rate": 1.3142618874729301e-05,
      "loss": 0.4239,
      "step": 13240
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5672411918640137,
      "learning_rate": 1.31416520443657e-05,
      "loss": 0.445,
      "step": 13241
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.2595021724700928,
      "learning_rate": 1.3140685181418138e-05,
      "loss": 0.4405,
      "step": 13242
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7651339769363403,
      "learning_rate": 1.3139718285896657e-05,
      "loss": 0.4722,
      "step": 13243
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5282491445541382,
      "learning_rate": 1.3138751357811269e-05,
      "loss": 0.4288,
      "step": 13244
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5541046857833862,
      "learning_rate": 1.3137784397172014e-05,
      "loss": 0.4985,
      "step": 13245
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5606828927993774,
      "learning_rate": 1.3136817403988918e-05,
      "loss": 0.4686,
      "step": 13246
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4798232316970825,
      "learning_rate": 1.3135850378272004e-05,
      "loss": 0.4595,
      "step": 13247
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7292944192886353,
      "learning_rate": 1.3134883320031313e-05,
      "loss": 0.4992,
      "step": 13248
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5655444860458374,
      "learning_rate": 1.3133916229276865e-05,
      "loss": 0.4689,
      "step": 13249
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5536638498306274,
      "learning_rate": 1.31329491060187e-05,
      "loss": 0.4298,
      "step": 13250
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.623121976852417,
      "learning_rate": 1.313198195026684e-05,
      "loss": 0.4761,
      "step": 13251
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.537137746810913,
      "learning_rate": 1.3131014762031317e-05,
      "loss": 0.4495,
      "step": 13252
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5988001823425293,
      "learning_rate": 1.3130047541322166e-05,
      "loss": 0.4781,
      "step": 13253
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.55906081199646,
      "learning_rate": 1.3129080288149416e-05,
      "loss": 0.4413,
      "step": 13254
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.526706576347351,
      "learning_rate": 1.3128113002523099e-05,
      "loss": 0.4939,
      "step": 13255
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.543483853340149,
      "learning_rate": 1.312714568445325e-05,
      "loss": 0.4314,
      "step": 13256
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.550545334815979,
      "learning_rate": 1.31261783339499e-05,
      "loss": 0.4644,
      "step": 13257
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.51768159866333,
      "learning_rate": 1.3125210951023082e-05,
      "loss": 0.4268,
      "step": 13258
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.2417585849761963,
      "learning_rate": 1.3124243535682826e-05,
      "loss": 1.0419,
      "step": 13259
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8117966651916504,
      "learning_rate": 1.312327608793917e-05,
      "loss": 1.0558,
      "step": 13260
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7163794040679932,
      "learning_rate": 1.3122308607802145e-05,
      "loss": 0.5142,
      "step": 13261
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6968663930892944,
      "learning_rate": 1.312134109528179e-05,
      "loss": 0.442,
      "step": 13262
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7566031217575073,
      "learning_rate": 1.3120373550388137e-05,
      "loss": 0.5229,
      "step": 13263
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.551666498184204,
      "learning_rate": 1.3119405973131219e-05,
      "loss": 0.4231,
      "step": 13264
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8188899755477905,
      "learning_rate": 1.311843836352107e-05,
      "loss": 0.5695,
      "step": 13265
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5408324003219604,
      "learning_rate": 1.3117470721567732e-05,
      "loss": 0.445,
      "step": 13266
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6580252647399902,
      "learning_rate": 1.3116503047281236e-05,
      "loss": 0.4671,
      "step": 13267
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5448626279830933,
      "learning_rate": 1.311553534067162e-05,
      "loss": 0.4493,
      "step": 13268
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6422239542007446,
      "learning_rate": 1.3114567601748922e-05,
      "loss": 0.4795,
      "step": 13269
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4896633625030518,
      "learning_rate": 1.3113599830523179e-05,
      "loss": 0.44,
      "step": 13270
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.911327600479126,
      "learning_rate": 1.3112632027004421e-05,
      "loss": 0.5171,
      "step": 13271
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4821491241455078,
      "learning_rate": 1.3111664191202697e-05,
      "loss": 0.4276,
      "step": 13272
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5950125455856323,
      "learning_rate": 1.3110696323128037e-05,
      "loss": 0.4772,
      "step": 13273
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6025128364562988,
      "learning_rate": 1.3109728422790484e-05,
      "loss": 0.4439,
      "step": 13274
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.569692850112915,
      "learning_rate": 1.310876049020007e-05,
      "loss": 0.4693,
      "step": 13275
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.429111957550049,
      "learning_rate": 1.3107792525366844e-05,
      "loss": 0.4719,
      "step": 13276
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.637985348701477,
      "learning_rate": 1.3106824528300836e-05,
      "loss": 0.462,
      "step": 13277
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5754499435424805,
      "learning_rate": 1.310585649901209e-05,
      "loss": 0.4267,
      "step": 13278
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5803298950195312,
      "learning_rate": 1.3104888437510648e-05,
      "loss": 0.4937,
      "step": 13279
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6112828254699707,
      "learning_rate": 1.3103920343806546e-05,
      "loss": 0.4423,
      "step": 13280
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5558804273605347,
      "learning_rate": 1.3102952217909827e-05,
      "loss": 0.5087,
      "step": 13281
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.506309986114502,
      "learning_rate": 1.3101984059830532e-05,
      "loss": 0.4219,
      "step": 13282
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6426976919174194,
      "learning_rate": 1.31010158695787e-05,
      "loss": 0.5148,
      "step": 13283
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6196969747543335,
      "learning_rate": 1.310004764716438e-05,
      "loss": 0.4498,
      "step": 13284
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.554103970527649,
      "learning_rate": 1.3099079392597602e-05,
      "loss": 0.4835,
      "step": 13285
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4825224876403809,
      "learning_rate": 1.309811110588842e-05,
      "loss": 0.4452,
      "step": 13286
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5372575521469116,
      "learning_rate": 1.309714278704687e-05,
      "loss": 0.4846,
      "step": 13287
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6036887168884277,
      "learning_rate": 1.3096174436082996e-05,
      "loss": 0.4931,
      "step": 13288
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6467602252960205,
      "learning_rate": 1.3095206053006843e-05,
      "loss": 0.5262,
      "step": 13289
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6295920610427856,
      "learning_rate": 1.3094237637828452e-05,
      "loss": 0.4607,
      "step": 13290
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6700527667999268,
      "learning_rate": 1.3093269190557869e-05,
      "loss": 0.4901,
      "step": 13291
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.530115008354187,
      "learning_rate": 1.3092300711205141e-05,
      "loss": 0.4091,
      "step": 13292
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6073644161224365,
      "learning_rate": 1.3091332199780307e-05,
      "loss": 0.4698,
      "step": 13293
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5919857025146484,
      "learning_rate": 1.309036365629342e-05,
      "loss": 0.4365,
      "step": 13294
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7304030656814575,
      "learning_rate": 1.3089395080754512e-05,
      "loss": 0.562,
      "step": 13295
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.583905816078186,
      "learning_rate": 1.308842647317364e-05,
      "loss": 0.4606,
      "step": 13296
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6233752965927124,
      "learning_rate": 1.3087457833560848e-05,
      "loss": 0.5388,
      "step": 13297
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6155717372894287,
      "learning_rate": 1.3086489161926181e-05,
      "loss": 0.4443,
      "step": 13298
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5796935558319092,
      "learning_rate": 1.3085520458279688e-05,
      "loss": 0.473,
      "step": 13299
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5670439004898071,
      "learning_rate": 1.3084551722631411e-05,
      "loss": 0.4353,
      "step": 13300
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.562994360923767,
      "learning_rate": 1.3083582954991398e-05,
      "loss": 0.4728,
      "step": 13301
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5319558382034302,
      "learning_rate": 1.3082614155369704e-05,
      "loss": 0.4255,
      "step": 13302
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.649209976196289,
      "learning_rate": 1.3081645323776369e-05,
      "loss": 0.5101,
      "step": 13303
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5967873334884644,
      "learning_rate": 1.3080676460221441e-05,
      "loss": 0.4008,
      "step": 13304
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6319465637207031,
      "learning_rate": 1.3079707564714979e-05,
      "loss": 0.5153,
      "step": 13305
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5168178081512451,
      "learning_rate": 1.3078738637267021e-05,
      "loss": 0.413,
      "step": 13306
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.993093490600586,
      "learning_rate": 1.3077769677887619e-05,
      "loss": 1.017,
      "step": 13307
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7170287370681763,
      "learning_rate": 1.3076800686586826e-05,
      "loss": 1.047,
      "step": 13308
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6344324350357056,
      "learning_rate": 1.3075831663374687e-05,
      "loss": 0.5139,
      "step": 13309
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6129465103149414,
      "learning_rate": 1.3074862608261261e-05,
      "loss": 0.4439,
      "step": 13310
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.735944390296936,
      "learning_rate": 1.3073893521256586e-05,
      "loss": 0.5715,
      "step": 13311
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6006395816802979,
      "learning_rate": 1.3072924402370723e-05,
      "loss": 0.4504,
      "step": 13312
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5506428480148315,
      "learning_rate": 1.307195525161372e-05,
      "loss": 0.4879,
      "step": 13313
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5162346363067627,
      "learning_rate": 1.307098606899563e-05,
      "loss": 0.4358,
      "step": 13314
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.585616111755371,
      "learning_rate": 1.3070016854526501e-05,
      "loss": 0.4714,
      "step": 13315
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5720453262329102,
      "learning_rate": 1.306904760821639e-05,
      "loss": 0.4686,
      "step": 13316
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5424249172210693,
      "learning_rate": 1.3068078330075347e-05,
      "loss": 0.4705,
      "step": 13317
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5222645998001099,
      "learning_rate": 1.3067109020113426e-05,
      "loss": 0.4108,
      "step": 13318
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7019357681274414,
      "learning_rate": 1.3066139678340676e-05,
      "loss": 0.5294,
      "step": 13319
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6117043495178223,
      "learning_rate": 1.3065170304767163e-05,
      "loss": 0.4359,
      "step": 13320
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.573542833328247,
      "learning_rate": 1.3064200899402925e-05,
      "loss": 0.5468,
      "step": 13321
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7191566228866577,
      "learning_rate": 1.3063231462258026e-05,
      "loss": 0.4679,
      "step": 13322
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6828559637069702,
      "learning_rate": 1.306226199334252e-05,
      "loss": 0.5244,
      "step": 13323
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.487247347831726,
      "learning_rate": 1.3061292492666459e-05,
      "loss": 0.4177,
      "step": 13324
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7007708549499512,
      "learning_rate": 1.30603229602399e-05,
      "loss": 0.4963,
      "step": 13325
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4811484813690186,
      "learning_rate": 1.3059353396072898e-05,
      "loss": 0.4394,
      "step": 13326
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5601330995559692,
      "learning_rate": 1.3058383800175507e-05,
      "loss": 0.4889,
      "step": 13327
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5154544115066528,
      "learning_rate": 1.3057414172557789e-05,
      "loss": 0.4386,
      "step": 13328
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5171358585357666,
      "learning_rate": 1.3056444513229793e-05,
      "loss": 0.4745,
      "step": 13329
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5074377059936523,
      "learning_rate": 1.3055474822201583e-05,
      "loss": 0.4429,
      "step": 13330
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5948344469070435,
      "learning_rate": 1.3054505099483212e-05,
      "loss": 0.5124,
      "step": 13331
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8342441320419312,
      "learning_rate": 1.305353534508474e-05,
      "loss": 0.4551,
      "step": 13332
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5232858657836914,
      "learning_rate": 1.3052565559016224e-05,
      "loss": 0.4797,
      "step": 13333
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6027476787567139,
      "learning_rate": 1.305159574128772e-05,
      "loss": 0.4575,
      "step": 13334
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5210177898406982,
      "learning_rate": 1.305062589190929e-05,
      "loss": 0.5061,
      "step": 13335
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.547545313835144,
      "learning_rate": 1.304965601089099e-05,
      "loss": 0.4466,
      "step": 13336
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5714266300201416,
      "learning_rate": 1.3048686098242877e-05,
      "loss": 0.5177,
      "step": 13337
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.674862027168274,
      "learning_rate": 1.304771615397502e-05,
      "loss": 0.5006,
      "step": 13338
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7704733610153198,
      "learning_rate": 1.3046746178097468e-05,
      "loss": 0.5264,
      "step": 13339
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5205974578857422,
      "learning_rate": 1.3045776170620292e-05,
      "loss": 0.4326,
      "step": 13340
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.595526933670044,
      "learning_rate": 1.3044806131553543e-05,
      "loss": 0.4845,
      "step": 13341
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5532503128051758,
      "learning_rate": 1.3043836060907287e-05,
      "loss": 0.4776,
      "step": 13342
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.715726613998413,
      "learning_rate": 1.3042865958691583e-05,
      "loss": 0.5273,
      "step": 13343
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5197949409484863,
      "learning_rate": 1.3041895824916493e-05,
      "loss": 0.438,
      "step": 13344
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6273757219314575,
      "learning_rate": 1.3040925659592077e-05,
      "loss": 0.4861,
      "step": 13345
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5629029273986816,
      "learning_rate": 1.3039955462728404e-05,
      "loss": 0.4417,
      "step": 13346
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.58552086353302,
      "learning_rate": 1.3038985234335526e-05,
      "loss": 0.4989,
      "step": 13347
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5127612352371216,
      "learning_rate": 1.3038014974423515e-05,
      "loss": 0.4257,
      "step": 13348
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6388980150222778,
      "learning_rate": 1.3037044683002431e-05,
      "loss": 0.491,
      "step": 13349
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5737982988357544,
      "learning_rate": 1.3036074360082336e-05,
      "loss": 0.4628,
      "step": 13350
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.506155252456665,
      "learning_rate": 1.3035104005673295e-05,
      "loss": 0.471,
      "step": 13351
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5927796363830566,
      "learning_rate": 1.3034133619785374e-05,
      "loss": 0.4857,
      "step": 13352
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5336594581604004,
      "learning_rate": 1.3033163202428634e-05,
      "loss": 0.4726,
      "step": 13353
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5200531482696533,
      "learning_rate": 1.3032192753613142e-05,
      "loss": 0.4256,
      "step": 13354
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6136753559112549,
      "learning_rate": 1.303122227334896e-05,
      "loss": 0.5032,
      "step": 13355
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.467988133430481,
      "learning_rate": 1.303025176164616e-05,
      "loss": 0.4329,
      "step": 13356
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6237605810165405,
      "learning_rate": 1.3029281218514797e-05,
      "loss": 0.5541,
      "step": 13357
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.563098430633545,
      "learning_rate": 1.302831064396495e-05,
      "loss": 0.424,
      "step": 13358
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5562937259674072,
      "learning_rate": 1.3027340038006678e-05,
      "loss": 0.4622,
      "step": 13359
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5788967609405518,
      "learning_rate": 1.3026369400650048e-05,
      "loss": 0.44,
      "step": 13360
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7376352548599243,
      "learning_rate": 1.3025398731905127e-05,
      "loss": 0.5275,
      "step": 13361
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5004727840423584,
      "learning_rate": 1.3024428031781984e-05,
      "loss": 0.4117,
      "step": 13362
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6840153932571411,
      "learning_rate": 1.3023457300290685e-05,
      "loss": 0.5521,
      "step": 13363
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6509603261947632,
      "learning_rate": 1.3022486537441302e-05,
      "loss": 0.5122,
      "step": 13364
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6787809133529663,
      "learning_rate": 1.3021515743243896e-05,
      "loss": 0.4906,
      "step": 13365
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.49312162399292,
      "learning_rate": 1.3020544917708543e-05,
      "loss": 0.4364,
      "step": 13366
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6568552255630493,
      "learning_rate": 1.3019574060845308e-05,
      "loss": 0.5407,
      "step": 13367
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4611233472824097,
      "learning_rate": 1.3018603172664264e-05,
      "loss": 0.416,
      "step": 13368
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.579675316810608,
      "learning_rate": 1.3017632253175476e-05,
      "loss": 0.4882,
      "step": 13369
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5213475227355957,
      "learning_rate": 1.3016661302389016e-05,
      "loss": 0.4275,
      "step": 13370
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5359734296798706,
      "learning_rate": 1.3015690320314952e-05,
      "loss": 0.4562,
      "step": 13371
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4760290384292603,
      "learning_rate": 1.3014719306963364e-05,
      "loss": 0.4172,
      "step": 13372
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6708910465240479,
      "learning_rate": 1.3013748262344308e-05,
      "loss": 0.5032,
      "step": 13373
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.591485857963562,
      "learning_rate": 1.3012777186467872e-05,
      "loss": 0.4191,
      "step": 13374
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5387647151947021,
      "learning_rate": 1.3011806079344112e-05,
      "loss": 0.4657,
      "step": 13375
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.562506914138794,
      "learning_rate": 1.3010834940983108e-05,
      "loss": 0.4652,
      "step": 13376
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.647855281829834,
      "learning_rate": 1.3009863771394932e-05,
      "loss": 0.5061,
      "step": 13377
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.678575038909912,
      "learning_rate": 1.3008892570589655e-05,
      "loss": 0.4657,
      "step": 13378
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6205692291259766,
      "learning_rate": 1.3007921338577352e-05,
      "loss": 0.4755,
      "step": 13379
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5360296964645386,
      "learning_rate": 1.3006950075368095e-05,
      "loss": 0.4594,
      "step": 13380
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6700679063796997,
      "learning_rate": 1.3005978780971954e-05,
      "loss": 0.5448,
      "step": 13381
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6132080554962158,
      "learning_rate": 1.300500745539901e-05,
      "loss": 0.4261,
      "step": 13382
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6713507175445557,
      "learning_rate": 1.3004036098659331e-05,
      "loss": 0.5052,
      "step": 13383
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5053125619888306,
      "learning_rate": 1.3003064710762994e-05,
      "loss": 0.4312,
      "step": 13384
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6458654403686523,
      "learning_rate": 1.3002093291720078e-05,
      "loss": 0.4682,
      "step": 13385
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5098905563354492,
      "learning_rate": 1.3001121841540649e-05,
      "loss": 0.4162,
      "step": 13386
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.476467251777649,
      "learning_rate": 1.3000150360234789e-05,
      "loss": 0.4419,
      "step": 13387
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5389121770858765,
      "learning_rate": 1.2999178847812572e-05,
      "loss": 0.4477,
      "step": 13388
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5600744485855103,
      "learning_rate": 1.2998207304284073e-05,
      "loss": 0.4881,
      "step": 13389
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5563913583755493,
      "learning_rate": 1.2997235729659377e-05,
      "loss": 0.4569,
      "step": 13390
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.579345703125,
      "learning_rate": 1.2996264123948545e-05,
      "loss": 0.5018,
      "step": 13391
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6629902124404907,
      "learning_rate": 1.2995292487161668e-05,
      "loss": 0.4677,
      "step": 13392
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6620210409164429,
      "learning_rate": 1.2994320819308815e-05,
      "loss": 0.5058,
      "step": 13393
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.511804223060608,
      "learning_rate": 1.2993349120400067e-05,
      "loss": 0.4505,
      "step": 13394
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.704365849494934,
      "learning_rate": 1.2992377390445505e-05,
      "loss": 0.5416,
      "step": 13395
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5163159370422363,
      "learning_rate": 1.2991405629455202e-05,
      "loss": 0.4238,
      "step": 13396
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.596336841583252,
      "learning_rate": 1.299043383743924e-05,
      "loss": 0.5479,
      "step": 13397
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6462829113006592,
      "learning_rate": 1.29894620144077e-05,
      "loss": 0.4458,
      "step": 13398
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6538283824920654,
      "learning_rate": 1.2988490160370652e-05,
      "loss": 0.5262,
      "step": 13399
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5814945697784424,
      "learning_rate": 1.2987518275338188e-05,
      "loss": 0.4351,
      "step": 13400
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5953665971755981,
      "learning_rate": 1.2986546359320378e-05,
      "loss": 0.5192,
      "step": 13401
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6242916584014893,
      "learning_rate": 1.298557441232731e-05,
      "loss": 0.4616,
      "step": 13402
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6433526277542114,
      "learning_rate": 1.2984602434369058e-05,
      "loss": 0.5302,
      "step": 13403
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5671298503875732,
      "learning_rate": 1.2983630425455709e-05,
      "loss": 0.4418,
      "step": 13404
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.510567307472229,
      "learning_rate": 1.298265838559734e-05,
      "loss": 0.4749,
      "step": 13405
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6487045288085938,
      "learning_rate": 1.2981686314804037e-05,
      "loss": 0.4413,
      "step": 13406
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.703683853149414,
      "learning_rate": 1.2980714213085872e-05,
      "loss": 0.5007,
      "step": 13407
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5271419286727905,
      "learning_rate": 1.2979742080452941e-05,
      "loss": 0.4264,
      "step": 13408
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6147692203521729,
      "learning_rate": 1.2978769916915316e-05,
      "loss": 0.4991,
      "step": 13409
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6088135242462158,
      "learning_rate": 1.2977797722483088e-05,
      "loss": 0.4154,
      "step": 13410
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6539949178695679,
      "learning_rate": 1.297682549716633e-05,
      "loss": 0.5447,
      "step": 13411
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5075362920761108,
      "learning_rate": 1.2975853240975134e-05,
      "loss": 0.4481,
      "step": 13412
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6518012285232544,
      "learning_rate": 1.2974880953919581e-05,
      "loss": 0.4965,
      "step": 13413
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6862165927886963,
      "learning_rate": 1.2973908636009757e-05,
      "loss": 0.5145,
      "step": 13414
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.709236979484558,
      "learning_rate": 1.2972936287255745e-05,
      "loss": 0.5832,
      "step": 13415
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7317478656768799,
      "learning_rate": 1.2971963907667626e-05,
      "loss": 0.4468,
      "step": 13416
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5575271844863892,
      "learning_rate": 1.2970991497255491e-05,
      "loss": 0.4831,
      "step": 13417
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.076634168624878,
      "learning_rate": 1.2970019056029425e-05,
      "loss": 0.4821,
      "step": 13418
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6432417631149292,
      "learning_rate": 1.296904658399951e-05,
      "loss": 0.5046,
      "step": 13419
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6072523593902588,
      "learning_rate": 1.2968074081175834e-05,
      "loss": 0.4374,
      "step": 13420
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5878640413284302,
      "learning_rate": 1.2967101547568485e-05,
      "loss": 0.4685,
      "step": 13421
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5306274890899658,
      "learning_rate": 1.2966128983187548e-05,
      "loss": 0.4231,
      "step": 13422
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5412344932556152,
      "learning_rate": 1.2965156388043112e-05,
      "loss": 0.5157,
      "step": 13423
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.498653769493103,
      "learning_rate": 1.2964183762145259e-05,
      "loss": 0.4187,
      "step": 13424
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5875256061553955,
      "learning_rate": 1.2963211105504083e-05,
      "loss": 0.4801,
      "step": 13425
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5743670463562012,
      "learning_rate": 1.296223841812967e-05,
      "loss": 0.426,
      "step": 13426
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5901505947113037,
      "learning_rate": 1.2961265700032104e-05,
      "loss": 0.4609,
      "step": 13427
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6340500116348267,
      "learning_rate": 1.2960292951221481e-05,
      "loss": 0.4581,
      "step": 13428
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.2082645893096924,
      "learning_rate": 1.2959320171707886e-05,
      "loss": 0.5148,
      "step": 13429
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4598602056503296,
      "learning_rate": 1.2958347361501406e-05,
      "loss": 0.4349,
      "step": 13430
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6556053161621094,
      "learning_rate": 1.2957374520612138e-05,
      "loss": 0.4843,
      "step": 13431
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5853949785232544,
      "learning_rate": 1.2956401649050163e-05,
      "loss": 0.4354,
      "step": 13432
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5451719760894775,
      "learning_rate": 1.2955428746825578e-05,
      "loss": 0.4751,
      "step": 13433
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9351446628570557,
      "learning_rate": 1.295445581394847e-05,
      "loss": 0.4366,
      "step": 13434
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5454304218292236,
      "learning_rate": 1.2953482850428927e-05,
      "loss": 0.5166,
      "step": 13435
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5620957612991333,
      "learning_rate": 1.2952509856277051e-05,
      "loss": 0.4256,
      "step": 13436
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4987679719924927,
      "learning_rate": 1.295153683150292e-05,
      "loss": 0.4482,
      "step": 13437
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5609838962554932,
      "learning_rate": 1.2950563776116636e-05,
      "loss": 0.4355,
      "step": 13438
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5114219188690186,
      "learning_rate": 1.2949590690128286e-05,
      "loss": 0.4727,
      "step": 13439
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5573229789733887,
      "learning_rate": 1.2948617573547964e-05,
      "loss": 0.4202,
      "step": 13440
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6385117769241333,
      "learning_rate": 1.2947644426385764e-05,
      "loss": 0.4849,
      "step": 13441
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4630444049835205,
      "learning_rate": 1.2946671248651778e-05,
      "loss": 0.4013,
      "step": 13442
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5223920345306396,
      "learning_rate": 1.2945698040356095e-05,
      "loss": 0.4733,
      "step": 13443
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4405529499053955,
      "learning_rate": 1.2944724801508818e-05,
      "loss": 0.4141,
      "step": 13444
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6464266777038574,
      "learning_rate": 1.2943751532120032e-05,
      "loss": 0.479,
      "step": 13445
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4896478652954102,
      "learning_rate": 1.294277823219984e-05,
      "loss": 0.4466,
      "step": 13446
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6873013973236084,
      "learning_rate": 1.2941804901758328e-05,
      "loss": 0.4951,
      "step": 13447
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6038731336593628,
      "learning_rate": 1.2940831540805596e-05,
      "loss": 0.4723,
      "step": 13448
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.535903811454773,
      "learning_rate": 1.2939858149351739e-05,
      "loss": 0.4678,
      "step": 13449
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6638245582580566,
      "learning_rate": 1.2938884727406854e-05,
      "loss": 0.4403,
      "step": 13450
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.296286106109619,
      "learning_rate": 1.293791127498103e-05,
      "loss": 0.467,
      "step": 13451
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5304208993911743,
      "learning_rate": 1.2936937792084373e-05,
      "loss": 0.4381,
      "step": 13452
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6639870405197144,
      "learning_rate": 1.2935964278726968e-05,
      "loss": 0.4947,
      "step": 13453
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5289337635040283,
      "learning_rate": 1.2934990734918926e-05,
      "loss": 0.4083,
      "step": 13454
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6479064226150513,
      "learning_rate": 1.293401716067033e-05,
      "loss": 0.4672,
      "step": 13455
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5270588397979736,
      "learning_rate": 1.2933043555991289e-05,
      "loss": 0.4307,
      "step": 13456
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5660905838012695,
      "learning_rate": 1.2932069920891897e-05,
      "loss": 0.4893,
      "step": 13457
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5932538509368896,
      "learning_rate": 1.293109625538225e-05,
      "loss": 0.4484,
      "step": 13458
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5560294389724731,
      "learning_rate": 1.2930122559472446e-05,
      "loss": 0.4737,
      "step": 13459
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6659719944000244,
      "learning_rate": 1.2929148833172587e-05,
      "loss": 0.437,
      "step": 13460
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5889450311660767,
      "learning_rate": 1.2928175076492768e-05,
      "loss": 0.5448,
      "step": 13461
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.512148141860962,
      "learning_rate": 1.2927201289443096e-05,
      "loss": 0.414,
      "step": 13462
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5267915725708008,
      "learning_rate": 1.292622747203366e-05,
      "loss": 0.4888,
      "step": 13463
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6453936100006104,
      "learning_rate": 1.292525362427457e-05,
      "loss": 0.4492,
      "step": 13464
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8160104751586914,
      "learning_rate": 1.2924279746175921e-05,
      "loss": 0.9903,
      "step": 13465
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6027454137802124,
      "learning_rate": 1.2923305837747817e-05,
      "loss": 1.0376,
      "step": 13466
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.596221923828125,
      "learning_rate": 1.2922331899000353e-05,
      "loss": 0.4909,
      "step": 13467
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5607783794403076,
      "learning_rate": 1.2921357929943636e-05,
      "loss": 0.4348,
      "step": 13468
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5821070671081543,
      "learning_rate": 1.2920383930587768e-05,
      "loss": 0.5021,
      "step": 13469
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.519888997077942,
      "learning_rate": 1.2919409900942847e-05,
      "loss": 0.4167,
      "step": 13470
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5998352766036987,
      "learning_rate": 1.2918435841018975e-05,
      "loss": 0.5056,
      "step": 13471
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5482505559921265,
      "learning_rate": 1.2917461750826259e-05,
      "loss": 0.4348,
      "step": 13472
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6256994009017944,
      "learning_rate": 1.2916487630374796e-05,
      "loss": 0.5042,
      "step": 13473
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5589942932128906,
      "learning_rate": 1.2915513479674698e-05,
      "loss": 0.4149,
      "step": 13474
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9135459661483765,
      "learning_rate": 1.291453929873606e-05,
      "loss": 0.5396,
      "step": 13475
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.649517297744751,
      "learning_rate": 1.2913565087568989e-05,
      "loss": 0.4716,
      "step": 13476
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.484947919845581,
      "learning_rate": 1.2912590846183588e-05,
      "loss": 0.481,
      "step": 13477
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.605940341949463,
      "learning_rate": 1.2911616574589964e-05,
      "loss": 0.4369,
      "step": 13478
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5665080547332764,
      "learning_rate": 1.2910642272798217e-05,
      "loss": 0.4904,
      "step": 13479
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.533793330192566,
      "learning_rate": 1.2909667940818462e-05,
      "loss": 0.4085,
      "step": 13480
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6226316690444946,
      "learning_rate": 1.2908693578660792e-05,
      "loss": 0.4682,
      "step": 13481
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5595052242279053,
      "learning_rate": 1.290771918633532e-05,
      "loss": 0.447,
      "step": 13482
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5867531299591064,
      "learning_rate": 1.290674476385215e-05,
      "loss": 0.5153,
      "step": 13483
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6096757650375366,
      "learning_rate": 1.2905770311221388e-05,
      "loss": 0.49,
      "step": 13484
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.558813214302063,
      "learning_rate": 1.2904795828453143e-05,
      "loss": 0.5043,
      "step": 13485
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5916584730148315,
      "learning_rate": 1.2903821315557519e-05,
      "loss": 0.4572,
      "step": 13486
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.509669542312622,
      "learning_rate": 1.2902846772544625e-05,
      "loss": 0.4384,
      "step": 13487
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4568120241165161,
      "learning_rate": 1.2901872199424567e-05,
      "loss": 0.4083,
      "step": 13488
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7044711112976074,
      "learning_rate": 1.2900897596207456e-05,
      "loss": 0.5224,
      "step": 13489
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5730668306350708,
      "learning_rate": 1.2899922962903397e-05,
      "loss": 0.4141,
      "step": 13490
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7114177942276,
      "learning_rate": 1.2898948299522501e-05,
      "loss": 0.5091,
      "step": 13491
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5758506059646606,
      "learning_rate": 1.2897973606074875e-05,
      "loss": 0.4132,
      "step": 13492
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.573873519897461,
      "learning_rate": 1.2896998882570628e-05,
      "loss": 0.5232,
      "step": 13493
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5049887895584106,
      "learning_rate": 1.2896024129019868e-05,
      "loss": 0.4528,
      "step": 13494
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.580851674079895,
      "learning_rate": 1.2895049345432711e-05,
      "loss": 0.4923,
      "step": 13495
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5129191875457764,
      "learning_rate": 1.2894074531819258e-05,
      "loss": 0.427,
      "step": 13496
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6071349382400513,
      "learning_rate": 1.2893099688189627e-05,
      "loss": 0.4941,
      "step": 13497
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6223489046096802,
      "learning_rate": 1.2892124814553925e-05,
      "loss": 0.4123,
      "step": 13498
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.644358515739441,
      "learning_rate": 1.2891149910922267e-05,
      "loss": 0.4683,
      "step": 13499
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8362360000610352,
      "learning_rate": 1.2890174977304759e-05,
      "loss": 0.4479,
      "step": 13500
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.675367832183838,
      "learning_rate": 1.2889200013711515e-05,
      "loss": 0.5097,
      "step": 13501
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.659291386604309,
      "learning_rate": 1.2888225020152645e-05,
      "loss": 0.4571,
      "step": 13502
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7059555053710938,
      "learning_rate": 1.2887249996638266e-05,
      "loss": 0.4773,
      "step": 13503
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6482678651809692,
      "learning_rate": 1.2886274943178484e-05,
      "loss": 0.4502,
      "step": 13504
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5103282928466797,
      "learning_rate": 1.288529985978342e-05,
      "loss": 0.4786,
      "step": 13505
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.610324740409851,
      "learning_rate": 1.288432474646318e-05,
      "loss": 0.4285,
      "step": 13506
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.616369605064392,
      "learning_rate": 1.2883349603227881e-05,
      "loss": 0.4926,
      "step": 13507
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.518005132675171,
      "learning_rate": 1.2882374430087636e-05,
      "loss": 0.4507,
      "step": 13508
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5530275106430054,
      "learning_rate": 1.288139922705256e-05,
      "loss": 0.4869,
      "step": 13509
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5713465213775635,
      "learning_rate": 1.2880423994132764e-05,
      "loss": 0.4393,
      "step": 13510
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.25246524810791,
      "learning_rate": 1.2879448731338366e-05,
      "loss": 0.4823,
      "step": 13511
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.618744969367981,
      "learning_rate": 1.2878473438679478e-05,
      "loss": 0.4424,
      "step": 13512
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5701603889465332,
      "learning_rate": 1.2877498116166223e-05,
      "loss": 0.5375,
      "step": 13513
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5062861442565918,
      "learning_rate": 1.2876522763808706e-05,
      "loss": 0.4486,
      "step": 13514
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6494340896606445,
      "learning_rate": 1.2875547381617051e-05,
      "loss": 0.5523,
      "step": 13515
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4986070394515991,
      "learning_rate": 1.287457196960137e-05,
      "loss": 0.4425,
      "step": 13516
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5567762851715088,
      "learning_rate": 1.2873596527771781e-05,
      "loss": 0.4454,
      "step": 13517
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.589282751083374,
      "learning_rate": 1.2872621056138401e-05,
      "loss": 0.4581,
      "step": 13518
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.511060357093811,
      "learning_rate": 1.2871645554711349e-05,
      "loss": 0.4995,
      "step": 13519
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6521313190460205,
      "learning_rate": 1.2870670023500738e-05,
      "loss": 0.4405,
      "step": 13520
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6068581342697144,
      "learning_rate": 1.2869694462516691e-05,
      "loss": 0.4925,
      "step": 13521
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5569031238555908,
      "learning_rate": 1.2868718871769322e-05,
      "loss": 0.4121,
      "step": 13522
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8630235195159912,
      "learning_rate": 1.2867743251268749e-05,
      "loss": 1.0043,
      "step": 13523
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6827054023742676,
      "learning_rate": 1.2866767601025096e-05,
      "loss": 1.0755,
      "step": 13524
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6722562313079834,
      "learning_rate": 1.2865791921048479e-05,
      "loss": 0.4571,
      "step": 13525
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5765575170516968,
      "learning_rate": 1.2864816211349015e-05,
      "loss": 0.4234,
      "step": 13526
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5758692026138306,
      "learning_rate": 1.2863840471936826e-05,
      "loss": 0.4741,
      "step": 13527
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5598986148834229,
      "learning_rate": 1.2862864702822034e-05,
      "loss": 0.4469,
      "step": 13528
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5797486305236816,
      "learning_rate": 1.2861888904014754e-05,
      "loss": 0.5121,
      "step": 13529
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6029953956604004,
      "learning_rate": 1.286091307552511e-05,
      "loss": 0.4608,
      "step": 13530
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6352710723876953,
      "learning_rate": 1.2859937217363225e-05,
      "loss": 0.4482,
      "step": 13531
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5742933750152588,
      "learning_rate": 1.2858961329539215e-05,
      "loss": 0.4279,
      "step": 13532
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.610568642616272,
      "learning_rate": 1.2857985412063206e-05,
      "loss": 0.5104,
      "step": 13533
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.561726450920105,
      "learning_rate": 1.2857009464945318e-05,
      "loss": 0.4399,
      "step": 13534
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5740489959716797,
      "learning_rate": 1.2856033488195675e-05,
      "loss": 0.4645,
      "step": 13535
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6106001138687134,
      "learning_rate": 1.2855057481824393e-05,
      "loss": 0.4329,
      "step": 13536
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.497769832611084,
      "learning_rate": 1.2854081445841603e-05,
      "loss": 0.4348,
      "step": 13537
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4967243671417236,
      "learning_rate": 1.2853105380257423e-05,
      "loss": 0.4076,
      "step": 13538
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6318445205688477,
      "learning_rate": 1.285212928508198e-05,
      "loss": 0.4866,
      "step": 13539
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5528112649917603,
      "learning_rate": 1.2851153160325393e-05,
      "loss": 0.4347,
      "step": 13540
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6341538429260254,
      "learning_rate": 1.2850177005997791e-05,
      "loss": 0.5506,
      "step": 13541
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5372291803359985,
      "learning_rate": 1.2849200822109293e-05,
      "loss": 0.3948,
      "step": 13542
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5556995868682861,
      "learning_rate": 1.2848224608670028e-05,
      "loss": 0.4419,
      "step": 13543
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6021267175674438,
      "learning_rate": 1.2847248365690121e-05,
      "loss": 0.5067,
      "step": 13544
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5948491096496582,
      "learning_rate": 1.2846272093179692e-05,
      "loss": 0.5376,
      "step": 13545
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5582406520843506,
      "learning_rate": 1.2845295791148873e-05,
      "loss": 0.4128,
      "step": 13546
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6330424547195435,
      "learning_rate": 1.2844319459607785e-05,
      "loss": 0.4605,
      "step": 13547
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.581087350845337,
      "learning_rate": 1.2843343098566555e-05,
      "loss": 0.4776,
      "step": 13548
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.524730920791626,
      "learning_rate": 1.2842366708035314e-05,
      "loss": 0.4989,
      "step": 13549
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6351335048675537,
      "learning_rate": 1.284139028802418e-05,
      "loss": 0.456,
      "step": 13550
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4993367195129395,
      "learning_rate": 1.2840413838543289e-05,
      "loss": 0.4742,
      "step": 13551
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6094846725463867,
      "learning_rate": 1.2839437359602763e-05,
      "loss": 0.4955,
      "step": 13552
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5920815467834473,
      "learning_rate": 1.2838460851212731e-05,
      "loss": 0.4993,
      "step": 13553
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6812472343444824,
      "learning_rate": 1.2837484313383322e-05,
      "loss": 0.464,
      "step": 13554
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6353265047073364,
      "learning_rate": 1.2836507746124663e-05,
      "loss": 0.4939,
      "step": 13555
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6560084819793701,
      "learning_rate": 1.283553114944688e-05,
      "loss": 0.4386,
      "step": 13556
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5611521005630493,
      "learning_rate": 1.283455452336011e-05,
      "loss": 0.4777,
      "step": 13557
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.070793390274048,
      "learning_rate": 1.2833577867874471e-05,
      "loss": 0.4476,
      "step": 13558
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5835052728652954,
      "learning_rate": 1.2832601183000102e-05,
      "loss": 0.4984,
      "step": 13559
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.648734450340271,
      "learning_rate": 1.2831624468747128e-05,
      "loss": 0.4608,
      "step": 13560
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5474121570587158,
      "learning_rate": 1.2830647725125683e-05,
      "loss": 0.5128,
      "step": 13561
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5584075450897217,
      "learning_rate": 1.2829670952145892e-05,
      "loss": 0.4479,
      "step": 13562
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5183815956115723,
      "learning_rate": 1.2828694149817889e-05,
      "loss": 0.4588,
      "step": 13563
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5378299951553345,
      "learning_rate": 1.2827717318151801e-05,
      "loss": 0.4531,
      "step": 13564
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6099509000778198,
      "learning_rate": 1.2826740457157771e-05,
      "loss": 0.4828,
      "step": 13565
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6212177276611328,
      "learning_rate": 1.2825763566845914e-05,
      "loss": 0.4209,
      "step": 13566
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5069091320037842,
      "learning_rate": 1.2824786647226377e-05,
      "loss": 0.4671,
      "step": 13567
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5526340007781982,
      "learning_rate": 1.282380969830928e-05,
      "loss": 0.4622,
      "step": 13568
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5678354501724243,
      "learning_rate": 1.2822832720104761e-05,
      "loss": 0.4882,
      "step": 13569
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4516364336013794,
      "learning_rate": 1.2821855712622954e-05,
      "loss": 0.423,
      "step": 13570
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7133122682571411,
      "learning_rate": 1.2820878675873993e-05,
      "loss": 0.9849,
      "step": 13571
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4916774034500122,
      "learning_rate": 1.2819901609868008e-05,
      "loss": 1.0596,
      "step": 13572
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6916638612747192,
      "learning_rate": 1.2818924514615134e-05,
      "loss": 0.5091,
      "step": 13573
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6506166458129883,
      "learning_rate": 1.2817947390125503e-05,
      "loss": 0.438,
      "step": 13574
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.628448247909546,
      "learning_rate": 1.2816970236409256e-05,
      "loss": 0.4941,
      "step": 13575
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5473108291625977,
      "learning_rate": 1.2815993053476519e-05,
      "loss": 0.4136,
      "step": 13576
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.517638087272644,
      "learning_rate": 1.2815015841337433e-05,
      "loss": 0.4832,
      "step": 13577
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5551598072052002,
      "learning_rate": 1.281403860000213e-05,
      "loss": 0.4505,
      "step": 13578
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6337072849273682,
      "learning_rate": 1.2813061329480747e-05,
      "loss": 0.5081,
      "step": 13579
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7760603427886963,
      "learning_rate": 1.281208402978342e-05,
      "loss": 0.5285,
      "step": 13580
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6146224737167358,
      "learning_rate": 1.2811106700920285e-05,
      "loss": 0.5128,
      "step": 13581
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6063611507415771,
      "learning_rate": 1.2810129342901479e-05,
      "loss": 0.4754,
      "step": 13582
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.678169846534729,
      "learning_rate": 1.2809151955737141e-05,
      "loss": 0.5195,
      "step": 13583
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.476165533065796,
      "learning_rate": 1.2808174539437397e-05,
      "loss": 0.412,
      "step": 13584
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5115249156951904,
      "learning_rate": 1.28071970940124e-05,
      "loss": 0.4337,
      "step": 13585
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.767914891242981,
      "learning_rate": 1.2806219619472276e-05,
      "loss": 0.475,
      "step": 13586
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5191072225570679,
      "learning_rate": 1.2805242115827169e-05,
      "loss": 0.5435,
      "step": 13587
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5678097009658813,
      "learning_rate": 1.2804264583087215e-05,
      "loss": 0.4334,
      "step": 13588
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5905499458312988,
      "learning_rate": 1.2803287021262554e-05,
      "loss": 0.4614,
      "step": 13589
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.494913101196289,
      "learning_rate": 1.2802309430363323e-05,
      "loss": 0.4343,
      "step": 13590
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4988386631011963,
      "learning_rate": 1.2801331810399662e-05,
      "loss": 0.4795,
      "step": 13591
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4628723859786987,
      "learning_rate": 1.2800354161381709e-05,
      "loss": 0.4175,
      "step": 13592
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6110374927520752,
      "learning_rate": 1.279937648331961e-05,
      "loss": 0.486,
      "step": 13593
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.593368411064148,
      "learning_rate": 1.2798398776223496e-05,
      "loss": 0.4504,
      "step": 13594
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5358034372329712,
      "learning_rate": 1.2797421040103513e-05,
      "loss": 0.4721,
      "step": 13595
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5470340251922607,
      "learning_rate": 1.27964432749698e-05,
      "loss": 0.467,
      "step": 13596
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5495682954788208,
      "learning_rate": 1.2795465480832499e-05,
      "loss": 0.4909,
      "step": 13597
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4574247598648071,
      "learning_rate": 1.2794487657701751e-05,
      "loss": 0.4168,
      "step": 13598
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7227363586425781,
      "learning_rate": 1.2793509805587698e-05,
      "loss": 0.5691,
      "step": 13599
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5979493856430054,
      "learning_rate": 1.2792531924500478e-05,
      "loss": 0.4384,
      "step": 13600
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.749621868133545,
      "learning_rate": 1.2791554014450242e-05,
      "loss": 0.4469,
      "step": 13601
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6379302740097046,
      "learning_rate": 1.2790576075447124e-05,
      "loss": 0.4083,
      "step": 13602
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6427521705627441,
      "learning_rate": 1.2789598107501271e-05,
      "loss": 0.484,
      "step": 13603
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5422550439834595,
      "learning_rate": 1.2788620110622821e-05,
      "loss": 0.4479,
      "step": 13604
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.743161916732788,
      "learning_rate": 1.2787642084821923e-05,
      "loss": 0.4848,
      "step": 13605
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5843724012374878,
      "learning_rate": 1.2786664030108722e-05,
      "loss": 0.4365,
      "step": 13606
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5261270999908447,
      "learning_rate": 1.2785685946493356e-05,
      "loss": 0.4688,
      "step": 13607
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6257859468460083,
      "learning_rate": 1.2784707833985972e-05,
      "loss": 0.4194,
      "step": 13608
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.639843225479126,
      "learning_rate": 1.2783729692596718e-05,
      "loss": 0.5445,
      "step": 13609
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4676611423492432,
      "learning_rate": 1.278275152233573e-05,
      "loss": 0.3983,
      "step": 13610
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6344871520996094,
      "learning_rate": 1.2781773323213164e-05,
      "loss": 0.5011,
      "step": 13611
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5728116035461426,
      "learning_rate": 1.2780795095239157e-05,
      "loss": 0.4419,
      "step": 13612
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6106599569320679,
      "learning_rate": 1.277981683842386e-05,
      "loss": 0.5204,
      "step": 13613
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5877723693847656,
      "learning_rate": 1.2778838552777417e-05,
      "loss": 0.4444,
      "step": 13614
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.475411057472229,
      "learning_rate": 1.2777860238309976e-05,
      "loss": 0.4638,
      "step": 13615
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5726244449615479,
      "learning_rate": 1.2776881895031681e-05,
      "loss": 0.4589,
      "step": 13616
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5717897415161133,
      "learning_rate": 1.2775903522952682e-05,
      "loss": 0.5091,
      "step": 13617
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4849807024002075,
      "learning_rate": 1.277492512208312e-05,
      "loss": 0.4295,
      "step": 13618
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.541182279586792,
      "learning_rate": 1.2773946692433152e-05,
      "loss": 0.5117,
      "step": 13619
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5247571468353271,
      "learning_rate": 1.2772968234012917e-05,
      "loss": 0.4475,
      "step": 13620
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.540117859840393,
      "learning_rate": 1.2771989746832569e-05,
      "loss": 0.4943,
      "step": 13621
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4898171424865723,
      "learning_rate": 1.2771011230902256e-05,
      "loss": 0.4135,
      "step": 13622
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5829603672027588,
      "learning_rate": 1.2770032686232123e-05,
      "loss": 0.4944,
      "step": 13623
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5266112089157104,
      "learning_rate": 1.2769054112832324e-05,
      "loss": 0.4599,
      "step": 13624
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.615538239479065,
      "learning_rate": 1.2768075510713005e-05,
      "loss": 0.5156,
      "step": 13625
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5440205335617065,
      "learning_rate": 1.2767096879884315e-05,
      "loss": 0.4679,
      "step": 13626
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.61681067943573,
      "learning_rate": 1.276611822035641e-05,
      "loss": 0.5165,
      "step": 13627
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5512323379516602,
      "learning_rate": 1.276513953213943e-05,
      "loss": 0.4156,
      "step": 13628
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5981543064117432,
      "learning_rate": 1.2764160815243536e-05,
      "loss": 0.44,
      "step": 13629
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5174890756607056,
      "learning_rate": 1.276318206967887e-05,
      "loss": 0.414,
      "step": 13630
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7168446779251099,
      "learning_rate": 1.2762203295455588e-05,
      "loss": 0.486,
      "step": 13631
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5454024076461792,
      "learning_rate": 1.2761224492583843e-05,
      "loss": 0.4507,
      "step": 13632
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.566328525543213,
      "learning_rate": 1.2760245661073784e-05,
      "loss": 0.4859,
      "step": 13633
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5947213172912598,
      "learning_rate": 1.2759266800935566e-05,
      "loss": 0.4434,
      "step": 13634
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5656952857971191,
      "learning_rate": 1.2758287912179336e-05,
      "loss": 0.4792,
      "step": 13635
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5438424348831177,
      "learning_rate": 1.2757308994815249e-05,
      "loss": 0.433,
      "step": 13636
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.498949646949768,
      "learning_rate": 1.2756330048853463e-05,
      "loss": 0.4678,
      "step": 13637
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.623868703842163,
      "learning_rate": 1.2755351074304121e-05,
      "loss": 0.4499,
      "step": 13638
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6114146709442139,
      "learning_rate": 1.2754372071177386e-05,
      "loss": 0.4866,
      "step": 13639
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5763697624206543,
      "learning_rate": 1.2753393039483406e-05,
      "loss": 0.4925,
      "step": 13640
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6189075708389282,
      "learning_rate": 1.275241397923234e-05,
      "loss": 0.4762,
      "step": 13641
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5872379541397095,
      "learning_rate": 1.2751434890434337e-05,
      "loss": 0.4319,
      "step": 13642
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.566235899925232,
      "learning_rate": 1.2750455773099557e-05,
      "loss": 0.4747,
      "step": 13643
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6093437671661377,
      "learning_rate": 1.2749476627238151e-05,
      "loss": 0.4197,
      "step": 13644
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5221821069717407,
      "learning_rate": 1.2748497452860275e-05,
      "loss": 0.4683,
      "step": 13645
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6198099851608276,
      "learning_rate": 1.2747518249976086e-05,
      "loss": 0.4325,
      "step": 13646
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6264384984970093,
      "learning_rate": 1.274653901859574e-05,
      "loss": 0.5015,
      "step": 13647
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5776755809783936,
      "learning_rate": 1.2745559758729391e-05,
      "loss": 0.4316,
      "step": 13648
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6342428922653198,
      "learning_rate": 1.2744580470387198e-05,
      "loss": 0.4923,
      "step": 13649
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.586553692817688,
      "learning_rate": 1.2743601153579317e-05,
      "loss": 0.4183,
      "step": 13650
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6029486656188965,
      "learning_rate": 1.2742621808315904e-05,
      "loss": 0.5263,
      "step": 13651
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5143851041793823,
      "learning_rate": 1.2741642434607118e-05,
      "loss": 0.4303,
      "step": 13652
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5991215705871582,
      "learning_rate": 1.2740663032463114e-05,
      "loss": 0.5205,
      "step": 13653
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5139057636260986,
      "learning_rate": 1.2739683601894053e-05,
      "loss": 0.4241,
      "step": 13654
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.626542329788208,
      "learning_rate": 1.2738704142910093e-05,
      "loss": 0.4922,
      "step": 13655
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4847604036331177,
      "learning_rate": 1.2737724655521389e-05,
      "loss": 0.4125,
      "step": 13656
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.581942081451416,
      "learning_rate": 1.2736745139738104e-05,
      "loss": 0.5293,
      "step": 13657
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5601097345352173,
      "learning_rate": 1.2735765595570397e-05,
      "loss": 0.4555,
      "step": 13658
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.503530740737915,
      "learning_rate": 1.2734786023028424e-05,
      "loss": 0.4554,
      "step": 13659
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5235732793807983,
      "learning_rate": 1.273380642212235e-05,
      "loss": 0.4185,
      "step": 13660
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6349081993103027,
      "learning_rate": 1.273282679286233e-05,
      "loss": 0.5194,
      "step": 13661
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5189054012298584,
      "learning_rate": 1.2731847135258523e-05,
      "loss": 0.4192,
      "step": 13662
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8983289003372192,
      "learning_rate": 1.2730867449321099e-05,
      "loss": 0.5356,
      "step": 13663
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4965277910232544,
      "learning_rate": 1.2729887735060205e-05,
      "loss": 0.3907,
      "step": 13664
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6848341226577759,
      "learning_rate": 1.2728907992486018e-05,
      "loss": 0.5064,
      "step": 13665
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5045524835586548,
      "learning_rate": 1.2727928221608685e-05,
      "loss": 0.4309,
      "step": 13666
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5347185134887695,
      "learning_rate": 1.2726948422438377e-05,
      "loss": 0.5365,
      "step": 13667
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4821491241455078,
      "learning_rate": 1.2725968594985252e-05,
      "loss": 0.4289,
      "step": 13668
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7437235116958618,
      "learning_rate": 1.2724988739259473e-05,
      "loss": 0.5215,
      "step": 13669
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5346038341522217,
      "learning_rate": 1.2724008855271206e-05,
      "loss": 0.4242,
      "step": 13670
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5172849893569946,
      "learning_rate": 1.2723028943030611e-05,
      "loss": 0.4662,
      "step": 13671
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.509555459022522,
      "learning_rate": 1.272204900254785e-05,
      "loss": 0.419,
      "step": 13672
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6235536336898804,
      "learning_rate": 1.2721069033833089e-05,
      "loss": 0.54,
      "step": 13673
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5433845520019531,
      "learning_rate": 1.272008903689649e-05,
      "loss": 0.4229,
      "step": 13674
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5685821771621704,
      "learning_rate": 1.2719109011748218e-05,
      "loss": 0.5147,
      "step": 13675
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5960602760314941,
      "learning_rate": 1.271812895839844e-05,
      "loss": 0.4796,
      "step": 13676
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.495924949645996,
      "learning_rate": 1.2717148876857315e-05,
      "loss": 0.4457,
      "step": 13677
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6087162494659424,
      "learning_rate": 1.2716168767135014e-05,
      "loss": 0.4891,
      "step": 13678
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5807462930679321,
      "learning_rate": 1.2715188629241695e-05,
      "loss": 0.4996,
      "step": 13679
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6140695810317993,
      "learning_rate": 1.2714208463187529e-05,
      "loss": 0.4353,
      "step": 13680
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5465304851531982,
      "learning_rate": 1.2713228268982687e-05,
      "loss": 0.5273,
      "step": 13681
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5508012771606445,
      "learning_rate": 1.2712248046637324e-05,
      "loss": 0.4235,
      "step": 13682
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.789409875869751,
      "learning_rate": 1.2711267796161613e-05,
      "loss": 0.5031,
      "step": 13683
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4484800100326538,
      "learning_rate": 1.2710287517565719e-05,
      "loss": 0.419,
      "step": 13684
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5701048374176025,
      "learning_rate": 1.2709307210859809e-05,
      "loss": 0.5255,
      "step": 13685
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6149203777313232,
      "learning_rate": 1.270832687605405e-05,
      "loss": 0.4465,
      "step": 13686
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5206716060638428,
      "learning_rate": 1.2707346513158614e-05,
      "loss": 0.4409,
      "step": 13687
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4962375164031982,
      "learning_rate": 1.2706366122183664e-05,
      "loss": 0.4322,
      "step": 13688
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.515352487564087,
      "learning_rate": 1.2705385703139367e-05,
      "loss": 0.4624,
      "step": 13689
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5887525081634521,
      "learning_rate": 1.2704405256035897e-05,
      "loss": 0.4662,
      "step": 13690
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.597694993019104,
      "learning_rate": 1.270342478088342e-05,
      "loss": 0.4959,
      "step": 13691
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5792229175567627,
      "learning_rate": 1.2702444277692102e-05,
      "loss": 0.4418,
      "step": 13692
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5443825721740723,
      "learning_rate": 1.2701463746472117e-05,
      "loss": 0.47,
      "step": 13693
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6690301895141602,
      "learning_rate": 1.2700483187233634e-05,
      "loss": 0.4871,
      "step": 13694
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5737439393997192,
      "learning_rate": 1.2699502599986823e-05,
      "loss": 0.4886,
      "step": 13695
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6096094846725464,
      "learning_rate": 1.2698521984741852e-05,
      "loss": 0.4765,
      "step": 13696
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5779494047164917,
      "learning_rate": 1.2697541341508894e-05,
      "loss": 0.4642,
      "step": 13697
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6385564804077148,
      "learning_rate": 1.2696560670298115e-05,
      "loss": 0.4597,
      "step": 13698
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9735814332962036,
      "learning_rate": 1.2695579971119693e-05,
      "loss": 0.4952,
      "step": 13699
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5817980766296387,
      "learning_rate": 1.2694599243983796e-05,
      "loss": 0.4514,
      "step": 13700
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6340980529785156,
      "learning_rate": 1.2693618488900596e-05,
      "loss": 0.5375,
      "step": 13701
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5548813343048096,
      "learning_rate": 1.2692637705880264e-05,
      "loss": 0.4222,
      "step": 13702
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.936113715171814,
      "learning_rate": 1.2691656894932973e-05,
      "loss": 1.0169,
      "step": 13703
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5110234022140503,
      "learning_rate": 1.2690676056068898e-05,
      "loss": 1.0254,
      "step": 13704
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6088645458221436,
      "learning_rate": 1.2689695189298207e-05,
      "loss": 0.4772,
      "step": 13705
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.674363374710083,
      "learning_rate": 1.2688714294631079e-05,
      "loss": 0.4326,
      "step": 13706
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5696747303009033,
      "learning_rate": 1.2687733372077681e-05,
      "loss": 0.4773,
      "step": 13707
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6670743227005005,
      "learning_rate": 1.2686752421648189e-05,
      "loss": 0.4401,
      "step": 13708
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7763912677764893,
      "learning_rate": 1.2685771443352786e-05,
      "loss": 0.5571,
      "step": 13709
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.496770977973938,
      "learning_rate": 1.2684790437201628e-05,
      "loss": 0.4218,
      "step": 13710
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6104105710983276,
      "learning_rate": 1.2683809403204908e-05,
      "loss": 0.5013,
      "step": 13711
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4984867572784424,
      "learning_rate": 1.2682828341372789e-05,
      "loss": 0.4341,
      "step": 13712
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.564233660697937,
      "learning_rate": 1.268184725171545e-05,
      "loss": 0.4602,
      "step": 13713
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.473136067390442,
      "learning_rate": 1.2680866134243068e-05,
      "loss": 0.4121,
      "step": 13714
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.734409213066101,
      "learning_rate": 1.2679884988965818e-05,
      "loss": 0.5157,
      "step": 13715
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6394423246383667,
      "learning_rate": 1.2678903815893871e-05,
      "loss": 0.4269,
      "step": 13716
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5304049253463745,
      "learning_rate": 1.2677922615037412e-05,
      "loss": 0.4831,
      "step": 13717
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4703404903411865,
      "learning_rate": 1.267694138640661e-05,
      "loss": 0.3989,
      "step": 13718
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7072880268096924,
      "learning_rate": 1.2675960130011649e-05,
      "loss": 0.5192,
      "step": 13719
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6124322414398193,
      "learning_rate": 1.2674978845862698e-05,
      "loss": 0.4551,
      "step": 13720
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7079734802246094,
      "learning_rate": 1.2673997533969942e-05,
      "loss": 0.4954,
      "step": 13721
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5140935182571411,
      "learning_rate": 1.2673016194343555e-05,
      "loss": 0.4492,
      "step": 13722
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.579454779624939,
      "learning_rate": 1.2672034826993716e-05,
      "loss": 0.4666,
      "step": 13723
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.658698558807373,
      "learning_rate": 1.2671053431930602e-05,
      "loss": 0.4379,
      "step": 13724
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6173819303512573,
      "learning_rate": 1.2670072009164394e-05,
      "loss": 0.4882,
      "step": 13725
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6962162256240845,
      "learning_rate": 1.2669090558705266e-05,
      "loss": 0.4039,
      "step": 13726
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5202562808990479,
      "learning_rate": 1.2668109080563407e-05,
      "loss": 0.4599,
      "step": 13727
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6275677680969238,
      "learning_rate": 1.2667127574748985e-05,
      "loss": 0.4664,
      "step": 13728
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.771654486656189,
      "learning_rate": 1.2666146041272187e-05,
      "loss": 0.4804,
      "step": 13729
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4687288999557495,
      "learning_rate": 1.2665164480143193e-05,
      "loss": 0.3912,
      "step": 13730
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6080520153045654,
      "learning_rate": 1.2664182891372177e-05,
      "loss": 0.5198,
      "step": 13731
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5763205289840698,
      "learning_rate": 1.266320127496933e-05,
      "loss": 0.4437,
      "step": 13732
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.536739706993103,
      "learning_rate": 1.2662219630944821e-05,
      "loss": 0.4869,
      "step": 13733
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5986850261688232,
      "learning_rate": 1.2661237959308841e-05,
      "loss": 0.4439,
      "step": 13734
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6609554290771484,
      "learning_rate": 1.2660256260071566e-05,
      "loss": 0.5172,
      "step": 13735
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.50784432888031,
      "learning_rate": 1.2659274533243181e-05,
      "loss": 0.4127,
      "step": 13736
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5998293161392212,
      "learning_rate": 1.2658292778833868e-05,
      "loss": 0.4992,
      "step": 13737
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5595157146453857,
      "learning_rate": 1.2657310996853806e-05,
      "loss": 0.4886,
      "step": 13738
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.592760443687439,
      "learning_rate": 1.2656329187313181e-05,
      "loss": 0.5578,
      "step": 13739
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.693363904953003,
      "learning_rate": 1.2655347350222173e-05,
      "loss": 0.4614,
      "step": 13740
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.622078537940979,
      "learning_rate": 1.2654365485590968e-05,
      "loss": 0.4755,
      "step": 13741
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.593578815460205,
      "learning_rate": 1.2653383593429748e-05,
      "loss": 0.4679,
      "step": 13742
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.51845383644104,
      "learning_rate": 1.2652401673748696e-05,
      "loss": 0.5168,
      "step": 13743
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6586755514144897,
      "learning_rate": 1.2651419726557999e-05,
      "loss": 0.4251,
      "step": 13744
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4606624841690063,
      "learning_rate": 1.265043775186784e-05,
      "loss": 0.4441,
      "step": 13745
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6161315441131592,
      "learning_rate": 1.2649455749688402e-05,
      "loss": 0.4505,
      "step": 13746
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5824933052062988,
      "learning_rate": 1.2648473720029873e-05,
      "loss": 0.4502,
      "step": 13747
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.569832682609558,
      "learning_rate": 1.2647491662902434e-05,
      "loss": 0.4328,
      "step": 13748
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5132344961166382,
      "learning_rate": 1.2646509578316271e-05,
      "loss": 0.4769,
      "step": 13749
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5106823444366455,
      "learning_rate": 1.2645527466281578e-05,
      "loss": 0.4254,
      "step": 13750
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.530625343322754,
      "learning_rate": 1.2644545326808529e-05,
      "loss": 0.4565,
      "step": 13751
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.681911826133728,
      "learning_rate": 1.2643563159907317e-05,
      "loss": 0.4449,
      "step": 13752
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5159322023391724,
      "learning_rate": 1.264258096558813e-05,
      "loss": 0.4671,
      "step": 13753
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5119235515594482,
      "learning_rate": 1.2641598743861155e-05,
      "loss": 0.445,
      "step": 13754
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6174087524414062,
      "learning_rate": 1.264061649473657e-05,
      "loss": 0.5053,
      "step": 13755
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5842937231063843,
      "learning_rate": 1.2639634218224574e-05,
      "loss": 0.4172,
      "step": 13756
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5326191186904907,
      "learning_rate": 1.2638651914335348e-05,
      "loss": 0.4557,
      "step": 13757
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5417810678482056,
      "learning_rate": 1.2637669583079086e-05,
      "loss": 0.4257,
      "step": 13758
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5820684432983398,
      "learning_rate": 1.2636687224465969e-05,
      "loss": 0.4842,
      "step": 13759
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5103223323822021,
      "learning_rate": 1.2635704838506193e-05,
      "loss": 0.438,
      "step": 13760
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5613818168640137,
      "learning_rate": 1.2634722425209938e-05,
      "loss": 0.5047,
      "step": 13761
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7053601741790771,
      "learning_rate": 1.26337399845874e-05,
      "loss": 0.4923,
      "step": 13762
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6213525533676147,
      "learning_rate": 1.2632757516648769e-05,
      "loss": 0.4925,
      "step": 13763
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5235143899917603,
      "learning_rate": 1.2631775021404231e-05,
      "loss": 0.3982,
      "step": 13764
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.547914743423462,
      "learning_rate": 1.2630792498863977e-05,
      "loss": 0.4978,
      "step": 13765
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4963077306747437,
      "learning_rate": 1.2629809949038201e-05,
      "loss": 0.4069,
      "step": 13766
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.847851037979126,
      "learning_rate": 1.2628827371937084e-05,
      "loss": 0.4761,
      "step": 13767
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6859526634216309,
      "learning_rate": 1.2627844767570832e-05,
      "loss": 0.4806,
      "step": 13768
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6928107738494873,
      "learning_rate": 1.262686213594962e-05,
      "loss": 0.5253,
      "step": 13769
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5840240716934204,
      "learning_rate": 1.2625879477083652e-05,
      "loss": 0.4008,
      "step": 13770
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8513368368148804,
      "learning_rate": 1.2624896790983113e-05,
      "loss": 0.9974,
      "step": 13771
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4929567575454712,
      "learning_rate": 1.2623914077658198e-05,
      "loss": 1.0334,
      "step": 13772
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6015733480453491,
      "learning_rate": 1.2622931337119095e-05,
      "loss": 0.4683,
      "step": 13773
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6147582530975342,
      "learning_rate": 1.2621948569376003e-05,
      "loss": 0.4361,
      "step": 13774
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5946934223175049,
      "learning_rate": 1.262096577443911e-05,
      "loss": 0.5203,
      "step": 13775
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.550524115562439,
      "learning_rate": 1.2619982952318613e-05,
      "loss": 0.4368,
      "step": 13776
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.619589924812317,
      "learning_rate": 1.26190001030247e-05,
      "loss": 0.5263,
      "step": 13777
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.516821265220642,
      "learning_rate": 1.2618017226567571e-05,
      "loss": 0.4168,
      "step": 13778
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7125240564346313,
      "learning_rate": 1.2617034322957413e-05,
      "loss": 0.5207,
      "step": 13779
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6154530048370361,
      "learning_rate": 1.2616051392204427e-05,
      "loss": 0.4397,
      "step": 13780
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5788830518722534,
      "learning_rate": 1.2615068434318805e-05,
      "loss": 0.503,
      "step": 13781
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.499907374382019,
      "learning_rate": 1.261408544931074e-05,
      "loss": 0.4238,
      "step": 13782
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5449291467666626,
      "learning_rate": 1.2613102437190431e-05,
      "loss": 0.5807,
      "step": 13783
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4818768501281738,
      "learning_rate": 1.261211939796807e-05,
      "loss": 0.4051,
      "step": 13784
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5694160461425781,
      "learning_rate": 1.2611136331653852e-05,
      "loss": 0.485,
      "step": 13785
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4670803546905518,
      "learning_rate": 1.261015323825798e-05,
      "loss": 0.4267,
      "step": 13786
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.702622890472412,
      "learning_rate": 1.2609170117790639e-05,
      "loss": 0.4816,
      "step": 13787
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6079238653182983,
      "learning_rate": 1.2608186970262035e-05,
      "loss": 0.4105,
      "step": 13788
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6227176189422607,
      "learning_rate": 1.2607203795682361e-05,
      "loss": 0.491,
      "step": 13789
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5975029468536377,
      "learning_rate": 1.2606220594061815e-05,
      "loss": 0.4398,
      "step": 13790
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7324587106704712,
      "learning_rate": 1.2605237365410595e-05,
      "loss": 0.5003,
      "step": 13791
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6392247676849365,
      "learning_rate": 1.2604254109738893e-05,
      "loss": 0.4336,
      "step": 13792
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6011461019515991,
      "learning_rate": 1.2603270827056913e-05,
      "loss": 0.4728,
      "step": 13793
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4991801977157593,
      "learning_rate": 1.2602287517374855e-05,
      "loss": 0.4271,
      "step": 13794
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5008292198181152,
      "learning_rate": 1.260130418070291e-05,
      "loss": 0.4636,
      "step": 13795
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5422999858856201,
      "learning_rate": 1.2600320817051286e-05,
      "loss": 0.4397,
      "step": 13796
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5893464088439941,
      "learning_rate": 1.2599337426430171e-05,
      "loss": 0.4957,
      "step": 13797
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5631909370422363,
      "learning_rate": 1.2598354008849775e-05,
      "loss": 0.4497,
      "step": 13798
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6377047300338745,
      "learning_rate": 1.259737056432029e-05,
      "loss": 0.506,
      "step": 13799
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6194738149642944,
      "learning_rate": 1.259638709285192e-05,
      "loss": 0.4856,
      "step": 13800
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6803234815597534,
      "learning_rate": 1.2595403594454868e-05,
      "loss": 0.5185,
      "step": 13801
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5778956413269043,
      "learning_rate": 1.2594420069139324e-05,
      "loss": 0.4139,
      "step": 13802
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.633966088294983,
      "learning_rate": 1.2593436516915496e-05,
      "loss": 0.4979,
      "step": 13803
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6878058910369873,
      "learning_rate": 1.2592452937793589e-05,
      "loss": 0.478,
      "step": 13804
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9351147413253784,
      "learning_rate": 1.2591469331783795e-05,
      "loss": 1.0133,
      "step": 13805
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5890929698944092,
      "learning_rate": 1.259048569889632e-05,
      "loss": 1.1031,
      "step": 13806
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6570653915405273,
      "learning_rate": 1.2589502039141368e-05,
      "loss": 0.5057,
      "step": 13807
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.592214584350586,
      "learning_rate": 1.258851835252914e-05,
      "loss": 0.4433,
      "step": 13808
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6695351600646973,
      "learning_rate": 1.2587534639069835e-05,
      "loss": 0.5484,
      "step": 13809
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5603463649749756,
      "learning_rate": 1.2586550898773659e-05,
      "loss": 0.4396,
      "step": 13810
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9074138402938843,
      "learning_rate": 1.2585567131650812e-05,
      "loss": 1.025,
      "step": 13811
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6610885858535767,
      "learning_rate": 1.2584583337711504e-05,
      "loss": 1.1796,
      "step": 13812
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5843050479888916,
      "learning_rate": 1.258359951696593e-05,
      "loss": 0.4945,
      "step": 13813
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5812699794769287,
      "learning_rate": 1.2582615669424299e-05,
      "loss": 0.422,
      "step": 13814
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6216496229171753,
      "learning_rate": 1.2581631795096814e-05,
      "loss": 0.5083,
      "step": 13815
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5395092964172363,
      "learning_rate": 1.2580647893993678e-05,
      "loss": 0.4332,
      "step": 13816
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5832605361938477,
      "learning_rate": 1.2579663966125098e-05,
      "loss": 0.4602,
      "step": 13817
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6117732524871826,
      "learning_rate": 1.2578680011501278e-05,
      "loss": 0.4439,
      "step": 13818
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7153109312057495,
      "learning_rate": 1.2577696030132422e-05,
      "loss": 0.476,
      "step": 13819
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5844595432281494,
      "learning_rate": 1.2576712022028734e-05,
      "loss": 0.4429,
      "step": 13820
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5155377388000488,
      "learning_rate": 1.2575727987200423e-05,
      "loss": 0.4767,
      "step": 13821
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5454635620117188,
      "learning_rate": 1.2574743925657697e-05,
      "loss": 0.4724,
      "step": 13822
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.526164174079895,
      "learning_rate": 1.2573759837410756e-05,
      "loss": 0.4592,
      "step": 13823
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6973848342895508,
      "learning_rate": 1.257277572246981e-05,
      "loss": 0.4421,
      "step": 13824
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5632511377334595,
      "learning_rate": 1.2571791580845067e-05,
      "loss": 0.4412,
      "step": 13825
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6183382272720337,
      "learning_rate": 1.2570807412546732e-05,
      "loss": 0.4426,
      "step": 13826
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6778510808944702,
      "learning_rate": 1.2569823217585016e-05,
      "loss": 0.5116,
      "step": 13827
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.519120693206787,
      "learning_rate": 1.2568838995970121e-05,
      "loss": 0.418,
      "step": 13828
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4967780113220215,
      "learning_rate": 1.2567854747712255e-05,
      "loss": 0.4771,
      "step": 13829
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.622245192527771,
      "learning_rate": 1.2566870472821636e-05,
      "loss": 0.436,
      "step": 13830
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6226496696472168,
      "learning_rate": 1.2565886171308459e-05,
      "loss": 0.4636,
      "step": 13831
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7001111507415771,
      "learning_rate": 1.2564901843182944e-05,
      "loss": 0.4641,
      "step": 13832
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6109564304351807,
      "learning_rate": 1.2563917488455293e-05,
      "loss": 0.5125,
      "step": 13833
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.602066159248352,
      "learning_rate": 1.2562933107135717e-05,
      "loss": 0.4462,
      "step": 13834
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6383302211761475,
      "learning_rate": 1.256194869923443e-05,
      "loss": 0.5014,
      "step": 13835
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.556018590927124,
      "learning_rate": 1.2560964264761636e-05,
      "loss": 0.4359,
      "step": 13836
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5762602090835571,
      "learning_rate": 1.2559979803727547e-05,
      "loss": 0.4957,
      "step": 13837
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.55526864528656,
      "learning_rate": 1.2558995316142375e-05,
      "loss": 0.4012,
      "step": 13838
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5606707334518433,
      "learning_rate": 1.2558010802016326e-05,
      "loss": 0.4758,
      "step": 13839
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4712132215499878,
      "learning_rate": 1.255702626135962e-05,
      "loss": 0.4384,
      "step": 13840
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6178396940231323,
      "learning_rate": 1.2556041694182457e-05,
      "loss": 0.5028,
      "step": 13841
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5769315958023071,
      "learning_rate": 1.2555057100495059e-05,
      "loss": 0.4607,
      "step": 13842
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.522218942642212,
      "learning_rate": 1.2554072480307631e-05,
      "loss": 0.5124,
      "step": 13843
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5712171792984009,
      "learning_rate": 1.2553087833630389e-05,
      "loss": 0.4723,
      "step": 13844
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6116023063659668,
      "learning_rate": 1.255210316047354e-05,
      "loss": 0.4733,
      "step": 13845
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6617079973220825,
      "learning_rate": 1.2551118460847303e-05,
      "loss": 0.4463,
      "step": 13846
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6764863729476929,
      "learning_rate": 1.2550133734761887e-05,
      "loss": 0.4738,
      "step": 13847
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5871108770370483,
      "learning_rate": 1.2549148982227508e-05,
      "loss": 0.4343,
      "step": 13848
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6924428939819336,
      "learning_rate": 1.2548164203254376e-05,
      "loss": 0.5052,
      "step": 13849
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5484477281570435,
      "learning_rate": 1.2547179397852707e-05,
      "loss": 0.4138,
      "step": 13850
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5726219415664673,
      "learning_rate": 1.2546194566032715e-05,
      "loss": 0.4363,
      "step": 13851
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.530392050743103,
      "learning_rate": 1.2545209707804613e-05,
      "loss": 0.4035,
      "step": 13852
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6749719381332397,
      "learning_rate": 1.2544224823178617e-05,
      "loss": 0.5508,
      "step": 13853
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5727643966674805,
      "learning_rate": 1.2543239912164941e-05,
      "loss": 0.4193,
      "step": 13854
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5375984907150269,
      "learning_rate": 1.2542254974773799e-05,
      "loss": 0.4725,
      "step": 13855
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6391127109527588,
      "learning_rate": 1.254127001101541e-05,
      "loss": 0.4828,
      "step": 13856
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5094982385635376,
      "learning_rate": 1.2540285020899986e-05,
      "loss": 0.4985,
      "step": 13857
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6360000371932983,
      "learning_rate": 1.2539300004437745e-05,
      "loss": 0.4566,
      "step": 13858
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.56049644947052,
      "learning_rate": 1.25383149616389e-05,
      "loss": 0.5158,
      "step": 13859
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5258010625839233,
      "learning_rate": 1.253732989251367e-05,
      "loss": 0.4499,
      "step": 13860
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7167924642562866,
      "learning_rate": 1.2536344797072275e-05,
      "loss": 0.5372,
      "step": 13861
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6092848777770996,
      "learning_rate": 1.2535359675324927e-05,
      "loss": 0.4289,
      "step": 13862
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7562094926834106,
      "learning_rate": 1.2534374527281845e-05,
      "loss": 1.0503,
      "step": 13863
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5298486948013306,
      "learning_rate": 1.2533389352953246e-05,
      "loss": 1.1196,
      "step": 13864
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6767687797546387,
      "learning_rate": 1.2532404152349346e-05,
      "loss": 0.5218,
      "step": 13865
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5919299125671387,
      "learning_rate": 1.2531418925480373e-05,
      "loss": 0.4329,
      "step": 13866
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6526533365249634,
      "learning_rate": 1.2530433672356529e-05,
      "loss": 0.5459,
      "step": 13867
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5712224245071411,
      "learning_rate": 1.2529448392988046e-05,
      "loss": 0.4262,
      "step": 13868
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5427229404449463,
      "learning_rate": 1.2528463087385136e-05,
      "loss": 0.4999,
      "step": 13869
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5600301027297974,
      "learning_rate": 1.2527477755558023e-05,
      "loss": 0.4029,
      "step": 13870
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6395974159240723,
      "learning_rate": 1.2526492397516924e-05,
      "loss": 0.4784,
      "step": 13871
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4888598918914795,
      "learning_rate": 1.2525507013272057e-05,
      "loss": 0.394,
      "step": 13872
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9167671203613281,
      "learning_rate": 1.2524521602833644e-05,
      "loss": 1.0391,
      "step": 13873
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.663430094718933,
      "learning_rate": 1.2523536166211908e-05,
      "loss": 1.0788,
      "step": 13874
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6658297777175903,
      "learning_rate": 1.2522550703417062e-05,
      "loss": 0.5589,
      "step": 13875
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5018874406814575,
      "learning_rate": 1.2521565214459335e-05,
      "loss": 0.4358,
      "step": 13876
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.685339331626892,
      "learning_rate": 1.2520579699348942e-05,
      "loss": 0.4761,
      "step": 13877
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6020803451538086,
      "learning_rate": 1.251959415809611e-05,
      "loss": 0.4185,
      "step": 13878
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5610030889511108,
      "learning_rate": 1.2518608590711054e-05,
      "loss": 0.442,
      "step": 13879
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5693222284317017,
      "learning_rate": 1.2517622997204002e-05,
      "loss": 0.4651,
      "step": 13880
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5709965229034424,
      "learning_rate": 1.2516637377585173e-05,
      "loss": 0.4547,
      "step": 13881
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.549026370048523,
      "learning_rate": 1.251565173186479e-05,
      "loss": 0.447,
      "step": 13882
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.570178508758545,
      "learning_rate": 1.2514666060053075e-05,
      "loss": 0.521,
      "step": 13883
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6293561458587646,
      "learning_rate": 1.2513680362160255e-05,
      "loss": 0.4648,
      "step": 13884
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5706162452697754,
      "learning_rate": 1.2512694638196545e-05,
      "loss": 0.451,
      "step": 13885
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.466283917427063,
      "learning_rate": 1.2511708888172178e-05,
      "loss": 0.4354,
      "step": 13886
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.618927001953125,
      "learning_rate": 1.2510723112097372e-05,
      "loss": 0.4796,
      "step": 13887
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4967329502105713,
      "learning_rate": 1.250973730998235e-05,
      "loss": 0.4322,
      "step": 13888
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5355781316757202,
      "learning_rate": 1.2508751481837342e-05,
      "loss": 0.4778,
      "step": 13889
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5319887399673462,
      "learning_rate": 1.250776562767257e-05,
      "loss": 0.4224,
      "step": 13890
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6276650428771973,
      "learning_rate": 1.2506779747498254e-05,
      "loss": 0.4789,
      "step": 13891
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5790213346481323,
      "learning_rate": 1.2505793841324629e-05,
      "loss": 0.4357,
      "step": 13892
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5529301166534424,
      "learning_rate": 1.250480790916191e-05,
      "loss": 0.5007,
      "step": 13893
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6450839042663574,
      "learning_rate": 1.2503821951020332e-05,
      "loss": 0.493,
      "step": 13894
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5482724905014038,
      "learning_rate": 1.2502835966910113e-05,
      "loss": 0.529,
      "step": 13895
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5192300081253052,
      "learning_rate": 1.2501849956841484e-05,
      "loss": 0.4187,
      "step": 13896
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5801483392715454,
      "learning_rate": 1.250086392082467e-05,
      "loss": 0.4852,
      "step": 13897
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.531345248222351,
      "learning_rate": 1.2499877858869898e-05,
      "loss": 0.4232,
      "step": 13898
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5451264381408691,
      "learning_rate": 1.2498891770987395e-05,
      "loss": 0.5026,
      "step": 13899
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5005122423171997,
      "learning_rate": 1.249790565718739e-05,
      "loss": 0.4331,
      "step": 13900
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6181726455688477,
      "learning_rate": 1.2496919517480106e-05,
      "loss": 0.5031,
      "step": 13901
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5106337070465088,
      "learning_rate": 1.2495933351875777e-05,
      "loss": 0.4037,
      "step": 13902
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4685118198394775,
      "learning_rate": 1.2494947160384625e-05,
      "loss": 0.4616,
      "step": 13903
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.574859857559204,
      "learning_rate": 1.2493960943016883e-05,
      "loss": 0.4464,
      "step": 13904
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5364750623703003,
      "learning_rate": 1.2492974699782779e-05,
      "loss": 0.4835,
      "step": 13905
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4479811191558838,
      "learning_rate": 1.249198843069254e-05,
      "loss": 0.4151,
      "step": 13906
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5669466257095337,
      "learning_rate": 1.2491002135756396e-05,
      "loss": 0.4973,
      "step": 13907
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5266234874725342,
      "learning_rate": 1.2490015814984578e-05,
      "loss": 0.4302,
      "step": 13908
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6476004123687744,
      "learning_rate": 1.248902946838731e-05,
      "loss": 0.5149,
      "step": 13909
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5935096740722656,
      "learning_rate": 1.248804309597483e-05,
      "loss": 0.4419,
      "step": 13910
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5474544763565063,
      "learning_rate": 1.2487056697757362e-05,
      "loss": 0.4625,
      "step": 13911
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5338644981384277,
      "learning_rate": 1.2486070273745143e-05,
      "loss": 0.4226,
      "step": 13912
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6551063060760498,
      "learning_rate": 1.2485083823948396e-05,
      "loss": 0.4875,
      "step": 13913
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.538346529006958,
      "learning_rate": 1.2484097348377358e-05,
      "loss": 0.4214,
      "step": 13914
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.491926670074463,
      "learning_rate": 1.2483110847042258e-05,
      "loss": 0.4623,
      "step": 13915
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6061538457870483,
      "learning_rate": 1.2482124319953326e-05,
      "loss": 0.464,
      "step": 13916
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.583544373512268,
      "learning_rate": 1.2481137767120797e-05,
      "loss": 0.5137,
      "step": 13917
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.574869155883789,
      "learning_rate": 1.2480151188554903e-05,
      "loss": 0.4239,
      "step": 13918
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5794320106506348,
      "learning_rate": 1.247916458426587e-05,
      "loss": 0.5178,
      "step": 13919
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4756618738174438,
      "learning_rate": 1.2478177954263944e-05,
      "loss": 0.425,
      "step": 13920
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5052030086517334,
      "learning_rate": 1.2477191298559344e-05,
      "loss": 0.4689,
      "step": 13921
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4284064769744873,
      "learning_rate": 1.247620461716231e-05,
      "loss": 0.4174,
      "step": 13922
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4676233530044556,
      "learning_rate": 1.2475217910083075e-05,
      "loss": 0.458,
      "step": 13923
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.563216209411621,
      "learning_rate": 1.2474231177331873e-05,
      "loss": 0.4533,
      "step": 13924
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9277560710906982,
      "learning_rate": 1.2473244418918938e-05,
      "loss": 0.5517,
      "step": 13925
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.438093662261963,
      "learning_rate": 1.24722576348545e-05,
      "loss": 0.3989,
      "step": 13926
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.220897674560547,
      "learning_rate": 1.2471270825148797e-05,
      "loss": 0.4884,
      "step": 13927
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6720094680786133,
      "learning_rate": 1.2470283989812068e-05,
      "loss": 0.4594,
      "step": 13928
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9623624086380005,
      "learning_rate": 1.2469297128854541e-05,
      "loss": 0.4669,
      "step": 13929
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5814710855484009,
      "learning_rate": 1.2468310242286456e-05,
      "loss": 0.4223,
      "step": 13930
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5585861206054688,
      "learning_rate": 1.2467323330118045e-05,
      "loss": 0.481,
      "step": 13931
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5319263935089111,
      "learning_rate": 1.2466336392359545e-05,
      "loss": 0.4534,
      "step": 13932
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5591953992843628,
      "learning_rate": 1.2465349429021197e-05,
      "loss": 0.4412,
      "step": 13933
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.614385962486267,
      "learning_rate": 1.2464362440113229e-05,
      "loss": 0.4454,
      "step": 13934
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5853569507598877,
      "learning_rate": 1.2463375425645882e-05,
      "loss": 0.4802,
      "step": 13935
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5451655387878418,
      "learning_rate": 1.2462388385629393e-05,
      "loss": 0.4899,
      "step": 13936
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.482163429260254,
      "learning_rate": 1.2461401320073998e-05,
      "loss": 0.4652,
      "step": 13937
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6099960803985596,
      "learning_rate": 1.2460414228989941e-05,
      "loss": 0.4332,
      "step": 13938
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.618947982788086,
      "learning_rate": 1.2459427112387447e-05,
      "loss": 0.4594,
      "step": 13939
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5239148139953613,
      "learning_rate": 1.2458439970276764e-05,
      "loss": 0.4208,
      "step": 13940
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5844093561172485,
      "learning_rate": 1.2457452802668125e-05,
      "loss": 0.512,
      "step": 13941
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4601014852523804,
      "learning_rate": 1.2456465609571775e-05,
      "loss": 0.4119,
      "step": 13942
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.507811427116394,
      "learning_rate": 1.2455478390997946e-05,
      "loss": 0.4659,
      "step": 13943
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5098971128463745,
      "learning_rate": 1.245449114695688e-05,
      "loss": 0.4391,
      "step": 13944
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.567125678062439,
      "learning_rate": 1.2453503877458815e-05,
      "loss": 0.487,
      "step": 13945
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.595647931098938,
      "learning_rate": 1.2452516582513994e-05,
      "loss": 0.4252,
      "step": 13946
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5837067365646362,
      "learning_rate": 1.2451529262132652e-05,
      "loss": 0.5376,
      "step": 13947
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6004983186721802,
      "learning_rate": 1.2450541916325029e-05,
      "loss": 0.4345,
      "step": 13948
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4780021905899048,
      "learning_rate": 1.2449554545101373e-05,
      "loss": 0.4581,
      "step": 13949
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4972128868103027,
      "learning_rate": 1.2448567148471917e-05,
      "loss": 0.4371,
      "step": 13950
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7167370319366455,
      "learning_rate": 1.2447579726446903e-05,
      "loss": 0.445,
      "step": 13951
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5887495279312134,
      "learning_rate": 1.2446592279036575e-05,
      "loss": 0.4553,
      "step": 13952
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5459954738616943,
      "learning_rate": 1.2445604806251173e-05,
      "loss": 0.5062,
      "step": 13953
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.498408555984497,
      "learning_rate": 1.2444617308100937e-05,
      "loss": 0.4252,
      "step": 13954
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6627495288848877,
      "learning_rate": 1.2443629784596109e-05,
      "loss": 0.4918,
      "step": 13955
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5048565864562988,
      "learning_rate": 1.2442642235746938e-05,
      "loss": 0.4293,
      "step": 13956
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6315512657165527,
      "learning_rate": 1.2441654661563656e-05,
      "loss": 0.4997,
      "step": 13957
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.565930962562561,
      "learning_rate": 1.2440667062056513e-05,
      "loss": 0.4127,
      "step": 13958
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6788041591644287,
      "learning_rate": 1.2439679437235749e-05,
      "loss": 0.4931,
      "step": 13959
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4770114421844482,
      "learning_rate": 1.2438691787111608e-05,
      "loss": 0.4331,
      "step": 13960
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5890023708343506,
      "learning_rate": 1.2437704111694334e-05,
      "loss": 0.5467,
      "step": 13961
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.542335033416748,
      "learning_rate": 1.243671641099417e-05,
      "loss": 0.4183,
      "step": 13962
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.974634885787964,
      "learning_rate": 1.243572868502136e-05,
      "loss": 0.4739,
      "step": 13963
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4522229433059692,
      "learning_rate": 1.2434740933786149e-05,
      "loss": 0.4196,
      "step": 13964
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.529963493347168,
      "learning_rate": 1.243375315729878e-05,
      "loss": 0.4892,
      "step": 13965
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5547428131103516,
      "learning_rate": 1.24327653555695e-05,
      "loss": 0.439,
      "step": 13966
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5839524269104004,
      "learning_rate": 1.2431777528608553e-05,
      "loss": 0.4854,
      "step": 13967
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4588301181793213,
      "learning_rate": 1.2430789676426185e-05,
      "loss": 0.4145,
      "step": 13968
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7677572965621948,
      "learning_rate": 1.242980179903264e-05,
      "loss": 0.4554,
      "step": 13969
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6028093099594116,
      "learning_rate": 1.2428813896438166e-05,
      "loss": 0.44,
      "step": 13970
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.630843162536621,
      "learning_rate": 1.2427825968653009e-05,
      "loss": 0.5025,
      "step": 13971
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.51360023021698,
      "learning_rate": 1.2426838015687411e-05,
      "loss": 0.4339,
      "step": 13972
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5942718982696533,
      "learning_rate": 1.2425850037551624e-05,
      "loss": 0.5212,
      "step": 13973
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5047955513000488,
      "learning_rate": 1.2424862034255894e-05,
      "loss": 0.4102,
      "step": 13974
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6457831859588623,
      "learning_rate": 1.2423874005810465e-05,
      "loss": 0.5137,
      "step": 13975
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5747404098510742,
      "learning_rate": 1.2422885952225588e-05,
      "loss": 0.4399,
      "step": 13976
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6325277090072632,
      "learning_rate": 1.2421897873511508e-05,
      "loss": 0.5327,
      "step": 13977
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4799970388412476,
      "learning_rate": 1.2420909769678473e-05,
      "loss": 0.4423,
      "step": 13978
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5840173959732056,
      "learning_rate": 1.2419921640736738e-05,
      "loss": 0.4708,
      "step": 13979
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5368932485580444,
      "learning_rate": 1.241893348669654e-05,
      "loss": 0.4332,
      "step": 13980
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5432862043380737,
      "learning_rate": 1.2417945307568137e-05,
      "loss": 0.4634,
      "step": 13981
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5841699838638306,
      "learning_rate": 1.2416957103361776e-05,
      "loss": 0.4184,
      "step": 13982
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5270063877105713,
      "learning_rate": 1.24159688740877e-05,
      "loss": 0.4665,
      "step": 13983
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4808058738708496,
      "learning_rate": 1.2414980619756166e-05,
      "loss": 0.4201,
      "step": 13984
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5143505334854126,
      "learning_rate": 1.2413992340377423e-05,
      "loss": 0.4601,
      "step": 13985
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4978537559509277,
      "learning_rate": 1.2413004035961716e-05,
      "loss": 0.3997,
      "step": 13986
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4854995012283325,
      "learning_rate": 1.2412015706519302e-05,
      "loss": 0.473,
      "step": 13987
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5863914489746094,
      "learning_rate": 1.2411027352060426e-05,
      "loss": 0.4253,
      "step": 13988
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6330220699310303,
      "learning_rate": 1.2410038972595345e-05,
      "loss": 0.5539,
      "step": 13989
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5628092288970947,
      "learning_rate": 1.2409050568134298e-05,
      "loss": 0.4264,
      "step": 13990
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6338520050048828,
      "learning_rate": 1.2408062138687549e-05,
      "loss": 0.4634,
      "step": 13991
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5577378273010254,
      "learning_rate": 1.2407073684265346e-05,
      "loss": 0.4217,
      "step": 13992
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5450894832611084,
      "learning_rate": 1.2406085204877938e-05,
      "loss": 0.4737,
      "step": 13993
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.518273949623108,
      "learning_rate": 1.240509670053558e-05,
      "loss": 0.4065,
      "step": 13994
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.631565809249878,
      "learning_rate": 1.2404108171248524e-05,
      "loss": 0.457,
      "step": 13995
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5903584957122803,
      "learning_rate": 1.2403119617027017e-05,
      "loss": 0.4655,
      "step": 13996
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5678235292434692,
      "learning_rate": 1.2402131037881322e-05,
      "loss": 0.4709,
      "step": 13997
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9996052980422974,
      "learning_rate": 1.2401142433821685e-05,
      "loss": 0.4338,
      "step": 13998
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6901435852050781,
      "learning_rate": 1.2400153804858362e-05,
      "loss": 0.5192,
      "step": 13999
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5030779838562012,
      "learning_rate": 1.2399165151001604e-05,
      "loss": 0.4305,
      "step": 14000
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6245505809783936,
      "learning_rate": 1.239817647226167e-05,
      "loss": 0.4972,
      "step": 14001
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.545807957649231,
      "learning_rate": 1.2397187768648808e-05,
      "loss": 0.4476,
      "step": 14002
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5454081296920776,
      "learning_rate": 1.2396199040173277e-05,
      "loss": 0.4967,
      "step": 14003
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5789666175842285,
      "learning_rate": 1.239521028684533e-05,
      "loss": 0.4198,
      "step": 14004
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6024113893508911,
      "learning_rate": 1.2394221508675225e-05,
      "loss": 0.5433,
      "step": 14005
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.551149606704712,
      "learning_rate": 1.239323270567321e-05,
      "loss": 0.4703,
      "step": 14006
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7057281732559204,
      "learning_rate": 1.2392243877849548e-05,
      "loss": 0.4671,
      "step": 14007
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5389407873153687,
      "learning_rate": 1.239125502521449e-05,
      "loss": 0.451,
      "step": 14008
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6151231527328491,
      "learning_rate": 1.2390266147778294e-05,
      "loss": 0.5629,
      "step": 14009
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4737659692764282,
      "learning_rate": 1.2389277245551216e-05,
      "loss": 0.4033,
      "step": 14010
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5175082683563232,
      "learning_rate": 1.2388288318543513e-05,
      "loss": 0.4928,
      "step": 14011
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5597282648086548,
      "learning_rate": 1.238729936676544e-05,
      "loss": 0.4252,
      "step": 14012
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6569833755493164,
      "learning_rate": 1.2386310390227255e-05,
      "loss": 0.5012,
      "step": 14013
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5372182130813599,
      "learning_rate": 1.2385321388939214e-05,
      "loss": 0.4643,
      "step": 14014
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7075895071029663,
      "learning_rate": 1.238433236291158e-05,
      "loss": 0.4879,
      "step": 14015
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6030759811401367,
      "learning_rate": 1.2383343312154604e-05,
      "loss": 0.4411,
      "step": 14016
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6496155261993408,
      "learning_rate": 1.2382354236678547e-05,
      "loss": 0.5647,
      "step": 14017
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5136841535568237,
      "learning_rate": 1.2381365136493666e-05,
      "loss": 0.4101,
      "step": 14018
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6055209636688232,
      "learning_rate": 1.2380376011610222e-05,
      "loss": 0.5183,
      "step": 14019
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4436627626419067,
      "learning_rate": 1.2379386862038472e-05,
      "loss": 0.471,
      "step": 14020
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7105169296264648,
      "learning_rate": 1.2378397687788675e-05,
      "loss": 0.5352,
      "step": 14021
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.497326374053955,
      "learning_rate": 1.2377408488871088e-05,
      "loss": 0.4372,
      "step": 14022
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6494545936584473,
      "learning_rate": 1.2376419265295978e-05,
      "loss": 0.4984,
      "step": 14023
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6154932975769043,
      "learning_rate": 1.2375430017073598e-05,
      "loss": 0.428,
      "step": 14024
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4659548997879028,
      "learning_rate": 1.237444074421421e-05,
      "loss": 0.4561,
      "step": 14025
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5096334218978882,
      "learning_rate": 1.2373451446728075e-05,
      "loss": 0.4232,
      "step": 14026
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5654727220535278,
      "learning_rate": 1.2372462124625453e-05,
      "loss": 0.4828,
      "step": 14027
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5944632291793823,
      "learning_rate": 1.2371472777916607e-05,
      "loss": 0.4707,
      "step": 14028
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6310116052627563,
      "learning_rate": 1.2370483406611795e-05,
      "loss": 0.4774,
      "step": 14029
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5086884498596191,
      "learning_rate": 1.2369494010721276e-05,
      "loss": 0.4263,
      "step": 14030
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5516120195388794,
      "learning_rate": 1.2368504590255319e-05,
      "loss": 0.4661,
      "step": 14031
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.495180606842041,
      "learning_rate": 1.236751514522418e-05,
      "loss": 0.4503,
      "step": 14032
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6208986043930054,
      "learning_rate": 1.2366525675638125e-05,
      "loss": 0.5257,
      "step": 14033
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6508177518844604,
      "learning_rate": 1.2365536181507415e-05,
      "loss": 0.4936,
      "step": 14034
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5949019193649292,
      "learning_rate": 1.236454666284231e-05,
      "loss": 0.5456,
      "step": 14035
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.458787202835083,
      "learning_rate": 1.2363557119653074e-05,
      "loss": 0.4007,
      "step": 14036
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5664374828338623,
      "learning_rate": 1.2362567551949973e-05,
      "loss": 0.4615,
      "step": 14037
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4685356616973877,
      "learning_rate": 1.236157795974327e-05,
      "loss": 0.4359,
      "step": 14038
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6044670343399048,
      "learning_rate": 1.2360588343043225e-05,
      "loss": 0.4968,
      "step": 14039
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6332019567489624,
      "learning_rate": 1.2359598701860103e-05,
      "loss": 0.4205,
      "step": 14040
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5671496391296387,
      "learning_rate": 1.2358609036204172e-05,
      "loss": 0.4613,
      "step": 14041
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5235403776168823,
      "learning_rate": 1.2357619346085692e-05,
      "loss": 0.4035,
      "step": 14042
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.608874797821045,
      "learning_rate": 1.235662963151493e-05,
      "loss": 0.5051,
      "step": 14043
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.477988600730896,
      "learning_rate": 1.235563989250215e-05,
      "loss": 0.4167,
      "step": 14044
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5042279958724976,
      "learning_rate": 1.235465012905762e-05,
      "loss": 0.4914,
      "step": 14045
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.56562077999115,
      "learning_rate": 1.23536603411916e-05,
      "loss": 0.4279,
      "step": 14046
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.542520523071289,
      "learning_rate": 1.235267052891436e-05,
      "loss": 0.4876,
      "step": 14047
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.711096167564392,
      "learning_rate": 1.2351680692236161e-05,
      "loss": 0.45,
      "step": 14048
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7866452932357788,
      "learning_rate": 1.2350690831167276e-05,
      "loss": 0.5232,
      "step": 14049
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6957286596298218,
      "learning_rate": 1.2349700945717965e-05,
      "loss": 0.4325,
      "step": 14050
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7110761404037476,
      "learning_rate": 1.2348711035898504e-05,
      "loss": 0.4622,
      "step": 14051
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5071771144866943,
      "learning_rate": 1.2347721101719148e-05,
      "loss": 0.4335,
      "step": 14052
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5950435400009155,
      "learning_rate": 1.2346731143190172e-05,
      "loss": 0.4853,
      "step": 14053
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5695005655288696,
      "learning_rate": 1.2345741160321842e-05,
      "loss": 0.4274,
      "step": 14054
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5815564393997192,
      "learning_rate": 1.2344751153124426e-05,
      "loss": 0.4783,
      "step": 14055
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5046677589416504,
      "learning_rate": 1.2343761121608188e-05,
      "loss": 0.4136,
      "step": 14056
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8345112800598145,
      "learning_rate": 1.2342771065783401e-05,
      "loss": 0.5211,
      "step": 14057
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.51547110080719,
      "learning_rate": 1.234178098566033e-05,
      "loss": 0.4437,
      "step": 14058
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5805085897445679,
      "learning_rate": 1.2340790881249251e-05,
      "loss": 0.4989,
      "step": 14059
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.583685040473938,
      "learning_rate": 1.233980075256042e-05,
      "loss": 0.4128,
      "step": 14060
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8088587522506714,
      "learning_rate": 1.2338810599604119e-05,
      "loss": 0.9429,
      "step": 14061
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4753637313842773,
      "learning_rate": 1.233782042239061e-05,
      "loss": 1.037,
      "step": 14062
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.625318169593811,
      "learning_rate": 1.2336830220930164e-05,
      "loss": 0.4782,
      "step": 14063
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5963200330734253,
      "learning_rate": 1.2335839995233054e-05,
      "loss": 0.4647,
      "step": 14064
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6398203372955322,
      "learning_rate": 1.233484974530955e-05,
      "loss": 0.5037,
      "step": 14065
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.573996901512146,
      "learning_rate": 1.2333859471169915e-05,
      "loss": 0.4442,
      "step": 14066
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5810998678207397,
      "learning_rate": 1.233286917282443e-05,
      "loss": 0.4876,
      "step": 14067
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5124342441558838,
      "learning_rate": 1.2331878850283356e-05,
      "loss": 0.4212,
      "step": 14068
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6063878536224365,
      "learning_rate": 1.2330888503556977e-05,
      "loss": 0.5215,
      "step": 14069
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6082098484039307,
      "learning_rate": 1.2329898132655551e-05,
      "loss": 0.4152,
      "step": 14070
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6800446510314941,
      "learning_rate": 1.2328907737589357e-05,
      "loss": 0.4878,
      "step": 14071
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5516600608825684,
      "learning_rate": 1.2327917318368666e-05,
      "loss": 0.4107,
      "step": 14072
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.599318504333496,
      "learning_rate": 1.232692687500375e-05,
      "loss": 0.5039,
      "step": 14073
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5263042449951172,
      "learning_rate": 1.2325936407504881e-05,
      "loss": 0.4301,
      "step": 14074
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5156525373458862,
      "learning_rate": 1.2324945915882334e-05,
      "loss": 0.4568,
      "step": 14075
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5785622596740723,
      "learning_rate": 1.2323955400146375e-05,
      "loss": 0.4522,
      "step": 14076
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6325730085372925,
      "learning_rate": 1.2322964860307289e-05,
      "loss": 0.5597,
      "step": 14077
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5519317388534546,
      "learning_rate": 1.2321974296375336e-05,
      "loss": 0.4582,
      "step": 14078
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.568352222442627,
      "learning_rate": 1.23209837083608e-05,
      "loss": 0.4941,
      "step": 14079
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6211258172988892,
      "learning_rate": 1.2319993096273953e-05,
      "loss": 0.4916,
      "step": 14080
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6292765140533447,
      "learning_rate": 1.2319002460125067e-05,
      "loss": 0.4998,
      "step": 14081
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5760499238967896,
      "learning_rate": 1.2318011799924417e-05,
      "loss": 0.4247,
      "step": 14082
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5935981273651123,
      "learning_rate": 1.2317021115682278e-05,
      "loss": 0.4687,
      "step": 14083
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6143831014633179,
      "learning_rate": 1.2316030407408921e-05,
      "loss": 0.4538,
      "step": 14084
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8421450853347778,
      "learning_rate": 1.231503967511463e-05,
      "loss": 1.0038,
      "step": 14085
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5891107320785522,
      "learning_rate": 1.2314048918809673e-05,
      "loss": 1.0953,
      "step": 14086
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6823056936264038,
      "learning_rate": 1.2313058138504333e-05,
      "loss": 0.4967,
      "step": 14087
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5892902612686157,
      "learning_rate": 1.2312067334208875e-05,
      "loss": 0.4323,
      "step": 14088
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6765456199645996,
      "learning_rate": 1.2311076505933587e-05,
      "loss": 0.4637,
      "step": 14089
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5808924436569214,
      "learning_rate": 1.2310085653688734e-05,
      "loss": 0.42,
      "step": 14090
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5900919437408447,
      "learning_rate": 1.2309094777484604e-05,
      "loss": 0.4906,
      "step": 14091
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5470554828643799,
      "learning_rate": 1.2308103877331468e-05,
      "loss": 0.4274,
      "step": 14092
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.623856782913208,
      "learning_rate": 1.2307112953239602e-05,
      "loss": 0.4558,
      "step": 14093
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6410915851593018,
      "learning_rate": 1.2306122005219282e-05,
      "loss": 0.4643,
      "step": 14094
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6634407043457031,
      "learning_rate": 1.2305131033280797e-05,
      "loss": 0.4573,
      "step": 14095
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7091342210769653,
      "learning_rate": 1.2304140037434413e-05,
      "loss": 0.4616,
      "step": 14096
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.520294189453125,
      "learning_rate": 1.2303149017690412e-05,
      "loss": 0.4788,
      "step": 14097
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.569922685623169,
      "learning_rate": 1.2302157974059075e-05,
      "loss": 0.4419,
      "step": 14098
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.590296983718872,
      "learning_rate": 1.2301166906550677e-05,
      "loss": 0.5041,
      "step": 14099
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5824295282363892,
      "learning_rate": 1.2300175815175499e-05,
      "loss": 0.437,
      "step": 14100
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5774664878845215,
      "learning_rate": 1.2299184699943818e-05,
      "loss": 0.4673,
      "step": 14101
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8810107707977295,
      "learning_rate": 1.2298193560865915e-05,
      "loss": 0.4373,
      "step": 14102
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5540151596069336,
      "learning_rate": 1.2297202397952073e-05,
      "loss": 0.468,
      "step": 14103
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6802144050598145,
      "learning_rate": 1.2296211211212566e-05,
      "loss": 0.4581,
      "step": 14104
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.529316782951355,
      "learning_rate": 1.2295220000657679e-05,
      "loss": 0.443,
      "step": 14105
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6443524360656738,
      "learning_rate": 1.2294228766297687e-05,
      "loss": 0.4729,
      "step": 14106
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5293304920196533,
      "learning_rate": 1.2293237508142877e-05,
      "loss": 0.4342,
      "step": 14107
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.571416974067688,
      "learning_rate": 1.2292246226203526e-05,
      "loss": 0.419,
      "step": 14108
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7242131233215332,
      "learning_rate": 1.2291254920489914e-05,
      "loss": 0.4986,
      "step": 14109
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.1466569900512695,
      "learning_rate": 1.2290263591012328e-05,
      "loss": 0.4453,
      "step": 14110
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.633086085319519,
      "learning_rate": 1.2289272237781045e-05,
      "loss": 0.4947,
      "step": 14111
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5532454252243042,
      "learning_rate": 1.2288280860806348e-05,
      "loss": 0.4222,
      "step": 14112
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5786665678024292,
      "learning_rate": 1.2287289460098519e-05,
      "loss": 0.4943,
      "step": 14113
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7691888809204102,
      "learning_rate": 1.228629803566784e-05,
      "loss": 0.4315,
      "step": 14114
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5892212390899658,
      "learning_rate": 1.2285306587524597e-05,
      "loss": 0.5158,
      "step": 14115
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4590568542480469,
      "learning_rate": 1.2284315115679069e-05,
      "loss": 0.4117,
      "step": 14116
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6028869152069092,
      "learning_rate": 1.2283323620141538e-05,
      "loss": 0.481,
      "step": 14117
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.338541030883789,
      "learning_rate": 1.2282332100922294e-05,
      "loss": 0.4365,
      "step": 14118
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.584194540977478,
      "learning_rate": 1.2281340558031614e-05,
      "loss": 0.5161,
      "step": 14119
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5594559907913208,
      "learning_rate": 1.2280348991479784e-05,
      "loss": 0.418,
      "step": 14120
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5329151153564453,
      "learning_rate": 1.2279357401277092e-05,
      "loss": 0.4689,
      "step": 14121
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5793232917785645,
      "learning_rate": 1.2278365787433812e-05,
      "loss": 0.4303,
      "step": 14122
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4926344156265259,
      "learning_rate": 1.2277374149960244e-05,
      "loss": 0.4283,
      "step": 14123
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5375617742538452,
      "learning_rate": 1.2276382488866659e-05,
      "loss": 0.4333,
      "step": 14124
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5064923763275146,
      "learning_rate": 1.2275390804163348e-05,
      "loss": 0.4995,
      "step": 14125
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.581276297569275,
      "learning_rate": 1.2274399095860597e-05,
      "loss": 0.441,
      "step": 14126
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6425660848617554,
      "learning_rate": 1.227340736396869e-05,
      "loss": 0.4926,
      "step": 14127
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.495300054550171,
      "learning_rate": 1.2272415608497914e-05,
      "loss": 0.4475,
      "step": 14128
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6255303621292114,
      "learning_rate": 1.2271423829458554e-05,
      "loss": 0.5783,
      "step": 14129
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.52699875831604,
      "learning_rate": 1.2270432026860891e-05,
      "loss": 0.4307,
      "step": 14130
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5770022869110107,
      "learning_rate": 1.2269440200715229e-05,
      "loss": 0.51,
      "step": 14131
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5135877132415771,
      "learning_rate": 1.2268448351031831e-05,
      "loss": 0.4266,
      "step": 14132
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.639483094215393,
      "learning_rate": 1.2267456477821003e-05,
      "loss": 0.49,
      "step": 14133
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.607995629310608,
      "learning_rate": 1.2266464581093025e-05,
      "loss": 0.4267,
      "step": 14134
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6205625534057617,
      "learning_rate": 1.2265472660858185e-05,
      "loss": 0.4677,
      "step": 14135
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5184683799743652,
      "learning_rate": 1.226448071712677e-05,
      "loss": 0.4282,
      "step": 14136
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5752005577087402,
      "learning_rate": 1.2263488749909069e-05,
      "loss": 0.4634,
      "step": 14137
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4576072692871094,
      "learning_rate": 1.2262496759215367e-05,
      "loss": 0.4062,
      "step": 14138
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5908513069152832,
      "learning_rate": 1.2261504745055963e-05,
      "loss": 0.4652,
      "step": 14139
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5377956628799438,
      "learning_rate": 1.2260512707441132e-05,
      "loss": 0.4186,
      "step": 14140
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.942636489868164,
      "learning_rate": 1.225952064638117e-05,
      "loss": 1.0285,
      "step": 14141
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6199032068252563,
      "learning_rate": 1.225852856188637e-05,
      "loss": 1.1112,
      "step": 14142
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6613624095916748,
      "learning_rate": 1.2257536453967014e-05,
      "loss": 0.486,
      "step": 14143
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.562725305557251,
      "learning_rate": 1.2256544322633397e-05,
      "loss": 0.4395,
      "step": 14144
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6024712324142456,
      "learning_rate": 1.2255552167895805e-05,
      "loss": 0.4724,
      "step": 14145
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.573686957359314,
      "learning_rate": 1.2254559989764532e-05,
      "loss": 0.4266,
      "step": 14146
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4889472723007202,
      "learning_rate": 1.2253567788249866e-05,
      "loss": 0.4434,
      "step": 14147
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4783707857131958,
      "learning_rate": 1.2252575563362095e-05,
      "loss": 0.4618,
      "step": 14148
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.663993239402771,
      "learning_rate": 1.225158331511152e-05,
      "loss": 0.4999,
      "step": 14149
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.544213056564331,
      "learning_rate": 1.225059104350842e-05,
      "loss": 0.4396,
      "step": 14150
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5271695852279663,
      "learning_rate": 1.2249598748563094e-05,
      "loss": 0.444,
      "step": 14151
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6114624738693237,
      "learning_rate": 1.224860643028583e-05,
      "loss": 0.4698,
      "step": 14152
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5335237979888916,
      "learning_rate": 1.2247614088686924e-05,
      "loss": 0.4611,
      "step": 14153
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5564405918121338,
      "learning_rate": 1.2246621723776666e-05,
      "loss": 0.4205,
      "step": 14154
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.669191598892212,
      "learning_rate": 1.2245629335565347e-05,
      "loss": 0.4929,
      "step": 14155
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5477828979492188,
      "learning_rate": 1.224463692406326e-05,
      "loss": 0.419,
      "step": 14156
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5423353910446167,
      "learning_rate": 1.2243644489280702e-05,
      "loss": 0.4495,
      "step": 14157
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5081663131713867,
      "learning_rate": 1.2242652031227958e-05,
      "loss": 0.4297,
      "step": 14158
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.616474986076355,
      "learning_rate": 1.224165954991533e-05,
      "loss": 0.4596,
      "step": 14159
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.508080005645752,
      "learning_rate": 1.2240667045353108e-05,
      "loss": 0.424,
      "step": 14160
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6812714338302612,
      "learning_rate": 1.2239674517551582e-05,
      "loss": 0.481,
      "step": 14161
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7077863216400146,
      "learning_rate": 1.2238681966521053e-05,
      "loss": 0.4428,
      "step": 14162
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6300808191299438,
      "learning_rate": 1.2237689392271811e-05,
      "loss": 0.5006,
      "step": 14163
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5202622413635254,
      "learning_rate": 1.2236696794814154e-05,
      "loss": 0.4169,
      "step": 14164
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6430772542953491,
      "learning_rate": 1.2235704174158375e-05,
      "loss": 0.5265,
      "step": 14165
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.617904782295227,
      "learning_rate": 1.2234711530314762e-05,
      "loss": 0.4845,
      "step": 14166
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5055574178695679,
      "learning_rate": 1.2233718863293624e-05,
      "loss": 0.4973,
      "step": 14167
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.495041847229004,
      "learning_rate": 1.2232726173105245e-05,
      "loss": 0.4253,
      "step": 14168
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5874799489974976,
      "learning_rate": 1.2231733459759926e-05,
      "loss": 0.476,
      "step": 14169
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5547547340393066,
      "learning_rate": 1.2230740723267967e-05,
      "loss": 0.45,
      "step": 14170
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.583789348602295,
      "learning_rate": 1.2229747963639656e-05,
      "loss": 0.4486,
      "step": 14171
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5033291578292847,
      "learning_rate": 1.2228755180885293e-05,
      "loss": 0.4115,
      "step": 14172
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5793592929840088,
      "learning_rate": 1.2227762375015175e-05,
      "loss": 0.5094,
      "step": 14173
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5325419902801514,
      "learning_rate": 1.2226769546039596e-05,
      "loss": 0.4647,
      "step": 14174
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6027840375900269,
      "learning_rate": 1.222577669396886e-05,
      "loss": 0.5135,
      "step": 14175
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6020053625106812,
      "learning_rate": 1.2224783818813259e-05,
      "loss": 0.432,
      "step": 14176
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6024219989776611,
      "learning_rate": 1.2223790920583092e-05,
      "loss": 0.487,
      "step": 14177
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5878193378448486,
      "learning_rate": 1.2222797999288659e-05,
      "loss": 0.4551,
      "step": 14178
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7955721616744995,
      "learning_rate": 1.2221805054940255e-05,
      "loss": 1.0358,
      "step": 14179
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6310240030288696,
      "learning_rate": 1.222081208754818e-05,
      "loss": 0.9684,
      "step": 14180
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6613985300064087,
      "learning_rate": 1.2219819097122732e-05,
      "loss": 0.4843,
      "step": 14181
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.553000569343567,
      "learning_rate": 1.2218826083674209e-05,
      "loss": 0.4452,
      "step": 14182
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5411760807037354,
      "learning_rate": 1.2217833047212916e-05,
      "loss": 0.4787,
      "step": 14183
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5697969198226929,
      "learning_rate": 1.2216839987749141e-05,
      "loss": 0.4453,
      "step": 14184
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.698582410812378,
      "learning_rate": 1.2215846905293196e-05,
      "loss": 0.4798,
      "step": 14185
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.486343502998352,
      "learning_rate": 1.221485379985537e-05,
      "loss": 0.4001,
      "step": 14186
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5258198976516724,
      "learning_rate": 1.2213860671445974e-05,
      "loss": 0.483,
      "step": 14187
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5548415184020996,
      "learning_rate": 1.22128675200753e-05,
      "loss": 0.4304,
      "step": 14188
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6016795635223389,
      "learning_rate": 1.2211874345753651e-05,
      "loss": 0.5146,
      "step": 14189
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5360685586929321,
      "learning_rate": 1.221088114849133e-05,
      "loss": 0.4237,
      "step": 14190
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.579851508140564,
      "learning_rate": 1.2209887928298634e-05,
      "loss": 0.476,
      "step": 14191
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5880494117736816,
      "learning_rate": 1.2208894685185865e-05,
      "loss": 0.4162,
      "step": 14192
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6175998449325562,
      "learning_rate": 1.220790141916333e-05,
      "loss": 0.4512,
      "step": 14193
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6527091264724731,
      "learning_rate": 1.2206908130241324e-05,
      "loss": 0.4247,
      "step": 14194
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5536061525344849,
      "learning_rate": 1.2205914818430152e-05,
      "loss": 0.4648,
      "step": 14195
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.617282748222351,
      "learning_rate": 1.2204921483740118e-05,
      "loss": 0.4494,
      "step": 14196
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.539086103439331,
      "learning_rate": 1.2203928126181519e-05,
      "loss": 0.4533,
      "step": 14197
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6264913082122803,
      "learning_rate": 1.2202934745764662e-05,
      "loss": 0.4285,
      "step": 14198
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6230273246765137,
      "learning_rate": 1.2201941342499852e-05,
      "loss": 0.4607,
      "step": 14199
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6233566999435425,
      "learning_rate": 1.2200947916397386e-05,
      "loss": 0.4671,
      "step": 14200
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5761823654174805,
      "learning_rate": 1.2199954467467573e-05,
      "loss": 0.4642,
      "step": 14201
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4810134172439575,
      "learning_rate": 1.219896099572071e-05,
      "loss": 0.4348,
      "step": 14202
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5122137069702148,
      "learning_rate": 1.2197967501167112e-05,
      "loss": 0.4718,
      "step": 14203
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5215470790863037,
      "learning_rate": 1.2196973983817069e-05,
      "loss": 0.4624,
      "step": 14204
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.591099739074707,
      "learning_rate": 1.2195980443680898e-05,
      "loss": 0.4482,
      "step": 14205
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5476391315460205,
      "learning_rate": 1.2194986880768895e-05,
      "loss": 0.4795,
      "step": 14206
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5361772775650024,
      "learning_rate": 1.2193993295091369e-05,
      "loss": 0.5149,
      "step": 14207
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.513521432876587,
      "learning_rate": 1.2192999686658625e-05,
      "loss": 0.4224,
      "step": 14208
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5692338943481445,
      "learning_rate": 1.2192006055480967e-05,
      "loss": 0.4777,
      "step": 14209
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4550436735153198,
      "learning_rate": 1.2191012401568698e-05,
      "loss": 0.4396,
      "step": 14210
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7511215209960938,
      "learning_rate": 1.2190018724932131e-05,
      "loss": 0.5307,
      "step": 14211
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5543761253356934,
      "learning_rate": 1.2189025025581567e-05,
      "loss": 0.4212,
      "step": 14212
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6481258869171143,
      "learning_rate": 1.2188031303527312e-05,
      "loss": 0.5015,
      "step": 14213
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5019100904464722,
      "learning_rate": 1.2187037558779675e-05,
      "loss": 0.4181,
      "step": 14214
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6728287935256958,
      "learning_rate": 1.2186043791348958e-05,
      "loss": 0.5309,
      "step": 14215
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5866882801055908,
      "learning_rate": 1.2185050001245475e-05,
      "loss": 0.4449,
      "step": 14216
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5441226959228516,
      "learning_rate": 1.2184056188479527e-05,
      "loss": 0.5332,
      "step": 14217
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.496904730796814,
      "learning_rate": 1.2183062353061426e-05,
      "loss": 0.421,
      "step": 14218
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7481316328048706,
      "learning_rate": 1.2182068495001476e-05,
      "loss": 0.5229,
      "step": 14219
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.505543828010559,
      "learning_rate": 1.2181074614309988e-05,
      "loss": 0.4195,
      "step": 14220
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6260915994644165,
      "learning_rate": 1.2180080710997266e-05,
      "loss": 0.4702,
      "step": 14221
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5677909851074219,
      "learning_rate": 1.2179086785073622e-05,
      "loss": 0.4539,
      "step": 14222
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.562374234199524,
      "learning_rate": 1.2178092836549364e-05,
      "loss": 0.5113,
      "step": 14223
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5782731771469116,
      "learning_rate": 1.21770988654348e-05,
      "loss": 0.4237,
      "step": 14224
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5802284479141235,
      "learning_rate": 1.2176104871740237e-05,
      "loss": 0.4718,
      "step": 14225
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5827192068099976,
      "learning_rate": 1.217511085547599e-05,
      "loss": 0.4281,
      "step": 14226
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.619228482246399,
      "learning_rate": 1.2174116816652364e-05,
      "loss": 0.5136,
      "step": 14227
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5008336305618286,
      "learning_rate": 1.217312275527967e-05,
      "loss": 0.4092,
      "step": 14228
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.704419732093811,
      "learning_rate": 1.2172128671368219e-05,
      "loss": 0.4702,
      "step": 14229
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5120455026626587,
      "learning_rate": 1.217113456492832e-05,
      "loss": 0.3997,
      "step": 14230
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.614970088005066,
      "learning_rate": 1.2170140435970284e-05,
      "loss": 0.4772,
      "step": 14231
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5930688381195068,
      "learning_rate": 1.216914628450442e-05,
      "loss": 0.4568,
      "step": 14232
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5699044466018677,
      "learning_rate": 1.216815211054104e-05,
      "loss": 0.5487,
      "step": 14233
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5860838890075684,
      "learning_rate": 1.2167157914090459e-05,
      "loss": 0.467,
      "step": 14234
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.517996072769165,
      "learning_rate": 1.2166163695162982e-05,
      "loss": 0.4693,
      "step": 14235
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5908347368240356,
      "learning_rate": 1.2165169453768923e-05,
      "loss": 0.4795,
      "step": 14236
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.538132905960083,
      "learning_rate": 1.2164175189918596e-05,
      "loss": 0.4611,
      "step": 14237
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7917951345443726,
      "learning_rate": 1.2163180903622313e-05,
      "loss": 0.4644,
      "step": 14238
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4719878435134888,
      "learning_rate": 1.2162186594890384e-05,
      "loss": 0.4942,
      "step": 14239
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5253283977508545,
      "learning_rate": 1.2161192263733122e-05,
      "loss": 0.4224,
      "step": 14240
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.818124532699585,
      "learning_rate": 1.216019791016084e-05,
      "loss": 0.979,
      "step": 14241
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6573759317398071,
      "learning_rate": 1.2159203534183853e-05,
      "loss": 1.0595,
      "step": 14242
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5887922048568726,
      "learning_rate": 1.215820913581247e-05,
      "loss": 0.5338,
      "step": 14243
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.515329360961914,
      "learning_rate": 1.2157214715057009e-05,
      "loss": 0.4067,
      "step": 14244
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6337844133377075,
      "learning_rate": 1.2156220271927778e-05,
      "loss": 0.5431,
      "step": 14245
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6103707551956177,
      "learning_rate": 1.21552258064351e-05,
      "loss": 0.4947,
      "step": 14246
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6734737157821655,
      "learning_rate": 1.2154231318589282e-05,
      "loss": 0.5023,
      "step": 14247
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5082210302352905,
      "learning_rate": 1.2153236808400641e-05,
      "loss": 0.4242,
      "step": 14248
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5419299602508545,
      "learning_rate": 1.2152242275879488e-05,
      "loss": 0.4222,
      "step": 14249
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4742727279663086,
      "learning_rate": 1.2151247721036146e-05,
      "loss": 0.3944,
      "step": 14250
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6866050958633423,
      "learning_rate": 1.2150253143880919e-05,
      "loss": 0.4649,
      "step": 14251
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5122939348220825,
      "learning_rate": 1.2149258544424136e-05,
      "loss": 0.4352,
      "step": 14252
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5455378293991089,
      "learning_rate": 1.2148263922676097e-05,
      "loss": 0.4641,
      "step": 14253
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5464940071105957,
      "learning_rate": 1.2147269278647131e-05,
      "loss": 0.4346,
      "step": 14254
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.558779001235962,
      "learning_rate": 1.2146274612347548e-05,
      "loss": 0.4501,
      "step": 14255
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6609302759170532,
      "learning_rate": 1.2145279923787667e-05,
      "loss": 0.4508,
      "step": 14256
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5935877561569214,
      "learning_rate": 1.21442852129778e-05,
      "loss": 0.4722,
      "step": 14257
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6570327281951904,
      "learning_rate": 1.2143290479928267e-05,
      "loss": 0.4448,
      "step": 14258
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6744508743286133,
      "learning_rate": 1.2142295724649384e-05,
      "loss": 0.4762,
      "step": 14259
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5215952396392822,
      "learning_rate": 1.2141300947151471e-05,
      "loss": 0.4066,
      "step": 14260
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5804494619369507,
      "learning_rate": 1.214030614744484e-05,
      "loss": 0.5089,
      "step": 14261
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.552212119102478,
      "learning_rate": 1.2139311325539816e-05,
      "loss": 0.4084,
      "step": 14262
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5006616115570068,
      "learning_rate": 1.2138316481446708e-05,
      "loss": 0.4538,
      "step": 14263
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.515621542930603,
      "learning_rate": 1.2137321615175839e-05,
      "loss": 0.424,
      "step": 14264
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6440094709396362,
      "learning_rate": 1.2136326726737528e-05,
      "loss": 0.5198,
      "step": 14265
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.515618920326233,
      "learning_rate": 1.2135331816142095e-05,
      "loss": 0.4186,
      "step": 14266
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6038286685943604,
      "learning_rate": 1.2134336883399855e-05,
      "loss": 0.5003,
      "step": 14267
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5458985567092896,
      "learning_rate": 1.2133341928521127e-05,
      "loss": 0.4188,
      "step": 14268
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5341745615005493,
      "learning_rate": 1.2132346951516231e-05,
      "loss": 0.4659,
      "step": 14269
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.6345949172973633,
      "learning_rate": 1.2131351952395494e-05,
      "loss": 0.4527,
      "step": 14270
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5682636499404907,
      "learning_rate": 1.2130356931169222e-05,
      "loss": 0.4502,
      "step": 14271
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6094377040863037,
      "learning_rate": 1.2129361887847746e-05,
      "loss": 0.442,
      "step": 14272
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6148691177368164,
      "learning_rate": 1.212836682244138e-05,
      "loss": 0.482,
      "step": 14273
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5266557931900024,
      "learning_rate": 1.2127371734960447e-05,
      "loss": 0.4362,
      "step": 14274
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5892680883407593,
      "learning_rate": 1.2126376625415268e-05,
      "loss": 0.4757,
      "step": 14275
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.595651388168335,
      "learning_rate": 1.2125381493816163e-05,
      "loss": 0.4394,
      "step": 14276
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5676621198654175,
      "learning_rate": 1.2124386340173452e-05,
      "loss": 0.4806,
      "step": 14277
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.566899061203003,
      "learning_rate": 1.212339116449746e-05,
      "loss": 0.436,
      "step": 14278
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5699310302734375,
      "learning_rate": 1.2122395966798504e-05,
      "loss": 0.5016,
      "step": 14279
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6523650884628296,
      "learning_rate": 1.2121400747086909e-05,
      "loss": 0.437,
      "step": 14280
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.644482970237732,
      "learning_rate": 1.2120405505372994e-05,
      "loss": 0.4651,
      "step": 14281
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5497158765792847,
      "learning_rate": 1.2119410241667087e-05,
      "loss": 0.4165,
      "step": 14282
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.694693684577942,
      "learning_rate": 1.2118414955979503e-05,
      "loss": 0.5195,
      "step": 14283
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.505888819694519,
      "learning_rate": 1.211741964832057e-05,
      "loss": 0.423,
      "step": 14284
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5167770385742188,
      "learning_rate": 1.211642431870061e-05,
      "loss": 0.4596,
      "step": 14285
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5954099893569946,
      "learning_rate": 1.2115428967129945e-05,
      "loss": 0.434,
      "step": 14286
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7459830045700073,
      "learning_rate": 1.2114433593618898e-05,
      "loss": 0.4946,
      "step": 14287
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6489161252975464,
      "learning_rate": 1.2113438198177794e-05,
      "loss": 0.4551,
      "step": 14288
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.623543381690979,
      "learning_rate": 1.2112442780816956e-05,
      "loss": 0.494,
      "step": 14289
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4912736415863037,
      "learning_rate": 1.211144734154671e-05,
      "loss": 0.425,
      "step": 14290
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5134477615356445,
      "learning_rate": 1.2110451880377376e-05,
      "loss": 0.4523,
      "step": 14291
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5208760499954224,
      "learning_rate": 1.2109456397319284e-05,
      "loss": 0.427,
      "step": 14292
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.562125563621521,
      "learning_rate": 1.2108460892382754e-05,
      "loss": 0.455,
      "step": 14293
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.559670329093933,
      "learning_rate": 1.2107465365578112e-05,
      "loss": 0.4428,
      "step": 14294
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4961451292037964,
      "learning_rate": 1.2106469816915685e-05,
      "loss": 0.4569,
      "step": 14295
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5402631759643555,
      "learning_rate": 1.2105474246405801e-05,
      "loss": 0.414,
      "step": 14296
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5474117994308472,
      "learning_rate": 1.2104478654058778e-05,
      "loss": 0.4799,
      "step": 14297
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.520592212677002,
      "learning_rate": 1.2103483039884949e-05,
      "loss": 0.4107,
      "step": 14298
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5752458572387695,
      "learning_rate": 1.2102487403894634e-05,
      "loss": 0.5025,
      "step": 14299
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.621809959411621,
      "learning_rate": 1.2101491746098166e-05,
      "loss": 0.4122,
      "step": 14300
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5165280103683472,
      "learning_rate": 1.2100496066505867e-05,
      "loss": 0.4682,
      "step": 14301
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6230173110961914,
      "learning_rate": 1.2099500365128065e-05,
      "loss": 0.4131,
      "step": 14302
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6305118799209595,
      "learning_rate": 1.2098504641975087e-05,
      "loss": 0.5204,
      "step": 14303
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5559282302856445,
      "learning_rate": 1.2097508897057259e-05,
      "loss": 0.4326,
      "step": 14304
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.646468997001648,
      "learning_rate": 1.2096513130384909e-05,
      "loss": 0.4881,
      "step": 14305
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4736697673797607,
      "learning_rate": 1.2095517341968368e-05,
      "loss": 0.414,
      "step": 14306
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6748485565185547,
      "learning_rate": 1.2094521531817957e-05,
      "loss": 0.537,
      "step": 14307
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5876314640045166,
      "learning_rate": 1.2093525699944013e-05,
      "loss": 0.424,
      "step": 14308
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5400052070617676,
      "learning_rate": 1.2092529846356858e-05,
      "loss": 0.4813,
      "step": 14309
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5858054161071777,
      "learning_rate": 1.2091533971066822e-05,
      "loss": 0.4778,
      "step": 14310
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5669623613357544,
      "learning_rate": 1.2090538074084237e-05,
      "loss": 0.4525,
      "step": 14311
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.494663119316101,
      "learning_rate": 1.2089542155419428e-05,
      "loss": 0.4174,
      "step": 14312
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5211173295974731,
      "learning_rate": 1.2088546215082721e-05,
      "loss": 0.5302,
      "step": 14313
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.564260482788086,
      "learning_rate": 1.2087550253084455e-05,
      "loss": 0.4116,
      "step": 14314
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7420753240585327,
      "learning_rate": 1.2086554269434951e-05,
      "loss": 0.5153,
      "step": 14315
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5257089138031006,
      "learning_rate": 1.2085558264144548e-05,
      "loss": 0.4349,
      "step": 14316
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5149632692337036,
      "learning_rate": 1.2084562237223565e-05,
      "loss": 0.5145,
      "step": 14317
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5814377069473267,
      "learning_rate": 1.2083566188682341e-05,
      "loss": 0.463,
      "step": 14318
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5950692892074585,
      "learning_rate": 1.2082570118531205e-05,
      "loss": 0.5137,
      "step": 14319
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5440173149108887,
      "learning_rate": 1.2081574026780485e-05,
      "loss": 0.4201,
      "step": 14320
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5540885925292969,
      "learning_rate": 1.2080577913440515e-05,
      "loss": 0.4497,
      "step": 14321
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6057157516479492,
      "learning_rate": 1.2079581778521622e-05,
      "loss": 0.405,
      "step": 14322
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.607479453086853,
      "learning_rate": 1.2078585622034143e-05,
      "loss": 0.4959,
      "step": 14323
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5638349056243896,
      "learning_rate": 1.2077589443988409e-05,
      "loss": 0.4134,
      "step": 14324
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7137271165847778,
      "learning_rate": 1.2076593244394745e-05,
      "loss": 0.4616,
      "step": 14325
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4961326122283936,
      "learning_rate": 1.2075597023263492e-05,
      "loss": 0.4198,
      "step": 14326
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5651792287826538,
      "learning_rate": 1.2074600780604979e-05,
      "loss": 0.4506,
      "step": 14327
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.519135594367981,
      "learning_rate": 1.2073604516429539e-05,
      "loss": 0.4333,
      "step": 14328
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5528382062911987,
      "learning_rate": 1.2072608230747503e-05,
      "loss": 0.4774,
      "step": 14329
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4864635467529297,
      "learning_rate": 1.2071611923569204e-05,
      "loss": 0.414,
      "step": 14330
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5770549774169922,
      "learning_rate": 1.2070615594904978e-05,
      "loss": 0.4838,
      "step": 14331
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.489181637763977,
      "learning_rate": 1.2069619244765158e-05,
      "loss": 0.4125,
      "step": 14332
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.585843563079834,
      "learning_rate": 1.2068622873160077e-05,
      "loss": 0.5394,
      "step": 14333
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6035503149032593,
      "learning_rate": 1.2067626480100067e-05,
      "loss": 0.4353,
      "step": 14334
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5287120342254639,
      "learning_rate": 1.2066630065595466e-05,
      "loss": 0.4965,
      "step": 14335
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.469120979309082,
      "learning_rate": 1.2065633629656606e-05,
      "loss": 0.4356,
      "step": 14336
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5825051069259644,
      "learning_rate": 1.2064637172293822e-05,
      "loss": 0.4501,
      "step": 14337
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5150476694107056,
      "learning_rate": 1.2063640693517448e-05,
      "loss": 0.4083,
      "step": 14338
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5867087841033936,
      "learning_rate": 1.2062644193337822e-05,
      "loss": 0.4735,
      "step": 14339
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5195316076278687,
      "learning_rate": 1.2061647671765278e-05,
      "loss": 0.4188,
      "step": 14340
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5221108198165894,
      "learning_rate": 1.2060651128810146e-05,
      "loss": 0.4863,
      "step": 14341
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5142472982406616,
      "learning_rate": 1.2059654564482773e-05,
      "loss": 0.4157,
      "step": 14342
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5165938138961792,
      "learning_rate": 1.2058657978793484e-05,
      "loss": 0.5065,
      "step": 14343
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6224662065505981,
      "learning_rate": 1.205766137175262e-05,
      "loss": 0.4665,
      "step": 14344
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5365031957626343,
      "learning_rate": 1.2056664743370522e-05,
      "loss": 0.4866,
      "step": 14345
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4837960004806519,
      "learning_rate": 1.2055668093657518e-05,
      "loss": 0.4378,
      "step": 14346
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.520166039466858,
      "learning_rate": 1.2054671422623949e-05,
      "loss": 0.519,
      "step": 14347
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4879813194274902,
      "learning_rate": 1.2053674730280152e-05,
      "loss": 0.4342,
      "step": 14348
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.60150146484375,
      "learning_rate": 1.2052678016636461e-05,
      "loss": 0.4631,
      "step": 14349
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5838745832443237,
      "learning_rate": 1.2051681281703222e-05,
      "loss": 0.4899,
      "step": 14350
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5371525287628174,
      "learning_rate": 1.2050684525490763e-05,
      "loss": 0.4445,
      "step": 14351
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5387828350067139,
      "learning_rate": 1.204968774800943e-05,
      "loss": 0.4378,
      "step": 14352
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6238020658493042,
      "learning_rate": 1.2048690949269554e-05,
      "loss": 0.5029,
      "step": 14353
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5547634363174438,
      "learning_rate": 1.204769412928148e-05,
      "loss": 0.4219,
      "step": 14354
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5272445678710938,
      "learning_rate": 1.204669728805554e-05,
      "loss": 0.4728,
      "step": 14355
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4349822998046875,
      "learning_rate": 1.2045700425602078e-05,
      "loss": 0.4057,
      "step": 14356
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6477090120315552,
      "learning_rate": 1.2044703541931432e-05,
      "loss": 0.4896,
      "step": 14357
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7012163400650024,
      "learning_rate": 1.204370663705394e-05,
      "loss": 0.4324,
      "step": 14358
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6010608673095703,
      "learning_rate": 1.204270971097994e-05,
      "loss": 0.4967,
      "step": 14359
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5159417390823364,
      "learning_rate": 1.2041712763719779e-05,
      "loss": 0.4243,
      "step": 14360
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6227704286575317,
      "learning_rate": 1.2040715795283789e-05,
      "loss": 0.4552,
      "step": 14361
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5807636976242065,
      "learning_rate": 1.2039718805682312e-05,
      "loss": 0.445,
      "step": 14362
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5848793983459473,
      "learning_rate": 1.2038721794925689e-05,
      "loss": 0.4868,
      "step": 14363
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.69203519821167,
      "learning_rate": 1.2037724763024263e-05,
      "loss": 0.4401,
      "step": 14364
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5882500410079956,
      "learning_rate": 1.2036727709988371e-05,
      "loss": 0.4808,
      "step": 14365
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5079078674316406,
      "learning_rate": 1.2035730635828358e-05,
      "loss": 0.4288,
      "step": 14366
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.51542067527771,
      "learning_rate": 1.203473354055456e-05,
      "loss": 0.4994,
      "step": 14367
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5162614583969116,
      "learning_rate": 1.2033736424177325e-05,
      "loss": 0.4388,
      "step": 14368
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.587725281715393,
      "learning_rate": 1.2032739286706987e-05,
      "loss": 0.5107,
      "step": 14369
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5520838499069214,
      "learning_rate": 1.2031742128153893e-05,
      "loss": 0.4125,
      "step": 14370
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.706386923789978,
      "learning_rate": 1.2030744948528387e-05,
      "loss": 0.5216,
      "step": 14371
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5854319334030151,
      "learning_rate": 1.2029747747840808e-05,
      "loss": 0.427,
      "step": 14372
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6584856510162354,
      "learning_rate": 1.2028750526101499e-05,
      "loss": 0.4866,
      "step": 14373
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.501886010169983,
      "learning_rate": 1.20277532833208e-05,
      "loss": 0.4057,
      "step": 14374
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6768007278442383,
      "learning_rate": 1.2026756019509059e-05,
      "loss": 0.4657,
      "step": 14375
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5459489822387695,
      "learning_rate": 1.2025758734676618e-05,
      "loss": 0.4578,
      "step": 14376
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.125023365020752,
      "learning_rate": 1.2024761428833815e-05,
      "loss": 0.5072,
      "step": 14377
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6470561027526855,
      "learning_rate": 1.2023764101991005e-05,
      "loss": 0.4318,
      "step": 14378
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5075238943099976,
      "learning_rate": 1.2022766754158522e-05,
      "loss": 0.4264,
      "step": 14379
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.470657467842102,
      "learning_rate": 1.2021769385346712e-05,
      "loss": 0.4393,
      "step": 14380
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5197376012802124,
      "learning_rate": 1.202077199556592e-05,
      "loss": 0.4646,
      "step": 14381
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.558563470840454,
      "learning_rate": 1.2019774584826494e-05,
      "loss": 0.4543,
      "step": 14382
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5448867082595825,
      "learning_rate": 1.2018777153138774e-05,
      "loss": 0.5502,
      "step": 14383
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5375539064407349,
      "learning_rate": 1.2017779700513108e-05,
      "loss": 0.4495,
      "step": 14384
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5958914756774902,
      "learning_rate": 1.2016782226959838e-05,
      "loss": 0.52,
      "step": 14385
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5839298963546753,
      "learning_rate": 1.2015784732489314e-05,
      "loss": 0.4509,
      "step": 14386
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4611029624938965,
      "learning_rate": 1.2014787217111876e-05,
      "loss": 0.4547,
      "step": 14387
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.560172438621521,
      "learning_rate": 1.2013789680837873e-05,
      "loss": 0.4464,
      "step": 14388
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5211116075515747,
      "learning_rate": 1.2012792123677652e-05,
      "loss": 0.4704,
      "step": 14389
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5286145210266113,
      "learning_rate": 1.201179454564156e-05,
      "loss": 0.4277,
      "step": 14390
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4888882637023926,
      "learning_rate": 1.2010796946739937e-05,
      "loss": 0.4516,
      "step": 14391
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5125758647918701,
      "learning_rate": 1.2009799326983138e-05,
      "loss": 0.4205,
      "step": 14392
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4919742345809937,
      "learning_rate": 1.2008801686381504e-05,
      "loss": 0.4556,
      "step": 14393
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4635611772537231,
      "learning_rate": 1.2007804024945386e-05,
      "loss": 0.4424,
      "step": 14394
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5633347034454346,
      "learning_rate": 1.2006806342685127e-05,
      "loss": 0.4527,
      "step": 14395
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.761504054069519,
      "learning_rate": 1.200580863961108e-05,
      "loss": 0.5065,
      "step": 14396
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5621649026870728,
      "learning_rate": 1.2004810915733589e-05,
      "loss": 0.4695,
      "step": 14397
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.572002649307251,
      "learning_rate": 1.2003813171063004e-05,
      "loss": 0.4448,
      "step": 14398
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5872617959976196,
      "learning_rate": 1.200281540560967e-05,
      "loss": 0.5103,
      "step": 14399
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0042824745178223,
      "learning_rate": 1.200181761938394e-05,
      "loss": 0.4375,
      "step": 14400
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.52580726146698,
      "learning_rate": 1.2000819812396158e-05,
      "loss": 0.4524,
      "step": 14401
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6032954454421997,
      "learning_rate": 1.1999821984656676e-05,
      "loss": 0.436,
      "step": 14402
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5749651193618774,
      "learning_rate": 1.199882413617584e-05,
      "loss": 0.4677,
      "step": 14403
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5010162591934204,
      "learning_rate": 1.1997826266964006e-05,
      "loss": 0.4259,
      "step": 14404
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5046453475952148,
      "learning_rate": 1.1996828377031514e-05,
      "loss": 0.4713,
      "step": 14405
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5418559312820435,
      "learning_rate": 1.1995830466388721e-05,
      "loss": 0.4383,
      "step": 14406
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.613053321838379,
      "learning_rate": 1.1994832535045974e-05,
      "loss": 0.5046,
      "step": 14407
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6330795288085938,
      "learning_rate": 1.1993834583013626e-05,
      "loss": 0.4656,
      "step": 14408
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5908180475234985,
      "learning_rate": 1.1992836610302023e-05,
      "loss": 0.5255,
      "step": 14409
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4785232543945312,
      "learning_rate": 1.1991838616921519e-05,
      "loss": 0.4256,
      "step": 14410
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.587762713432312,
      "learning_rate": 1.199084060288246e-05,
      "loss": 0.4635,
      "step": 14411
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6341447830200195,
      "learning_rate": 1.1989842568195205e-05,
      "loss": 0.4611,
      "step": 14412
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6159287691116333,
      "learning_rate": 1.1988844512870098e-05,
      "loss": 0.4651,
      "step": 14413
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5027867555618286,
      "learning_rate": 1.1987846436917494e-05,
      "loss": 0.4127,
      "step": 14414
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5774667263031006,
      "learning_rate": 1.1986848340347742e-05,
      "loss": 0.4476,
      "step": 14415
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5353784561157227,
      "learning_rate": 1.1985850223171196e-05,
      "loss": 0.4302,
      "step": 14416
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5325130224227905,
      "learning_rate": 1.1984852085398209e-05,
      "loss": 0.4833,
      "step": 14417
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.51686692237854,
      "learning_rate": 1.1983853927039131e-05,
      "loss": 0.407,
      "step": 14418
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7471319437026978,
      "learning_rate": 1.1982855748104316e-05,
      "loss": 0.545,
      "step": 14419
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4904004335403442,
      "learning_rate": 1.1981857548604116e-05,
      "loss": 0.4247,
      "step": 14420
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5905941724777222,
      "learning_rate": 1.1980859328548883e-05,
      "loss": 0.4913,
      "step": 14421
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.541147232055664,
      "learning_rate": 1.1979861087948974e-05,
      "loss": 0.4101,
      "step": 14422
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.526959776878357,
      "learning_rate": 1.1978862826814736e-05,
      "loss": 0.4689,
      "step": 14423
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4938019514083862,
      "learning_rate": 1.1977864545156528e-05,
      "loss": 0.4418,
      "step": 14424
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.593640685081482,
      "learning_rate": 1.1976866242984704e-05,
      "loss": 0.512,
      "step": 14425
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5067236423492432,
      "learning_rate": 1.1975867920309611e-05,
      "loss": 0.4481,
      "step": 14426
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5714142322540283,
      "learning_rate": 1.1974869577141612e-05,
      "loss": 0.4837,
      "step": 14427
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5893824100494385,
      "learning_rate": 1.1973871213491058e-05,
      "loss": 0.4553,
      "step": 14428
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6260541677474976,
      "learning_rate": 1.1972872829368299e-05,
      "loss": 0.5366,
      "step": 14429
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.505661129951477,
      "learning_rate": 1.1971874424783698e-05,
      "loss": 0.4213,
      "step": 14430
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6603800058364868,
      "learning_rate": 1.1970875999747605e-05,
      "loss": 0.5525,
      "step": 14431
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6357793807983398,
      "learning_rate": 1.1969877554270377e-05,
      "loss": 0.4287,
      "step": 14432
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5689899921417236,
      "learning_rate": 1.1968879088362364e-05,
      "loss": 0.5125,
      "step": 14433
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.495030403137207,
      "learning_rate": 1.1967880602033932e-05,
      "loss": 0.4202,
      "step": 14434
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9237803220748901,
      "learning_rate": 1.196688209529543e-05,
      "loss": 0.4746,
      "step": 14435
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5082026720046997,
      "learning_rate": 1.1965883568157214e-05,
      "loss": 0.4164,
      "step": 14436
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5803217887878418,
      "learning_rate": 1.1964885020629642e-05,
      "loss": 0.4694,
      "step": 14437
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.582617998123169,
      "learning_rate": 1.1963886452723072e-05,
      "loss": 0.4535,
      "step": 14438
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5194581747055054,
      "learning_rate": 1.1962887864447856e-05,
      "loss": 0.4949,
      "step": 14439
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6641367673873901,
      "learning_rate": 1.1961889255814358e-05,
      "loss": 0.441,
      "step": 14440
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.573187232017517,
      "learning_rate": 1.1960890626832928e-05,
      "loss": 0.4387,
      "step": 14441
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6025497913360596,
      "learning_rate": 1.1959891977513928e-05,
      "loss": 0.4437,
      "step": 14442
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.564945936203003,
      "learning_rate": 1.1958893307867714e-05,
      "loss": 0.4638,
      "step": 14443
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4780330657958984,
      "learning_rate": 1.1957894617904644e-05,
      "loss": 0.4044,
      "step": 14444
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.675304889678955,
      "learning_rate": 1.1956895907635079e-05,
      "loss": 0.5085,
      "step": 14445
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5021485090255737,
      "learning_rate": 1.1955897177069368e-05,
      "loss": 0.4335,
      "step": 14446
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6027004718780518,
      "learning_rate": 1.1954898426217878e-05,
      "loss": 0.5012,
      "step": 14447
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6457397937774658,
      "learning_rate": 1.1953899655090968e-05,
      "loss": 0.4163,
      "step": 14448
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6802419424057007,
      "learning_rate": 1.195290086369899e-05,
      "loss": 0.5225,
      "step": 14449
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5470446348190308,
      "learning_rate": 1.1951902052052313e-05,
      "loss": 0.4365,
      "step": 14450
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5202165842056274,
      "learning_rate": 1.1950903220161286e-05,
      "loss": 0.4588,
      "step": 14451
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4666712284088135,
      "learning_rate": 1.194990436803627e-05,
      "loss": 0.4014,
      "step": 14452
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5963466167449951,
      "learning_rate": 1.1948905495687635e-05,
      "loss": 0.5203,
      "step": 14453
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.516978144645691,
      "learning_rate": 1.1947906603125727e-05,
      "loss": 0.4351,
      "step": 14454
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7483131885528564,
      "learning_rate": 1.1946907690360919e-05,
      "loss": 0.5748,
      "step": 14455
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5099873542785645,
      "learning_rate": 1.1945908757403558e-05,
      "loss": 0.4184,
      "step": 14456
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.537808895111084,
      "learning_rate": 1.1944909804264015e-05,
      "loss": 0.4657,
      "step": 14457
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5438766479492188,
      "learning_rate": 1.194391083095265e-05,
      "loss": 0.4368,
      "step": 14458
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5829676389694214,
      "learning_rate": 1.1942911837479819e-05,
      "loss": 0.4502,
      "step": 14459
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4668728113174438,
      "learning_rate": 1.1941912823855884e-05,
      "loss": 0.4156,
      "step": 14460
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5901063680648804,
      "learning_rate": 1.1940913790091208e-05,
      "loss": 0.4713,
      "step": 14461
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.523322343826294,
      "learning_rate": 1.1939914736196152e-05,
      "loss": 0.4129,
      "step": 14462
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6938834190368652,
      "learning_rate": 1.193891566218108e-05,
      "loss": 0.4777,
      "step": 14463
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5075862407684326,
      "learning_rate": 1.1937916568056351e-05,
      "loss": 0.4307,
      "step": 14464
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6633590459823608,
      "learning_rate": 1.1936917453832329e-05,
      "loss": 0.4714,
      "step": 14465
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4857732057571411,
      "learning_rate": 1.1935918319519376e-05,
      "loss": 0.4096,
      "step": 14466
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.559062123298645,
      "learning_rate": 1.1934919165127854e-05,
      "loss": 0.5156,
      "step": 14467
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6539064645767212,
      "learning_rate": 1.1933919990668125e-05,
      "loss": 0.4326,
      "step": 14468
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6134389638900757,
      "learning_rate": 1.1932920796150555e-05,
      "loss": 0.5103,
      "step": 14469
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5672919750213623,
      "learning_rate": 1.19319215815855e-05,
      "loss": 0.4544,
      "step": 14470
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6221964359283447,
      "learning_rate": 1.1930922346983338e-05,
      "loss": 0.4914,
      "step": 14471
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6022385358810425,
      "learning_rate": 1.1929923092354417e-05,
      "loss": 0.4565,
      "step": 14472
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5304052829742432,
      "learning_rate": 1.192892381770911e-05,
      "loss": 0.4638,
      "step": 14473
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.51836097240448,
      "learning_rate": 1.1927924523057777e-05,
      "loss": 0.4245,
      "step": 14474
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5159552097320557,
      "learning_rate": 1.1926925208410784e-05,
      "loss": 0.4602,
      "step": 14475
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5880693197250366,
      "learning_rate": 1.1925925873778496e-05,
      "loss": 0.51,
      "step": 14476
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5949891805648804,
      "learning_rate": 1.1924926519171279e-05,
      "loss": 0.4636,
      "step": 14477
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5573054552078247,
      "learning_rate": 1.1923927144599495e-05,
      "loss": 0.4168,
      "step": 14478
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5947316884994507,
      "learning_rate": 1.1922927750073507e-05,
      "loss": 0.4737,
      "step": 14479
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6631635427474976,
      "learning_rate": 1.1921928335603683e-05,
      "loss": 0.4662,
      "step": 14480
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7304290533065796,
      "learning_rate": 1.1920928901200395e-05,
      "loss": 0.4869,
      "step": 14481
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5726879835128784,
      "learning_rate": 1.1919929446873995e-05,
      "loss": 0.4517,
      "step": 14482
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6266506910324097,
      "learning_rate": 1.191892997263486e-05,
      "loss": 0.4654,
      "step": 14483
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5879473686218262,
      "learning_rate": 1.1917930478493354e-05,
      "loss": 0.4588,
      "step": 14484
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5245720148086548,
      "learning_rate": 1.191693096445984e-05,
      "loss": 0.5056,
      "step": 14485
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5148115158081055,
      "learning_rate": 1.191593143054469e-05,
      "loss": 0.42,
      "step": 14486
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5337018966674805,
      "learning_rate": 1.1914931876758262e-05,
      "loss": 0.4868,
      "step": 14487
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5985413789749146,
      "learning_rate": 1.191393230311093e-05,
      "loss": 0.4544,
      "step": 14488
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5047731399536133,
      "learning_rate": 1.1912932709613063e-05,
      "loss": 0.4829,
      "step": 14489
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5383845567703247,
      "learning_rate": 1.191193309627502e-05,
      "loss": 0.4251,
      "step": 14490
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5710504055023193,
      "learning_rate": 1.191093346310718e-05,
      "loss": 0.475,
      "step": 14491
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4893372058868408,
      "learning_rate": 1.1909933810119896e-05,
      "loss": 0.4358,
      "step": 14492
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.657301664352417,
      "learning_rate": 1.190893413732355e-05,
      "loss": 0.551,
      "step": 14493
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5324420928955078,
      "learning_rate": 1.1907934444728503e-05,
      "loss": 0.4311,
      "step": 14494
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5092213153839111,
      "learning_rate": 1.1906934732345124e-05,
      "loss": 0.4626,
      "step": 14495
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6567507982254028,
      "learning_rate": 1.1905935000183782e-05,
      "loss": 0.4432,
      "step": 14496
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5440188646316528,
      "learning_rate": 1.1904935248254848e-05,
      "loss": 0.4766,
      "step": 14497
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5815249681472778,
      "learning_rate": 1.1903935476568687e-05,
      "loss": 0.4516,
      "step": 14498
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5793312788009644,
      "learning_rate": 1.1902935685135674e-05,
      "loss": 0.5209,
      "step": 14499
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5110092163085938,
      "learning_rate": 1.190193587396617e-05,
      "loss": 0.3903,
      "step": 14500
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6018229722976685,
      "learning_rate": 1.1900936043070555e-05,
      "loss": 0.4768,
      "step": 14501
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5698500871658325,
      "learning_rate": 1.1899936192459192e-05,
      "loss": 0.4354,
      "step": 14502
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8276153802871704,
      "learning_rate": 1.1898936322142451e-05,
      "loss": 0.5124,
      "step": 14503
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5676900148391724,
      "learning_rate": 1.1897936432130706e-05,
      "loss": 0.4274,
      "step": 14504
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.53216552734375,
      "learning_rate": 1.1896936522434324e-05,
      "loss": 0.4698,
      "step": 14505
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.529756784439087,
      "learning_rate": 1.1895936593063675e-05,
      "loss": 0.4247,
      "step": 14506
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5631071329116821,
      "learning_rate": 1.1894936644029135e-05,
      "loss": 0.4813,
      "step": 14507
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6286449432373047,
      "learning_rate": 1.189393667534107e-05,
      "loss": 0.4683,
      "step": 14508
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.620524287223816,
      "learning_rate": 1.1892936687009854e-05,
      "loss": 0.5062,
      "step": 14509
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6648128032684326,
      "learning_rate": 1.1891936679045857e-05,
      "loss": 0.4187,
      "step": 14510
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5336416959762573,
      "learning_rate": 1.189093665145945e-05,
      "loss": 0.5022,
      "step": 14511
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6212809085845947,
      "learning_rate": 1.1889936604261007e-05,
      "loss": 0.4403,
      "step": 14512
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5921449661254883,
      "learning_rate": 1.1888936537460902e-05,
      "loss": 0.534,
      "step": 14513
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5386265516281128,
      "learning_rate": 1.1887936451069502e-05,
      "loss": 0.4145,
      "step": 14514
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.537095069885254,
      "learning_rate": 1.1886936345097182e-05,
      "loss": 0.4517,
      "step": 14515
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5397627353668213,
      "learning_rate": 1.1885936219554312e-05,
      "loss": 0.4188,
      "step": 14516
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4818960428237915,
      "learning_rate": 1.1884936074451274e-05,
      "loss": 0.4675,
      "step": 14517
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6912999153137207,
      "learning_rate": 1.188393590979843e-05,
      "loss": 0.4518,
      "step": 14518
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7977958917617798,
      "learning_rate": 1.188293572560616e-05,
      "loss": 0.5459,
      "step": 14519
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5585020780563354,
      "learning_rate": 1.1881935521884835e-05,
      "loss": 0.4211,
      "step": 14520
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5644454956054688,
      "learning_rate": 1.1880935298644831e-05,
      "loss": 0.5117,
      "step": 14521
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5725858211517334,
      "learning_rate": 1.187993505589652e-05,
      "loss": 0.4437,
      "step": 14522
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.552587866783142,
      "learning_rate": 1.1878934793650273e-05,
      "loss": 0.4557,
      "step": 14523
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5509077310562134,
      "learning_rate": 1.1877934511916469e-05,
      "loss": 0.4271,
      "step": 14524
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5654616355895996,
      "learning_rate": 1.1876934210705483e-05,
      "loss": 0.4611,
      "step": 14525
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4962445497512817,
      "learning_rate": 1.1875933890027684e-05,
      "loss": 0.436,
      "step": 14526
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5010555982589722,
      "learning_rate": 1.1874933549893455e-05,
      "loss": 0.456,
      "step": 14527
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5019886493682861,
      "learning_rate": 1.1873933190313165e-05,
      "loss": 0.4173,
      "step": 14528
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5733311176300049,
      "learning_rate": 1.1872932811297192e-05,
      "loss": 0.4693,
      "step": 14529
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.683302402496338,
      "learning_rate": 1.1871932412855908e-05,
      "loss": 0.5196,
      "step": 14530
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.475622296333313,
      "learning_rate": 1.1870931994999694e-05,
      "loss": 0.4345,
      "step": 14531
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.635183572769165,
      "learning_rate": 1.1869931557738925e-05,
      "loss": 0.4315,
      "step": 14532
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.64070725440979,
      "learning_rate": 1.1868931101083974e-05,
      "loss": 0.451,
      "step": 14533
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.085330009460449,
      "learning_rate": 1.1867930625045216e-05,
      "loss": 0.4272,
      "step": 14534
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6392555236816406,
      "learning_rate": 1.1866930129633036e-05,
      "loss": 0.5603,
      "step": 14535
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5026886463165283,
      "learning_rate": 1.1865929614857802e-05,
      "loss": 0.4168,
      "step": 14536
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.496266484260559,
      "learning_rate": 1.1864929080729892e-05,
      "loss": 0.4653,
      "step": 14537
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5203396081924438,
      "learning_rate": 1.186392852725969e-05,
      "loss": 0.4332,
      "step": 14538
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5976756811141968,
      "learning_rate": 1.1862927954457565e-05,
      "loss": 0.4912,
      "step": 14539
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5121455192565918,
      "learning_rate": 1.1861927362333901e-05,
      "loss": 0.4259,
      "step": 14540
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9899965524673462,
      "learning_rate": 1.1860926750899071e-05,
      "loss": 1.0701,
      "step": 14541
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6066374778747559,
      "learning_rate": 1.1859926120163454e-05,
      "loss": 1.024,
      "step": 14542
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5509904623031616,
      "learning_rate": 1.1858925470137432e-05,
      "loss": 0.4829,
      "step": 14543
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6705312728881836,
      "learning_rate": 1.1857924800831375e-05,
      "loss": 0.4411,
      "step": 14544
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6992237567901611,
      "learning_rate": 1.1856924112255673e-05,
      "loss": 0.5007,
      "step": 14545
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4742579460144043,
      "learning_rate": 1.1855923404420694e-05,
      "loss": 0.4088,
      "step": 14546
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5607316493988037,
      "learning_rate": 1.1854922677336824e-05,
      "loss": 0.4846,
      "step": 14547
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6251994371414185,
      "learning_rate": 1.185392193101444e-05,
      "loss": 0.422,
      "step": 14548
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6587947607040405,
      "learning_rate": 1.185292116546392e-05,
      "loss": 0.512,
      "step": 14549
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.493316411972046,
      "learning_rate": 1.1851920380695645e-05,
      "loss": 0.4453,
      "step": 14550
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6099015474319458,
      "learning_rate": 1.1850919576719994e-05,
      "loss": 0.5192,
      "step": 14551
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.26717472076416,
      "learning_rate": 1.1849918753547345e-05,
      "loss": 0.4286,
      "step": 14552
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4903607368469238,
      "learning_rate": 1.1848917911188084e-05,
      "loss": 0.4631,
      "step": 14553
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.589402675628662,
      "learning_rate": 1.1847917049652584e-05,
      "loss": 0.4771,
      "step": 14554
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9466439485549927,
      "learning_rate": 1.1846916168951232e-05,
      "loss": 0.5141,
      "step": 14555
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.501152753829956,
      "learning_rate": 1.1845915269094406e-05,
      "loss": 0.4154,
      "step": 14556
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5625736713409424,
      "learning_rate": 1.1844914350092487e-05,
      "loss": 0.488,
      "step": 14557
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4927469491958618,
      "learning_rate": 1.1843913411955854e-05,
      "loss": 0.4259,
      "step": 14558
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5386723279953003,
      "learning_rate": 1.1842912454694893e-05,
      "loss": 0.4841,
      "step": 14559
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6956590414047241,
      "learning_rate": 1.1841911478319979e-05,
      "loss": 0.4607,
      "step": 14560
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5636875629425049,
      "learning_rate": 1.1840910482841502e-05,
      "loss": 0.4509,
      "step": 14561
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5733546018600464,
      "learning_rate": 1.1839909468269837e-05,
      "loss": 0.4425,
      "step": 14562
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5318002700805664,
      "learning_rate": 1.1838908434615368e-05,
      "loss": 0.4832,
      "step": 14563
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.530620813369751,
      "learning_rate": 1.183790738188848e-05,
      "loss": 0.4477,
      "step": 14564
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5389469861984253,
      "learning_rate": 1.1836906310099551e-05,
      "loss": 0.4713,
      "step": 14565
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5795639753341675,
      "learning_rate": 1.183590521925897e-05,
      "loss": 0.4279,
      "step": 14566
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5491079092025757,
      "learning_rate": 1.1834904109377114e-05,
      "loss": 0.497,
      "step": 14567
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5067083835601807,
      "learning_rate": 1.183390298046437e-05,
      "loss": 0.4084,
      "step": 14568
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5738753080368042,
      "learning_rate": 1.1832901832531116e-05,
      "loss": 0.4662,
      "step": 14569
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5709720849990845,
      "learning_rate": 1.1831900665587739e-05,
      "loss": 0.4244,
      "step": 14570
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5362660884857178,
      "learning_rate": 1.1830899479644627e-05,
      "loss": 0.4895,
      "step": 14571
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5249481201171875,
      "learning_rate": 1.1829898274712154e-05,
      "loss": 0.4466,
      "step": 14572
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5824252367019653,
      "learning_rate": 1.1828897050800714e-05,
      "loss": 0.4709,
      "step": 14573
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6223504543304443,
      "learning_rate": 1.1827895807920686e-05,
      "loss": 0.4061,
      "step": 14574
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5505977869033813,
      "learning_rate": 1.1826894546082457e-05,
      "loss": 0.4481,
      "step": 14575
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.558117389678955,
      "learning_rate": 1.1825893265296408e-05,
      "loss": 0.4626,
      "step": 14576
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.48970365524292,
      "learning_rate": 1.1824891965572926e-05,
      "loss": 0.4493,
      "step": 14577
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6389069557189941,
      "learning_rate": 1.1823890646922396e-05,
      "loss": 0.4702,
      "step": 14578
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5644570589065552,
      "learning_rate": 1.1822889309355206e-05,
      "loss": 0.4802,
      "step": 14579
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5808149576187134,
      "learning_rate": 1.1821887952881737e-05,
      "loss": 0.4452,
      "step": 14580
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.604141354560852,
      "learning_rate": 1.1820886577512377e-05,
      "loss": 0.4666,
      "step": 14581
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4001392126083374,
      "learning_rate": 1.1819885183257512e-05,
      "loss": 0.3917,
      "step": 14582
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.584722876548767,
      "learning_rate": 1.1818883770127526e-05,
      "loss": 0.5173,
      "step": 14583
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4742501974105835,
      "learning_rate": 1.1817882338132806e-05,
      "loss": 0.4085,
      "step": 14584
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5921778678894043,
      "learning_rate": 1.1816880887283742e-05,
      "loss": 0.4516,
      "step": 14585
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5395158529281616,
      "learning_rate": 1.1815879417590715e-05,
      "loss": 0.4093,
      "step": 14586
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.627326488494873,
      "learning_rate": 1.1814877929064119e-05,
      "loss": 0.5411,
      "step": 14587
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.480425477027893,
      "learning_rate": 1.1813876421714331e-05,
      "loss": 0.413,
      "step": 14588
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5548352003097534,
      "learning_rate": 1.1812874895551749e-05,
      "loss": 0.5239,
      "step": 14589
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5450365543365479,
      "learning_rate": 1.1811873350586753e-05,
      "loss": 0.4105,
      "step": 14590
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.728642463684082,
      "learning_rate": 1.1810871786829735e-05,
      "loss": 0.5083,
      "step": 14591
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4736316204071045,
      "learning_rate": 1.1809870204291079e-05,
      "loss": 0.4244,
      "step": 14592
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6474109888076782,
      "learning_rate": 1.1808868602981173e-05,
      "loss": 0.5381,
      "step": 14593
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6050641536712646,
      "learning_rate": 1.180786698291041e-05,
      "loss": 0.4261,
      "step": 14594
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9266568422317505,
      "learning_rate": 1.1806865344089176e-05,
      "loss": 0.9864,
      "step": 14595
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7709890604019165,
      "learning_rate": 1.1805863686527856e-05,
      "loss": 1.0738,
      "step": 14596
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6192189455032349,
      "learning_rate": 1.1804862010236846e-05,
      "loss": 0.4868,
      "step": 14597
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6552793979644775,
      "learning_rate": 1.1803860315226524e-05,
      "loss": 0.4659,
      "step": 14598
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5687119960784912,
      "learning_rate": 1.1802858601507292e-05,
      "loss": 0.4703,
      "step": 14599
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6255829334259033,
      "learning_rate": 1.1801856869089532e-05,
      "loss": 0.4696,
      "step": 14600
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5029793977737427,
      "learning_rate": 1.1800855117983635e-05,
      "loss": 0.4184,
      "step": 14601
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5778844356536865,
      "learning_rate": 1.1799853348199993e-05,
      "loss": 0.4322,
      "step": 14602
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6778959035873413,
      "learning_rate": 1.179885155974899e-05,
      "loss": 0.5101,
      "step": 14603
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5940539836883545,
      "learning_rate": 1.179784975264102e-05,
      "loss": 0.4541,
      "step": 14604
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6189755201339722,
      "learning_rate": 1.1796847926886474e-05,
      "loss": 0.4712,
      "step": 14605
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5813889503479004,
      "learning_rate": 1.1795846082495741e-05,
      "loss": 0.4236,
      "step": 14606
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6944069862365723,
      "learning_rate": 1.1794844219479216e-05,
      "loss": 0.4618,
      "step": 14607
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5434330701828003,
      "learning_rate": 1.179384233784728e-05,
      "loss": 0.4224,
      "step": 14608
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5719813108444214,
      "learning_rate": 1.1792840437610336e-05,
      "loss": 0.4616,
      "step": 14609
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.551692247390747,
      "learning_rate": 1.179183851877877e-05,
      "loss": 0.4112,
      "step": 14610
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9600657224655151,
      "learning_rate": 1.179083658136297e-05,
      "loss": 1.0491,
      "step": 14611
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.724038004875183,
      "learning_rate": 1.178983462537333e-05,
      "loss": 1.0258,
      "step": 14612
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6957377195358276,
      "learning_rate": 1.1788832650820246e-05,
      "loss": 0.4743,
      "step": 14613
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5942798852920532,
      "learning_rate": 1.1787830657714104e-05,
      "loss": 0.4118,
      "step": 14614
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6185030937194824,
      "learning_rate": 1.1786828646065302e-05,
      "loss": 0.4623,
      "step": 14615
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.488769769668579,
      "learning_rate": 1.1785826615884228e-05,
      "loss": 0.4184,
      "step": 14616
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.726711630821228,
      "learning_rate": 1.1784824567181276e-05,
      "loss": 0.5071,
      "step": 14617
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.629335880279541,
      "learning_rate": 1.178382249996684e-05,
      "loss": 0.4647,
      "step": 14618
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7880769968032837,
      "learning_rate": 1.1782820414251315e-05,
      "loss": 0.499,
      "step": 14619
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5260549783706665,
      "learning_rate": 1.1781818310045088e-05,
      "loss": 0.4206,
      "step": 14620
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.535546064376831,
      "learning_rate": 1.1780816187358556e-05,
      "loss": 0.44,
      "step": 14621
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5097486972808838,
      "learning_rate": 1.1779814046202112e-05,
      "loss": 0.4263,
      "step": 14622
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5471398830413818,
      "learning_rate": 1.1778811886586155e-05,
      "loss": 0.5195,
      "step": 14623
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.53960120677948,
      "learning_rate": 1.1777809708521067e-05,
      "loss": 0.4326,
      "step": 14624
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.601802945137024,
      "learning_rate": 1.1776807512017257e-05,
      "loss": 0.4538,
      "step": 14625
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6081104278564453,
      "learning_rate": 1.1775805297085106e-05,
      "loss": 0.42,
      "step": 14626
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6081091165542603,
      "learning_rate": 1.1774803063735017e-05,
      "loss": 0.478,
      "step": 14627
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6201456785202026,
      "learning_rate": 1.177380081197738e-05,
      "loss": 0.4393,
      "step": 14628
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5584770441055298,
      "learning_rate": 1.1772798541822595e-05,
      "loss": 0.4947,
      "step": 14629
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5271344184875488,
      "learning_rate": 1.1771796253281051e-05,
      "loss": 0.4128,
      "step": 14630
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4984874725341797,
      "learning_rate": 1.1770793946363151e-05,
      "loss": 0.4818,
      "step": 14631
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5003094673156738,
      "learning_rate": 1.1769791621079281e-05,
      "loss": 0.4395,
      "step": 14632
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.529685616493225,
      "learning_rate": 1.1768789277439848e-05,
      "loss": 0.48,
      "step": 14633
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5936477184295654,
      "learning_rate": 1.1767786915455235e-05,
      "loss": 0.4657,
      "step": 14634
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5555311441421509,
      "learning_rate": 1.176678453513585e-05,
      "loss": 0.4938,
      "step": 14635
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.517449140548706,
      "learning_rate": 1.1765782136492081e-05,
      "loss": 0.4509,
      "step": 14636
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.710592269897461,
      "learning_rate": 1.176477971953433e-05,
      "loss": 0.5081,
      "step": 14637
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.071092128753662,
      "learning_rate": 1.1763777284272988e-05,
      "loss": 0.4409,
      "step": 14638
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6691731214523315,
      "learning_rate": 1.1762774830718458e-05,
      "loss": 0.511,
      "step": 14639
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5017199516296387,
      "learning_rate": 1.1761772358881132e-05,
      "loss": 0.4136,
      "step": 14640
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5852673053741455,
      "learning_rate": 1.1760769868771414e-05,
      "loss": 0.4803,
      "step": 14641
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5622931718826294,
      "learning_rate": 1.175976736039969e-05,
      "loss": 0.4352,
      "step": 14642
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4755642414093018,
      "learning_rate": 1.1758764833776373e-05,
      "loss": 0.4579,
      "step": 14643
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5174353122711182,
      "learning_rate": 1.1757762288911844e-05,
      "loss": 0.4747,
      "step": 14644
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5077787637710571,
      "learning_rate": 1.1756759725816514e-05,
      "loss": 0.4889,
      "step": 14645
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5196841955184937,
      "learning_rate": 1.1755757144500774e-05,
      "loss": 0.4335,
      "step": 14646
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4960191249847412,
      "learning_rate": 1.1754754544975027e-05,
      "loss": 0.4722,
      "step": 14647
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4544198513031006,
      "learning_rate": 1.1753751927249668e-05,
      "loss": 0.435,
      "step": 14648
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5403748750686646,
      "learning_rate": 1.1752749291335098e-05,
      "loss": 0.5022,
      "step": 14649
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5217372179031372,
      "learning_rate": 1.1751746637241713e-05,
      "loss": 0.4284,
      "step": 14650
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.489835500717163,
      "learning_rate": 1.1750743964979919e-05,
      "loss": 0.4739,
      "step": 14651
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5629068613052368,
      "learning_rate": 1.1749741274560105e-05,
      "loss": 0.4465,
      "step": 14652
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6516368389129639,
      "learning_rate": 1.1748738565992676e-05,
      "loss": 0.5007,
      "step": 14653
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.49203622341156,
      "learning_rate": 1.1747735839288036e-05,
      "loss": 0.4106,
      "step": 14654
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5759587287902832,
      "learning_rate": 1.1746733094456577e-05,
      "loss": 0.5016,
      "step": 14655
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6062730550765991,
      "learning_rate": 1.1745730331508704e-05,
      "loss": 0.4517,
      "step": 14656
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6267141103744507,
      "learning_rate": 1.1744727550454818e-05,
      "loss": 0.5518,
      "step": 14657
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.617769479751587,
      "learning_rate": 1.1743724751305315e-05,
      "loss": 0.4253,
      "step": 14658
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4847984313964844,
      "learning_rate": 1.17427219340706e-05,
      "loss": 0.4928,
      "step": 14659
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4602309465408325,
      "learning_rate": 1.1741719098761067e-05,
      "loss": 0.406,
      "step": 14660
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6286641359329224,
      "learning_rate": 1.1740716245387125e-05,
      "loss": 0.4993,
      "step": 14661
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6076308488845825,
      "learning_rate": 1.1739713373959172e-05,
      "loss": 0.4395,
      "step": 14662
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.589699149131775,
      "learning_rate": 1.173871048448761e-05,
      "loss": 0.4448,
      "step": 14663
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5490939617156982,
      "learning_rate": 1.173770757698284e-05,
      "loss": 0.451,
      "step": 14664
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5184441804885864,
      "learning_rate": 1.1736704651455262e-05,
      "loss": 0.4655,
      "step": 14665
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6328681707382202,
      "learning_rate": 1.1735701707915281e-05,
      "loss": 0.4297,
      "step": 14666
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5949517488479614,
      "learning_rate": 1.17346987463733e-05,
      "loss": 0.4566,
      "step": 14667
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4598218202590942,
      "learning_rate": 1.1733695766839712e-05,
      "loss": 0.4337,
      "step": 14668
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5982880592346191,
      "learning_rate": 1.1732692769324934e-05,
      "loss": 0.5235,
      "step": 14669
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5456337928771973,
      "learning_rate": 1.1731689753839355e-05,
      "loss": 0.405,
      "step": 14670
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5552517175674438,
      "learning_rate": 1.173068672039339e-05,
      "loss": 0.4852,
      "step": 14671
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5392762422561646,
      "learning_rate": 1.172968366899743e-05,
      "loss": 0.4311,
      "step": 14672
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6042622327804565,
      "learning_rate": 1.172868059966189e-05,
      "loss": 0.4863,
      "step": 14673
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5188658237457275,
      "learning_rate": 1.1727677512397167e-05,
      "loss": 0.4199,
      "step": 14674
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5831267833709717,
      "learning_rate": 1.1726674407213666e-05,
      "loss": 0.465,
      "step": 14675
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5273123979568481,
      "learning_rate": 1.1725671284121785e-05,
      "loss": 0.4174,
      "step": 14676
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5539671182632446,
      "learning_rate": 1.1724668143131938e-05,
      "loss": 0.5454,
      "step": 14677
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5894510746002197,
      "learning_rate": 1.1723664984254523e-05,
      "loss": 0.4117,
      "step": 14678
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5777300596237183,
      "learning_rate": 1.1722661807499946e-05,
      "loss": 0.5043,
      "step": 14679
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6376214027404785,
      "learning_rate": 1.1721658612878612e-05,
      "loss": 0.4599,
      "step": 14680
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4725364446640015,
      "learning_rate": 1.1720655400400927e-05,
      "loss": 0.4806,
      "step": 14681
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5322282314300537,
      "learning_rate": 1.1719652170077291e-05,
      "loss": 0.4185,
      "step": 14682
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5365664958953857,
      "learning_rate": 1.1718648921918112e-05,
      "loss": 0.4745,
      "step": 14683
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5322141647338867,
      "learning_rate": 1.1717645655933797e-05,
      "loss": 0.4245,
      "step": 14684
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.527745008468628,
      "learning_rate": 1.171664237213475e-05,
      "loss": 0.4694,
      "step": 14685
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.645704984664917,
      "learning_rate": 1.1715639070531373e-05,
      "loss": 0.4462,
      "step": 14686
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.715684413909912,
      "learning_rate": 1.1714635751134083e-05,
      "loss": 0.5624,
      "step": 14687
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5392905473709106,
      "learning_rate": 1.171363241395327e-05,
      "loss": 0.4025,
      "step": 14688
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6573023796081543,
      "learning_rate": 1.1712629058999351e-05,
      "loss": 0.4791,
      "step": 14689
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5066235065460205,
      "learning_rate": 1.1711625686282733e-05,
      "loss": 0.447,
      "step": 14690
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6173386573791504,
      "learning_rate": 1.1710622295813818e-05,
      "loss": 0.4787,
      "step": 14691
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.545896291732788,
      "learning_rate": 1.1709618887603013e-05,
      "loss": 0.4116,
      "step": 14692
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.772255539894104,
      "learning_rate": 1.1708615461660728e-05,
      "loss": 0.9848,
      "step": 14693
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7445604801177979,
      "learning_rate": 1.1707612017997367e-05,
      "loss": 1.0917,
      "step": 14694
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6443204879760742,
      "learning_rate": 1.170660855662334e-05,
      "loss": 0.4704,
      "step": 14695
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5620285272598267,
      "learning_rate": 1.1705605077549053e-05,
      "loss": 0.4213,
      "step": 14696
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6107525825500488,
      "learning_rate": 1.1704601580784916e-05,
      "loss": 0.4704,
      "step": 14697
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5968626737594604,
      "learning_rate": 1.1703598066341334e-05,
      "loss": 0.4505,
      "step": 14698
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.572561264038086,
      "learning_rate": 1.170259453422871e-05,
      "loss": 0.4543,
      "step": 14699
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5287996530532837,
      "learning_rate": 1.1701590984457467e-05,
      "loss": 0.4145,
      "step": 14700
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6201468706130981,
      "learning_rate": 1.1700587417037999e-05,
      "loss": 0.5257,
      "step": 14701
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5052194595336914,
      "learning_rate": 1.1699583831980724e-05,
      "loss": 0.4335,
      "step": 14702
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.45806884765625,
      "learning_rate": 1.169858022929604e-05,
      "loss": 0.4706,
      "step": 14703
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6216773986816406,
      "learning_rate": 1.1697576608994367e-05,
      "loss": 0.4281,
      "step": 14704
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.627790093421936,
      "learning_rate": 1.169657297108611e-05,
      "loss": 0.4706,
      "step": 14705
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5932480096817017,
      "learning_rate": 1.1695569315581681e-05,
      "loss": 0.4483,
      "step": 14706
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5045113563537598,
      "learning_rate": 1.1694565642491483e-05,
      "loss": 0.4674,
      "step": 14707
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.571089506149292,
      "learning_rate": 1.1693561951825932e-05,
      "loss": 0.434,
      "step": 14708
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6126850843429565,
      "learning_rate": 1.1692558243595431e-05,
      "loss": 0.4801,
      "step": 14709
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6080634593963623,
      "learning_rate": 1.1691554517810401e-05,
      "loss": 0.4036,
      "step": 14710
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4871339797973633,
      "learning_rate": 1.1690550774481239e-05,
      "loss": 0.4452,
      "step": 14711
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5737050771713257,
      "learning_rate": 1.1689547013618368e-05,
      "loss": 0.4518,
      "step": 14712
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5980387926101685,
      "learning_rate": 1.168854323523219e-05,
      "loss": 0.5064,
      "step": 14713
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5200326442718506,
      "learning_rate": 1.168753943933312e-05,
      "loss": 0.4127,
      "step": 14714
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6008466482162476,
      "learning_rate": 1.1686535625931566e-05,
      "loss": 0.4948,
      "step": 14715
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5689843893051147,
      "learning_rate": 1.1685531795037942e-05,
      "loss": 0.4448,
      "step": 14716
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5897362232208252,
      "learning_rate": 1.1684527946662653e-05,
      "loss": 0.4863,
      "step": 14717
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5036940574645996,
      "learning_rate": 1.1683524080816121e-05,
      "loss": 0.435,
      "step": 14718
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6768882274627686,
      "learning_rate": 1.168252019750875e-05,
      "loss": 0.5923,
      "step": 14719
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5070345401763916,
      "learning_rate": 1.1681516296750956e-05,
      "loss": 0.415,
      "step": 14720
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6520315408706665,
      "learning_rate": 1.1680512378553148e-05,
      "loss": 0.4771,
      "step": 14721
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.521913766860962,
      "learning_rate": 1.167950844292574e-05,
      "loss": 0.4371,
      "step": 14722
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5760607719421387,
      "learning_rate": 1.1678504489879143e-05,
      "loss": 0.4864,
      "step": 14723
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5012446641921997,
      "learning_rate": 1.1677500519423771e-05,
      "loss": 0.4226,
      "step": 14724
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6149147748947144,
      "learning_rate": 1.1676496531570034e-05,
      "loss": 0.4714,
      "step": 14725
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.573647379875183,
      "learning_rate": 1.1675492526328349e-05,
      "loss": 0.4395,
      "step": 14726
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.455221176147461,
      "learning_rate": 1.1674488503709125e-05,
      "loss": 0.4369,
      "step": 14727
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5992134809494019,
      "learning_rate": 1.1673484463722782e-05,
      "loss": 0.4641,
      "step": 14728
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5733193159103394,
      "learning_rate": 1.1672480406379726e-05,
      "loss": 0.5238,
      "step": 14729
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5213220119476318,
      "learning_rate": 1.1671476331690375e-05,
      "loss": 0.3968,
      "step": 14730
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4504259824752808,
      "learning_rate": 1.1670472239665141e-05,
      "loss": 0.4595,
      "step": 14731
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5174505710601807,
      "learning_rate": 1.1669468130314438e-05,
      "loss": 0.4423,
      "step": 14732
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5987963676452637,
      "learning_rate": 1.166846400364868e-05,
      "loss": 0.5144,
      "step": 14733
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5120586156845093,
      "learning_rate": 1.1667459859678284e-05,
      "loss": 0.4328,
      "step": 14734
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6010711193084717,
      "learning_rate": 1.166645569841366e-05,
      "loss": 0.4764,
      "step": 14735
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6171941757202148,
      "learning_rate": 1.1665451519865233e-05,
      "loss": 0.4536,
      "step": 14736
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6361136436462402,
      "learning_rate": 1.1664447324043404e-05,
      "loss": 0.5035,
      "step": 14737
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6141446828842163,
      "learning_rate": 1.1663443110958597e-05,
      "loss": 0.4492,
      "step": 14738
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5646909475326538,
      "learning_rate": 1.1662438880621223e-05,
      "loss": 0.4571,
      "step": 14739
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5494221448898315,
      "learning_rate": 1.1661434633041702e-05,
      "loss": 0.4139,
      "step": 14740
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5514363050460815,
      "learning_rate": 1.1660430368230447e-05,
      "loss": 0.447,
      "step": 14741
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5965347290039062,
      "learning_rate": 1.165942608619787e-05,
      "loss": 0.448,
      "step": 14742
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5899566411972046,
      "learning_rate": 1.1658421786954395e-05,
      "loss": 0.5147,
      "step": 14743
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4824272394180298,
      "learning_rate": 1.1657417470510435e-05,
      "loss": 0.4274,
      "step": 14744
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.607075810432434,
      "learning_rate": 1.16564131368764e-05,
      "loss": 0.4991,
      "step": 14745
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5612987279891968,
      "learning_rate": 1.1655408786062718e-05,
      "loss": 0.4378,
      "step": 14746
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5296313762664795,
      "learning_rate": 1.1654404418079795e-05,
      "loss": 0.5103,
      "step": 14747
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5140454769134521,
      "learning_rate": 1.1653400032938056e-05,
      "loss": 0.4126,
      "step": 14748
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5145758390426636,
      "learning_rate": 1.1652395630647912e-05,
      "loss": 0.468,
      "step": 14749
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5347739458084106,
      "learning_rate": 1.1651391211219787e-05,
      "loss": 0.415,
      "step": 14750
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7104411125183105,
      "learning_rate": 1.1650386774664092e-05,
      "loss": 0.5074,
      "step": 14751
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.536024570465088,
      "learning_rate": 1.1649382320991246e-05,
      "loss": 0.4349,
      "step": 14752
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5988723039627075,
      "learning_rate": 1.1648377850211668e-05,
      "loss": 0.4887,
      "step": 14753
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5918382406234741,
      "learning_rate": 1.164737336233578e-05,
      "loss": 0.4253,
      "step": 14754
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6274052858352661,
      "learning_rate": 1.1646368857373988e-05,
      "loss": 0.477,
      "step": 14755
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.4114904403686523,
      "learning_rate": 1.1645364335336724e-05,
      "loss": 0.4696,
      "step": 14756
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6984964609146118,
      "learning_rate": 1.1644359796234399e-05,
      "loss": 0.5049,
      "step": 14757
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5166593790054321,
      "learning_rate": 1.1643355240077435e-05,
      "loss": 0.4203,
      "step": 14758
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6113494634628296,
      "learning_rate": 1.164235066687625e-05,
      "loss": 0.5049,
      "step": 14759
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5721049308776855,
      "learning_rate": 1.164134607664126e-05,
      "loss": 0.4265,
      "step": 14760
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6252752542495728,
      "learning_rate": 1.1640341469382887e-05,
      "loss": 0.5444,
      "step": 14761
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5099623203277588,
      "learning_rate": 1.163933684511155e-05,
      "loss": 0.4271,
      "step": 14762
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5417982339859009,
      "learning_rate": 1.1638332203837668e-05,
      "loss": 0.5588,
      "step": 14763
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4854973554611206,
      "learning_rate": 1.1637327545571663e-05,
      "loss": 0.4193,
      "step": 14764
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6081305742263794,
      "learning_rate": 1.163632287032395e-05,
      "loss": 0.4973,
      "step": 14765
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5699366331100464,
      "learning_rate": 1.1635318178104955e-05,
      "loss": 0.4495,
      "step": 14766
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7111903429031372,
      "learning_rate": 1.1634313468925095e-05,
      "loss": 0.4655,
      "step": 14767
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6044598817825317,
      "learning_rate": 1.163330874279479e-05,
      "loss": 0.4282,
      "step": 14768
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5565192699432373,
      "learning_rate": 1.1632303999724463e-05,
      "loss": 0.4702,
      "step": 14769
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7037127017974854,
      "learning_rate": 1.1631299239724534e-05,
      "loss": 0.4466,
      "step": 14770
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5812236070632935,
      "learning_rate": 1.1630294462805419e-05,
      "loss": 0.5042,
      "step": 14771
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5930440425872803,
      "learning_rate": 1.1629289668977549e-05,
      "loss": 0.4327,
      "step": 14772
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7149070501327515,
      "learning_rate": 1.1628284858251336e-05,
      "loss": 0.4705,
      "step": 14773
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6758736371994019,
      "learning_rate": 1.1627280030637205e-05,
      "loss": 0.4423,
      "step": 14774
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5682586431503296,
      "learning_rate": 1.1626275186145579e-05,
      "loss": 0.5421,
      "step": 14775
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5121207237243652,
      "learning_rate": 1.1625270324786879e-05,
      "loss": 0.3996,
      "step": 14776
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5810418128967285,
      "learning_rate": 1.1624265446571527e-05,
      "loss": 0.5597,
      "step": 14777
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5272575616836548,
      "learning_rate": 1.1623260551509947e-05,
      "loss": 0.4303,
      "step": 14778
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5994120836257935,
      "learning_rate": 1.1622255639612553e-05,
      "loss": 0.4853,
      "step": 14779
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.593002200126648,
      "learning_rate": 1.1621250710889782e-05,
      "loss": 0.4655,
      "step": 14780
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6627837419509888,
      "learning_rate": 1.1620245765352043e-05,
      "loss": 0.5031,
      "step": 14781
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6050549745559692,
      "learning_rate": 1.1619240803009768e-05,
      "loss": 0.4194,
      "step": 14782
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5245579481124878,
      "learning_rate": 1.1618235823873371e-05,
      "loss": 0.4342,
      "step": 14783
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4871643781661987,
      "learning_rate": 1.1617230827953288e-05,
      "loss": 0.4281,
      "step": 14784
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6158539056777954,
      "learning_rate": 1.1616225815259932e-05,
      "loss": 0.5301,
      "step": 14785
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5403680801391602,
      "learning_rate": 1.1615220785803729e-05,
      "loss": 0.4262,
      "step": 14786
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5507547855377197,
      "learning_rate": 1.1614215739595104e-05,
      "loss": 0.4584,
      "step": 14787
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4841253757476807,
      "learning_rate": 1.161321067664448e-05,
      "loss": 0.4357,
      "step": 14788
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6341763734817505,
      "learning_rate": 1.1612205596962281e-05,
      "loss": 0.4807,
      "step": 14789
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.567997932434082,
      "learning_rate": 1.1611200500558936e-05,
      "loss": 0.4582,
      "step": 14790
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5419197082519531,
      "learning_rate": 1.161019538744486e-05,
      "loss": 0.4857,
      "step": 14791
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5384260416030884,
      "learning_rate": 1.1609190257630487e-05,
      "loss": 0.437,
      "step": 14792
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5690220594406128,
      "learning_rate": 1.1608185111126236e-05,
      "loss": 0.4517,
      "step": 14793
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6749769449234009,
      "learning_rate": 1.1607179947942536e-05,
      "loss": 0.4705,
      "step": 14794
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.623615026473999,
      "learning_rate": 1.1606174768089807e-05,
      "loss": 0.4485,
      "step": 14795
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5501035451889038,
      "learning_rate": 1.1605169571578479e-05,
      "loss": 0.4089,
      "step": 14796
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.653019666671753,
      "learning_rate": 1.1604164358418972e-05,
      "loss": 0.5021,
      "step": 14797
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6020417213439941,
      "learning_rate": 1.1603159128621721e-05,
      "loss": 0.4188,
      "step": 14798
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6394144296646118,
      "learning_rate": 1.1602153882197143e-05,
      "loss": 0.4764,
      "step": 14799
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5372796058654785,
      "learning_rate": 1.1601148619155671e-05,
      "loss": 0.4604,
      "step": 14800
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6785355806350708,
      "learning_rate": 1.160014333950772e-05,
      "loss": 0.992,
      "step": 14801
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.50752592086792,
      "learning_rate": 1.1599138043263732e-05,
      "loss": 1.0101,
      "step": 14802
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7004750967025757,
      "learning_rate": 1.159813273043412e-05,
      "loss": 0.5143,
      "step": 14803
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5942928791046143,
      "learning_rate": 1.1597127401029318e-05,
      "loss": 0.4159,
      "step": 14804
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6345020532608032,
      "learning_rate": 1.1596122055059751e-05,
      "loss": 0.4582,
      "step": 14805
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.777523398399353,
      "learning_rate": 1.1595116692535846e-05,
      "loss": 0.4376,
      "step": 14806
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6200569868087769,
      "learning_rate": 1.1594111313468028e-05,
      "loss": 0.4311,
      "step": 14807
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7449500560760498,
      "learning_rate": 1.159310591786673e-05,
      "loss": 0.4318,
      "step": 14808
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6007188558578491,
      "learning_rate": 1.1592100505742372e-05,
      "loss": 0.4524,
      "step": 14809
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6088770627975464,
      "learning_rate": 1.159109507710539e-05,
      "loss": 0.4333,
      "step": 14810
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6160995960235596,
      "learning_rate": 1.1590089631966206e-05,
      "loss": 0.4547,
      "step": 14811
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.528153657913208,
      "learning_rate": 1.158908417033525e-05,
      "loss": 0.4094,
      "step": 14812
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.63935387134552,
      "learning_rate": 1.1588078692222951e-05,
      "loss": 0.5636,
      "step": 14813
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.552991271018982,
      "learning_rate": 1.1587073197639735e-05,
      "loss": 0.411,
      "step": 14814
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5696290731430054,
      "learning_rate": 1.1586067686596029e-05,
      "loss": 0.4881,
      "step": 14815
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5427948236465454,
      "learning_rate": 1.1585062159102272e-05,
      "loss": 0.3931,
      "step": 14816
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6515153646469116,
      "learning_rate": 1.158405661516888e-05,
      "loss": 0.5621,
      "step": 14817
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6500393152236938,
      "learning_rate": 1.1583051054806293e-05,
      "loss": 0.4618,
      "step": 14818
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6470052003860474,
      "learning_rate": 1.1582045478024931e-05,
      "loss": 0.4925,
      "step": 14819
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5744435787200928,
      "learning_rate": 1.1581039884835228e-05,
      "loss": 0.4022,
      "step": 14820
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5375686883926392,
      "learning_rate": 1.1580034275247614e-05,
      "loss": 0.4573,
      "step": 14821
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5123212337493896,
      "learning_rate": 1.157902864927252e-05,
      "loss": 0.4154,
      "step": 14822
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6888514757156372,
      "learning_rate": 1.1578023006920369e-05,
      "loss": 0.4997,
      "step": 14823
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4462034702301025,
      "learning_rate": 1.15770173482016e-05,
      "loss": 0.3902,
      "step": 14824
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6046642065048218,
      "learning_rate": 1.1576011673126637e-05,
      "loss": 0.4514,
      "step": 14825
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.512251615524292,
      "learning_rate": 1.1575005981705916e-05,
      "loss": 0.4059,
      "step": 14826
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5736335515975952,
      "learning_rate": 1.1574000273949859e-05,
      "loss": 0.4825,
      "step": 14827
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4749205112457275,
      "learning_rate": 1.1572994549868908e-05,
      "loss": 0.4165,
      "step": 14828
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5740102529525757,
      "learning_rate": 1.1571988809473485e-05,
      "loss": 0.4615,
      "step": 14829
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5426054000854492,
      "learning_rate": 1.1570983052774025e-05,
      "loss": 0.4397,
      "step": 14830
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5125985145568848,
      "learning_rate": 1.1569977279780959e-05,
      "loss": 0.4448,
      "step": 14831
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5703381299972534,
      "learning_rate": 1.156897149050472e-05,
      "loss": 0.451,
      "step": 14832
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.562495470046997,
      "learning_rate": 1.1567965684955731e-05,
      "loss": 0.5173,
      "step": 14833
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5824313163757324,
      "learning_rate": 1.1566959863144438e-05,
      "loss": 0.4398,
      "step": 14834
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4970686435699463,
      "learning_rate": 1.156595402508126e-05,
      "loss": 0.4582,
      "step": 14835
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5565227270126343,
      "learning_rate": 1.156494817077664e-05,
      "loss": 0.4403,
      "step": 14836
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.520786166191101,
      "learning_rate": 1.1563942300241e-05,
      "loss": 0.4808,
      "step": 14837
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.472529649734497,
      "learning_rate": 1.156293641348478e-05,
      "loss": 0.4021,
      "step": 14838
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.542878270149231,
      "learning_rate": 1.1561930510518411e-05,
      "loss": 0.4763,
      "step": 14839
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7644939422607422,
      "learning_rate": 1.1560924591352322e-05,
      "loss": 0.4602,
      "step": 14840
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5640811920166016,
      "learning_rate": 1.1559918655996953e-05,
      "loss": 0.5078,
      "step": 14841
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.485783338546753,
      "learning_rate": 1.155891270446273e-05,
      "loss": 0.4306,
      "step": 14842
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4912159442901611,
      "learning_rate": 1.1557906736760089e-05,
      "loss": 0.4723,
      "step": 14843
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5612229108810425,
      "learning_rate": 1.1556900752899466e-05,
      "loss": 0.4393,
      "step": 14844
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7324527502059937,
      "learning_rate": 1.155589475289129e-05,
      "loss": 0.5037,
      "step": 14845
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7537661790847778,
      "learning_rate": 1.1554888736745999e-05,
      "loss": 0.4509,
      "step": 14846
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5738575458526611,
      "learning_rate": 1.1553882704474026e-05,
      "loss": 0.5199,
      "step": 14847
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4950255155563354,
      "learning_rate": 1.1552876656085805e-05,
      "loss": 0.4331,
      "step": 14848
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5212570428848267,
      "learning_rate": 1.155187059159177e-05,
      "loss": 0.4419,
      "step": 14849
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4782917499542236,
      "learning_rate": 1.1550864511002356e-05,
      "loss": 0.4424,
      "step": 14850
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.513197660446167,
      "learning_rate": 1.1549858414327992e-05,
      "loss": 0.4861,
      "step": 14851
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5058482885360718,
      "learning_rate": 1.1548852301579123e-05,
      "loss": 0.4189,
      "step": 14852
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5347614288330078,
      "learning_rate": 1.1547846172766177e-05,
      "loss": 0.447,
      "step": 14853
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4879438877105713,
      "learning_rate": 1.154684002789959e-05,
      "loss": 0.416,
      "step": 14854
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5366584062576294,
      "learning_rate": 1.1545833866989802e-05,
      "loss": 0.4653,
      "step": 14855
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5076669454574585,
      "learning_rate": 1.1544827690047243e-05,
      "loss": 0.4421,
      "step": 14856
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4957866668701172,
      "learning_rate": 1.1543821497082348e-05,
      "loss": 0.4593,
      "step": 14857
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.530756950378418,
      "learning_rate": 1.1542815288105558e-05,
      "loss": 0.4268,
      "step": 14858
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6544464826583862,
      "learning_rate": 1.1541809063127307e-05,
      "loss": 0.4736,
      "step": 14859
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.626821756362915,
      "learning_rate": 1.1540802822158028e-05,
      "loss": 0.3949,
      "step": 14860
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6654645204544067,
      "learning_rate": 1.1539796565208162e-05,
      "loss": 0.4966,
      "step": 14861
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.51166832447052,
      "learning_rate": 1.1538790292288144e-05,
      "loss": 0.3961,
      "step": 14862
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7026702165603638,
      "learning_rate": 1.1537784003408406e-05,
      "loss": 0.4978,
      "step": 14863
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4998400211334229,
      "learning_rate": 1.1536777698579393e-05,
      "loss": 0.4256,
      "step": 14864
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6338353157043457,
      "learning_rate": 1.1535771377811536e-05,
      "loss": 0.5215,
      "step": 14865
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5968143939971924,
      "learning_rate": 1.1534765041115275e-05,
      "loss": 0.407,
      "step": 14866
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5544633865356445,
      "learning_rate": 1.1533758688501044e-05,
      "loss": 0.4935,
      "step": 14867
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5226936340332031,
      "learning_rate": 1.1532752319979286e-05,
      "loss": 0.4208,
      "step": 14868
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5441465377807617,
      "learning_rate": 1.1531745935560433e-05,
      "loss": 0.4763,
      "step": 14869
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.706085443496704,
      "learning_rate": 1.1530739535254927e-05,
      "loss": 0.4615,
      "step": 14870
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7126576900482178,
      "learning_rate": 1.1529733119073203e-05,
      "loss": 0.4903,
      "step": 14871
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5041440725326538,
      "learning_rate": 1.15287266870257e-05,
      "loss": 0.3895,
      "step": 14872
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4923834800720215,
      "learning_rate": 1.1527720239122857e-05,
      "loss": 0.4951,
      "step": 14873
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.681159257888794,
      "learning_rate": 1.1526713775375113e-05,
      "loss": 0.5162,
      "step": 14874
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6054375171661377,
      "learning_rate": 1.1525707295792907e-05,
      "loss": 0.4747,
      "step": 14875
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5885639190673828,
      "learning_rate": 1.1524700800386675e-05,
      "loss": 0.4446,
      "step": 14876
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5760494470596313,
      "learning_rate": 1.1523694289166858e-05,
      "loss": 0.4702,
      "step": 14877
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6437519788742065,
      "learning_rate": 1.1522687762143893e-05,
      "loss": 0.4672,
      "step": 14878
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6819788217544556,
      "learning_rate": 1.1521681219328223e-05,
      "loss": 0.4713,
      "step": 14879
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5455060005187988,
      "learning_rate": 1.1520674660730287e-05,
      "loss": 0.4079,
      "step": 14880
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5140798091888428,
      "learning_rate": 1.1519668086360518e-05,
      "loss": 0.4925,
      "step": 14881
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4733353853225708,
      "learning_rate": 1.1518661496229363e-05,
      "loss": 0.4063,
      "step": 14882
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0237479209899902,
      "learning_rate": 1.151765489034726e-05,
      "loss": 0.4952,
      "step": 14883
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5034152269363403,
      "learning_rate": 1.151664826872465e-05,
      "loss": 0.4121,
      "step": 14884
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.54054594039917,
      "learning_rate": 1.1515641631371975e-05,
      "loss": 0.4665,
      "step": 14885
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5427844524383545,
      "learning_rate": 1.1514634978299669e-05,
      "loss": 0.4173,
      "step": 14886
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.501227855682373,
      "learning_rate": 1.1513628309518174e-05,
      "loss": 0.435,
      "step": 14887
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5157121419906616,
      "learning_rate": 1.1512621625037939e-05,
      "loss": 0.4807,
      "step": 14888
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5612719058990479,
      "learning_rate": 1.1511614924869392e-05,
      "loss": 0.4833,
      "step": 14889
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.580006718635559,
      "learning_rate": 1.1510608209022984e-05,
      "loss": 0.4334,
      "step": 14890
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5606026649475098,
      "learning_rate": 1.1509601477509156e-05,
      "loss": 0.444,
      "step": 14891
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6423512697219849,
      "learning_rate": 1.1508594730338345e-05,
      "loss": 0.4513,
      "step": 14892
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6947433948516846,
      "learning_rate": 1.1507587967520995e-05,
      "loss": 0.9211,
      "step": 14893
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5164246559143066,
      "learning_rate": 1.1506581189067544e-05,
      "loss": 1.0486,
      "step": 14894
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.648985505104065,
      "learning_rate": 1.1505574394988439e-05,
      "loss": 0.5154,
      "step": 14895
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5109394788742065,
      "learning_rate": 1.1504567585294122e-05,
      "loss": 0.3948,
      "step": 14896
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6279313564300537,
      "learning_rate": 1.1503560759995026e-05,
      "loss": 0.4784,
      "step": 14897
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5396454334259033,
      "learning_rate": 1.1502553919101611e-05,
      "loss": 0.5158,
      "step": 14898
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5866667032241821,
      "learning_rate": 1.15015470626243e-05,
      "loss": 0.4701,
      "step": 14899
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.427636742591858,
      "learning_rate": 1.150054019057355e-05,
      "loss": 0.3768,
      "step": 14900
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5150835514068604,
      "learning_rate": 1.1499533302959798e-05,
      "loss": 0.4467,
      "step": 14901
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.533813714981079,
      "learning_rate": 1.1498526399793488e-05,
      "loss": 0.4272,
      "step": 14902
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5240789651870728,
      "learning_rate": 1.1497519481085063e-05,
      "loss": 0.4377,
      "step": 14903
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5443867444992065,
      "learning_rate": 1.1496512546844964e-05,
      "loss": 0.4088,
      "step": 14904
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.665191888809204,
      "learning_rate": 1.1495505597083635e-05,
      "loss": 0.5006,
      "step": 14905
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.601415991783142,
      "learning_rate": 1.1494498631811531e-05,
      "loss": 0.4279,
      "step": 14906
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.495867133140564,
      "learning_rate": 1.1493491651039077e-05,
      "loss": 0.4357,
      "step": 14907
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7713557481765747,
      "learning_rate": 1.149248465477673e-05,
      "loss": 0.472,
      "step": 14908
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6587693691253662,
      "learning_rate": 1.149147764303493e-05,
      "loss": 0.4919,
      "step": 14909
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5739067792892456,
      "learning_rate": 1.1490470615824121e-05,
      "loss": 0.418,
      "step": 14910
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5962505340576172,
      "learning_rate": 1.1489463573154752e-05,
      "loss": 0.4631,
      "step": 14911
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5073940753936768,
      "learning_rate": 1.148845651503726e-05,
      "loss": 0.4058,
      "step": 14912
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.58720862865448,
      "learning_rate": 1.1487449441482092e-05,
      "loss": 0.4574,
      "step": 14913
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5109493732452393,
      "learning_rate": 1.14864423524997e-05,
      "loss": 0.448,
      "step": 14914
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6631897687911987,
      "learning_rate": 1.148543524810052e-05,
      "loss": 0.467,
      "step": 14915
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6023904085159302,
      "learning_rate": 1.1484428128295006e-05,
      "loss": 0.444,
      "step": 14916
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5620003938674927,
      "learning_rate": 1.1483420993093595e-05,
      "loss": 0.4466,
      "step": 14917
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5181140899658203,
      "learning_rate": 1.1482413842506736e-05,
      "loss": 0.4135,
      "step": 14918
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6606014966964722,
      "learning_rate": 1.1481406676544874e-05,
      "loss": 0.4891,
      "step": 14919
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6963785886764526,
      "learning_rate": 1.148039949521846e-05,
      "loss": 0.4289,
      "step": 14920
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5808359384536743,
      "learning_rate": 1.1479392298537933e-05,
      "loss": 0.4722,
      "step": 14921
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5619093179702759,
      "learning_rate": 1.1478385086513742e-05,
      "loss": 0.4295,
      "step": 14922
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6764637231826782,
      "learning_rate": 1.1477377859156333e-05,
      "loss": 0.5071,
      "step": 14923
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4973437786102295,
      "learning_rate": 1.1476370616476156e-05,
      "loss": 0.4209,
      "step": 14924
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6213796138763428,
      "learning_rate": 1.1475363358483651e-05,
      "loss": 0.5035,
      "step": 14925
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.517102837562561,
      "learning_rate": 1.147435608518927e-05,
      "loss": 0.4295,
      "step": 14926
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.632692813873291,
      "learning_rate": 1.147334879660346e-05,
      "loss": 0.4858,
      "step": 14927
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.526242733001709,
      "learning_rate": 1.147234149273667e-05,
      "loss": 0.4203,
      "step": 14928
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5537422895431519,
      "learning_rate": 1.147133417359934e-05,
      "loss": 0.4754,
      "step": 14929
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.607957363128662,
      "learning_rate": 1.1470326839201925e-05,
      "loss": 0.4638,
      "step": 14930
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5323002338409424,
      "learning_rate": 1.1469319489554863e-05,
      "loss": 0.4905,
      "step": 14931
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5651413202285767,
      "learning_rate": 1.1468312124668614e-05,
      "loss": 0.4381,
      "step": 14932
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6360687017440796,
      "learning_rate": 1.1467304744553618e-05,
      "loss": 0.5115,
      "step": 14933
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4819039106369019,
      "learning_rate": 1.1466297349220328e-05,
      "loss": 0.4193,
      "step": 14934
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.710752010345459,
      "learning_rate": 1.1465289938679186e-05,
      "loss": 0.4744,
      "step": 14935
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.554944634437561,
      "learning_rate": 1.1464282512940648e-05,
      "loss": 0.4373,
      "step": 14936
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5838054418563843,
      "learning_rate": 1.1463275072015159e-05,
      "loss": 0.4748,
      "step": 14937
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.475182294845581,
      "learning_rate": 1.1462267615913164e-05,
      "loss": 0.398,
      "step": 14938
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9630485773086548,
      "learning_rate": 1.1461260144645118e-05,
      "loss": 0.4935,
      "step": 14939
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5145537853240967,
      "learning_rate": 1.1460252658221465e-05,
      "loss": 0.4165,
      "step": 14940
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.610319972038269,
      "learning_rate": 1.145924515665266e-05,
      "loss": 0.5155,
      "step": 14941
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5233601331710815,
      "learning_rate": 1.1458237639949148e-05,
      "loss": 0.4537,
      "step": 14942
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5433727502822876,
      "learning_rate": 1.145723010812138e-05,
      "loss": 0.5001,
      "step": 14943
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5646799802780151,
      "learning_rate": 1.1456222561179806e-05,
      "loss": 0.4478,
      "step": 14944
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.504607081413269,
      "learning_rate": 1.1455214999134876e-05,
      "loss": 0.4562,
      "step": 14945
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5655810832977295,
      "learning_rate": 1.1454207421997037e-05,
      "loss": 0.4333,
      "step": 14946
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4802995920181274,
      "learning_rate": 1.1453199829776744e-05,
      "loss": 0.4311,
      "step": 14947
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6070936918258667,
      "learning_rate": 1.1452192222484442e-05,
      "loss": 0.4182,
      "step": 14948
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5329524278640747,
      "learning_rate": 1.1451184600130588e-05,
      "loss": 0.4768,
      "step": 14949
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.532390832901001,
      "learning_rate": 1.1450176962725627e-05,
      "loss": 0.4085,
      "step": 14950
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6805007457733154,
      "learning_rate": 1.1449169310280012e-05,
      "loss": 0.4999,
      "step": 14951
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5071072578430176,
      "learning_rate": 1.1448161642804196e-05,
      "loss": 0.4118,
      "step": 14952
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5606400966644287,
      "learning_rate": 1.1447153960308628e-05,
      "loss": 0.4866,
      "step": 14953
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5621988773345947,
      "learning_rate": 1.1446146262803757e-05,
      "loss": 0.4226,
      "step": 14954
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5690844058990479,
      "learning_rate": 1.1445138550300039e-05,
      "loss": 0.475,
      "step": 14955
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4786325693130493,
      "learning_rate": 1.144413082280792e-05,
      "loss": 0.4212,
      "step": 14956
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5362261533737183,
      "learning_rate": 1.1443123080337858e-05,
      "loss": 0.4371,
      "step": 14957
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5046299695968628,
      "learning_rate": 1.14421153229003e-05,
      "loss": 0.4158,
      "step": 14958
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4972021579742432,
      "learning_rate": 1.1441107550505702e-05,
      "loss": 0.4801,
      "step": 14959
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5771268606185913,
      "learning_rate": 1.1440099763164513e-05,
      "loss": 0.4244,
      "step": 14960
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5938222408294678,
      "learning_rate": 1.1439091960887188e-05,
      "loss": 0.4648,
      "step": 14961
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5677413940429688,
      "learning_rate": 1.1438084143684177e-05,
      "loss": 0.4382,
      "step": 14962
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.601320743560791,
      "learning_rate": 1.1437076311565934e-05,
      "loss": 0.4716,
      "step": 14963
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.506003975868225,
      "learning_rate": 1.1436068464542909e-05,
      "loss": 0.4126,
      "step": 14964
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.707062840461731,
      "learning_rate": 1.1435060602625564e-05,
      "loss": 0.5096,
      "step": 14965
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.1023967266082764,
      "learning_rate": 1.1434052725824338e-05,
      "loss": 0.399,
      "step": 14966
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5992785692214966,
      "learning_rate": 1.1433044834149696e-05,
      "loss": 0.4958,
      "step": 14967
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5785484313964844,
      "learning_rate": 1.143203692761209e-05,
      "loss": 0.434,
      "step": 14968
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5411514043807983,
      "learning_rate": 1.1431029006221967e-05,
      "loss": 0.4728,
      "step": 14969
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5595571994781494,
      "learning_rate": 1.1430021069989787e-05,
      "loss": 0.4468,
      "step": 14970
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5145593881607056,
      "learning_rate": 1.1429013118926002e-05,
      "loss": 0.5256,
      "step": 14971
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4756040573120117,
      "learning_rate": 1.1428005153041064e-05,
      "loss": 0.4043,
      "step": 14972
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6641067266464233,
      "learning_rate": 1.1426997172345434e-05,
      "loss": 0.5242,
      "step": 14973
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5014861822128296,
      "learning_rate": 1.1425989176849555e-05,
      "loss": 0.4518,
      "step": 14974
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5768027305603027,
      "learning_rate": 1.142498116656389e-05,
      "loss": 0.4628,
      "step": 14975
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5697804689407349,
      "learning_rate": 1.1423973141498892e-05,
      "loss": 0.4688,
      "step": 14976
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.70546293258667,
      "learning_rate": 1.1422965101665016e-05,
      "loss": 0.522,
      "step": 14977
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5305964946746826,
      "learning_rate": 1.1421957047072717e-05,
      "loss": 0.4241,
      "step": 14978
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6117874383926392,
      "learning_rate": 1.142094897773245e-05,
      "loss": 0.4602,
      "step": 14979
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5808067321777344,
      "learning_rate": 1.1419940893654669e-05,
      "loss": 0.4355,
      "step": 14980
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6349936723709106,
      "learning_rate": 1.1418932794849828e-05,
      "loss": 0.4921,
      "step": 14981
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5351933240890503,
      "learning_rate": 1.1417924681328385e-05,
      "loss": 0.426,
      "step": 14982
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6046814918518066,
      "learning_rate": 1.1416916553100801e-05,
      "loss": 0.4702,
      "step": 14983
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.6278223991394043,
      "learning_rate": 1.1415908410177521e-05,
      "loss": 0.4422,
      "step": 14984
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5169029235839844,
      "learning_rate": 1.141490025256901e-05,
      "loss": 0.4664,
      "step": 14985
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5450013875961304,
      "learning_rate": 1.141389208028572e-05,
      "loss": 0.4235,
      "step": 14986
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5171676874160767,
      "learning_rate": 1.1412883893338106e-05,
      "loss": 0.4574,
      "step": 14987
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5593267679214478,
      "learning_rate": 1.141187569173663e-05,
      "loss": 0.4513,
      "step": 14988
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6106257438659668,
      "learning_rate": 1.1410867475491744e-05,
      "loss": 0.4427,
      "step": 14989
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6074117422103882,
      "learning_rate": 1.1409859244613906e-05,
      "loss": 0.4594,
      "step": 14990
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5497024059295654,
      "learning_rate": 1.1408850999113574e-05,
      "loss": 0.4646,
      "step": 14991
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5627323389053345,
      "learning_rate": 1.1407842739001202e-05,
      "loss": 0.4386,
      "step": 14992
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5734949111938477,
      "learning_rate": 1.1406834464287253e-05,
      "loss": 0.4924,
      "step": 14993
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5968823432922363,
      "learning_rate": 1.1405826174982178e-05,
      "loss": 0.4866,
      "step": 14994
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6525825262069702,
      "learning_rate": 1.1404817871096438e-05,
      "loss": 0.5147,
      "step": 14995
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6758514642715454,
      "learning_rate": 1.1403809552640492e-05,
      "loss": 0.4262,
      "step": 14996
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.617781162261963,
      "learning_rate": 1.1402801219624796e-05,
      "loss": 0.4756,
      "step": 14997
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.560471534729004,
      "learning_rate": 1.1401792872059808e-05,
      "loss": 0.4456,
      "step": 14998
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5061904191970825,
      "learning_rate": 1.1400784509955985e-05,
      "loss": 0.5033,
      "step": 14999
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6081534624099731,
      "learning_rate": 1.1399776133323785e-05,
      "loss": 0.42,
      "step": 15000
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7448347806930542,
      "learning_rate": 1.1398767742173673e-05,
      "loss": 0.4662,
      "step": 15001
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6836040019989014,
      "learning_rate": 1.1397759336516099e-05,
      "loss": 0.4539,
      "step": 15002
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.549330234527588,
      "learning_rate": 1.1396750916361526e-05,
      "loss": 0.4363,
      "step": 15003
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6181246042251587,
      "learning_rate": 1.1395742481720414e-05,
      "loss": 0.4994,
      "step": 15004
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5877858400344849,
      "learning_rate": 1.1394734032603219e-05,
      "loss": 0.4451,
      "step": 15005
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6990290880203247,
      "learning_rate": 1.1393725569020404e-05,
      "loss": 0.4208,
      "step": 15006
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6471587419509888,
      "learning_rate": 1.1392717090982423e-05,
      "loss": 0.459,
      "step": 15007
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.56844961643219,
      "learning_rate": 1.139170859849974e-05,
      "loss": 0.441,
      "step": 15008
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5053809881210327,
      "learning_rate": 1.1390700091582816e-05,
      "loss": 0.5377,
      "step": 15009
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6228500604629517,
      "learning_rate": 1.1389691570242103e-05,
      "loss": 0.4398,
      "step": 15010
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7811779975891113,
      "learning_rate": 1.1388683034488073e-05,
      "loss": 1.0415,
      "step": 15011
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5995442867279053,
      "learning_rate": 1.1387674484331173e-05,
      "loss": 1.0331,
      "step": 15012
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5799018144607544,
      "learning_rate": 1.1386665919781871e-05,
      "loss": 0.5027,
      "step": 15013
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7937536239624023,
      "learning_rate": 1.138565734085063e-05,
      "loss": 0.4116,
      "step": 15014
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6366406679153442,
      "learning_rate": 1.1384648747547902e-05,
      "loss": 0.4886,
      "step": 15015
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5855703353881836,
      "learning_rate": 1.1383640139884156e-05,
      "loss": 0.4546,
      "step": 15016
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6031808853149414,
      "learning_rate": 1.1382631517869847e-05,
      "loss": 0.4769,
      "step": 15017
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5691794157028198,
      "learning_rate": 1.1381622881515435e-05,
      "loss": 0.4765,
      "step": 15018
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5288206338882446,
      "learning_rate": 1.138061423083139e-05,
      "loss": 0.4558,
      "step": 15019
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6131359338760376,
      "learning_rate": 1.1379605565828165e-05,
      "loss": 0.4718,
      "step": 15020
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5763616561889648,
      "learning_rate": 1.1378596886516225e-05,
      "loss": 0.5074,
      "step": 15021
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5036277770996094,
      "learning_rate": 1.1377588192906031e-05,
      "loss": 0.4047,
      "step": 15022
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.035783290863037,
      "learning_rate": 1.1376579485008043e-05,
      "loss": 0.493,
      "step": 15023
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6772857904434204,
      "learning_rate": 1.1375570762832725e-05,
      "loss": 0.4669,
      "step": 15024
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6633983850479126,
      "learning_rate": 1.137456202639054e-05,
      "loss": 0.4491,
      "step": 15025
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.494416356086731,
      "learning_rate": 1.1373553275691947e-05,
      "loss": 0.3966,
      "step": 15026
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6452525854110718,
      "learning_rate": 1.1372544510747412e-05,
      "loss": 0.4087,
      "step": 15027
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4963818788528442,
      "learning_rate": 1.1371535731567393e-05,
      "loss": 0.4316,
      "step": 15028
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5403658151626587,
      "learning_rate": 1.137052693816236e-05,
      "loss": 0.4681,
      "step": 15029
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.503732681274414,
      "learning_rate": 1.1369518130542766e-05,
      "loss": 0.4099,
      "step": 15030
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6425176858901978,
      "learning_rate": 1.136850930871908e-05,
      "loss": 0.5246,
      "step": 15031
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5337938070297241,
      "learning_rate": 1.1367500472701765e-05,
      "loss": 0.4206,
      "step": 15032
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5532523393630981,
      "learning_rate": 1.1366491622501283e-05,
      "loss": 0.4603,
      "step": 15033
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5000101327896118,
      "learning_rate": 1.1365482758128099e-05,
      "loss": 0.4061,
      "step": 15034
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6452000141143799,
      "learning_rate": 1.1364473879592674e-05,
      "loss": 0.4956,
      "step": 15035
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4474194049835205,
      "learning_rate": 1.1363464986905472e-05,
      "loss": 0.4252,
      "step": 15036
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.511240839958191,
      "learning_rate": 1.1362456080076963e-05,
      "loss": 0.4407,
      "step": 15037
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6258642673492432,
      "learning_rate": 1.13614471591176e-05,
      "loss": 0.4619,
      "step": 15038
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.037933349609375,
      "learning_rate": 1.1360438224037856e-05,
      "loss": 0.4736,
      "step": 15039
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6320639848709106,
      "learning_rate": 1.1359429274848188e-05,
      "loss": 0.4117,
      "step": 15040
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8484482765197754,
      "learning_rate": 1.1358420311559068e-05,
      "loss": 0.5299,
      "step": 15041
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6078511476516724,
      "learning_rate": 1.1357411334180958e-05,
      "loss": 0.4392,
      "step": 15042
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7179139852523804,
      "learning_rate": 1.1356402342724322e-05,
      "loss": 0.5281,
      "step": 15043
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4550427198410034,
      "learning_rate": 1.135539333719962e-05,
      "loss": 0.4177,
      "step": 15044
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7909166812896729,
      "learning_rate": 1.1354384317617328e-05,
      "loss": 1.036,
      "step": 15045
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4600952863693237,
      "learning_rate": 1.13533752839879e-05,
      "loss": 0.928,
      "step": 15046
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6186232566833496,
      "learning_rate": 1.1352366236321808e-05,
      "loss": 0.4582,
      "step": 15047
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.549196481704712,
      "learning_rate": 1.1351357174629517e-05,
      "loss": 0.4006,
      "step": 15048
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6238702535629272,
      "learning_rate": 1.1350348098921488e-05,
      "loss": 0.4917,
      "step": 15049
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6023712158203125,
      "learning_rate": 1.1349339009208193e-05,
      "loss": 0.4709,
      "step": 15050
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5143951177597046,
      "learning_rate": 1.1348329905500093e-05,
      "loss": 0.4689,
      "step": 15051
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4531991481781006,
      "learning_rate": 1.1347320787807657e-05,
      "loss": 0.4006,
      "step": 15052
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4961087703704834,
      "learning_rate": 1.1346311656141349e-05,
      "loss": 0.4481,
      "step": 15053
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5589648485183716,
      "learning_rate": 1.1345302510511635e-05,
      "loss": 0.4166,
      "step": 15054
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4893200397491455,
      "learning_rate": 1.1344293350928986e-05,
      "loss": 0.44,
      "step": 15055
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5322145223617554,
      "learning_rate": 1.1343284177403861e-05,
      "loss": 0.4303,
      "step": 15056
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6146385669708252,
      "learning_rate": 1.1342274989946733e-05,
      "loss": 0.4946,
      "step": 15057
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.544589638710022,
      "learning_rate": 1.134126578856807e-05,
      "loss": 0.4334,
      "step": 15058
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6115999221801758,
      "learning_rate": 1.1340256573278333e-05,
      "loss": 0.4676,
      "step": 15059
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5397197008132935,
      "learning_rate": 1.1339247344087992e-05,
      "loss": 0.4356,
      "step": 15060
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5828790664672852,
      "learning_rate": 1.1338238101007515e-05,
      "loss": 0.4732,
      "step": 15061
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.711403727531433,
      "learning_rate": 1.1337228844047368e-05,
      "loss": 0.5181,
      "step": 15062
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9812062978744507,
      "learning_rate": 1.1336219573218024e-05,
      "loss": 1.0741,
      "step": 15063
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.626451015472412,
      "learning_rate": 1.1335210288529939e-05,
      "loss": 1.0332,
      "step": 15064
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7958561182022095,
      "learning_rate": 1.1334200989993593e-05,
      "loss": 0.9544,
      "step": 15065
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4506903886795044,
      "learning_rate": 1.1333191677619448e-05,
      "loss": 1.0381,
      "step": 15066
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9243375062942505,
      "learning_rate": 1.1332182351417975e-05,
      "loss": 0.4837,
      "step": 15067
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7650634050369263,
      "learning_rate": 1.1331173011399639e-05,
      "loss": 0.4336,
      "step": 15068
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6682223081588745,
      "learning_rate": 1.133016365757491e-05,
      "loss": 0.5069,
      "step": 15069
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5604777336120605,
      "learning_rate": 1.1329154289954258e-05,
      "loss": 0.4523,
      "step": 15070
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5977534055709839,
      "learning_rate": 1.132814490854815e-05,
      "loss": 0.4619,
      "step": 15071
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6088967323303223,
      "learning_rate": 1.1327135513367057e-05,
      "loss": 0.4084,
      "step": 15072
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6119087934494019,
      "learning_rate": 1.1326126104421447e-05,
      "loss": 0.5655,
      "step": 15073
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.661124587059021,
      "learning_rate": 1.1325116681721784e-05,
      "loss": 0.4242,
      "step": 15074
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5669348239898682,
      "learning_rate": 1.1324107245278544e-05,
      "loss": 0.4334,
      "step": 15075
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5423362255096436,
      "learning_rate": 1.1323097795102196e-05,
      "loss": 0.4233,
      "step": 15076
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6074109077453613,
      "learning_rate": 1.1322088331203206e-05,
      "loss": 0.4945,
      "step": 15077
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5299625396728516,
      "learning_rate": 1.1321078853592049e-05,
      "loss": 0.4146,
      "step": 15078
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6007646322250366,
      "learning_rate": 1.1320069362279192e-05,
      "loss": 0.4809,
      "step": 15079
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.526436686515808,
      "learning_rate": 1.1319059857275101e-05,
      "loss": 0.4176,
      "step": 15080
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.514603614807129,
      "learning_rate": 1.1318050338590254e-05,
      "loss": 0.506,
      "step": 15081
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.608847737312317,
      "learning_rate": 1.1317040806235114e-05,
      "loss": 0.4427,
      "step": 15082
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6445633172988892,
      "learning_rate": 1.1316031260220156e-05,
      "loss": 0.4442,
      "step": 15083
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4998129606246948,
      "learning_rate": 1.1315021700555848e-05,
      "loss": 0.427,
      "step": 15084
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5493206977844238,
      "learning_rate": 1.1314012127252665e-05,
      "loss": 0.4328,
      "step": 15085
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6988706588745117,
      "learning_rate": 1.1313002540321074e-05,
      "loss": 0.4441,
      "step": 15086
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.512986660003662,
      "learning_rate": 1.1311992939771548e-05,
      "loss": 0.4815,
      "step": 15087
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4740523099899292,
      "learning_rate": 1.1310983325614555e-05,
      "loss": 0.416,
      "step": 15088
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5592914819717407,
      "learning_rate": 1.130997369786057e-05,
      "loss": 0.4862,
      "step": 15089
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.546330213546753,
      "learning_rate": 1.130896405652006e-05,
      "loss": 0.4535,
      "step": 15090
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5739034414291382,
      "learning_rate": 1.1307954401603504e-05,
      "loss": 0.4414,
      "step": 15091
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5254982709884644,
      "learning_rate": 1.1306944733121365e-05,
      "loss": 0.4494,
      "step": 15092
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.592980146408081,
      "learning_rate": 1.1305935051084122e-05,
      "loss": 0.4705,
      "step": 15093
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4792505502700806,
      "learning_rate": 1.1304925355502245e-05,
      "loss": 0.4386,
      "step": 15094
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5014448165893555,
      "learning_rate": 1.1303915646386205e-05,
      "loss": 0.4752,
      "step": 15095
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.667467474937439,
      "learning_rate": 1.1302905923746475e-05,
      "loss": 0.4654,
      "step": 15096
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5478425025939941,
      "learning_rate": 1.1301896187593524e-05,
      "loss": 0.4912,
      "step": 15097
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.537108302116394,
      "learning_rate": 1.1300886437937828e-05,
      "loss": 0.4117,
      "step": 15098
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5540170669555664,
      "learning_rate": 1.1299876674789864e-05,
      "loss": 0.522,
      "step": 15099
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5275331735610962,
      "learning_rate": 1.1298866898160096e-05,
      "loss": 0.4192,
      "step": 15100
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6566359996795654,
      "learning_rate": 1.1297857108059003e-05,
      "loss": 0.5244,
      "step": 15101
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5404654741287231,
      "learning_rate": 1.1296847304497056e-05,
      "loss": 0.4059,
      "step": 15102
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6280012130737305,
      "learning_rate": 1.129583748748473e-05,
      "loss": 0.5165,
      "step": 15103
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5673787593841553,
      "learning_rate": 1.1294827657032494e-05,
      "loss": 0.4353,
      "step": 15104
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6477395296096802,
      "learning_rate": 1.129381781315083e-05,
      "loss": 0.536,
      "step": 15105
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5719801187515259,
      "learning_rate": 1.1292807955850198e-05,
      "loss": 0.4324,
      "step": 15106
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.506110429763794,
      "learning_rate": 1.1291798085141087e-05,
      "loss": 0.44,
      "step": 15107
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.533146858215332,
      "learning_rate": 1.1290788201033962e-05,
      "loss": 0.4215,
      "step": 15108
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5224275588989258,
      "learning_rate": 1.1289778303539303e-05,
      "loss": 0.4533,
      "step": 15109
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5290517807006836,
      "learning_rate": 1.1288768392667575e-05,
      "loss": 0.4337,
      "step": 15110
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.487545371055603,
      "learning_rate": 1.128775846842926e-05,
      "loss": 0.4595,
      "step": 15111
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5939891338348389,
      "learning_rate": 1.1286748530834831e-05,
      "loss": 0.4396,
      "step": 15112
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5453999042510986,
      "learning_rate": 1.1285738579894761e-05,
      "loss": 0.4569,
      "step": 15113
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.517132043838501,
      "learning_rate": 1.1284728615619527e-05,
      "loss": 0.3977,
      "step": 15114
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.562888741493225,
      "learning_rate": 1.1283718638019603e-05,
      "loss": 0.4384,
      "step": 15115
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5961424112319946,
      "learning_rate": 1.128270864710546e-05,
      "loss": 0.4136,
      "step": 15116
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6274757385253906,
      "learning_rate": 1.1281698642887584e-05,
      "loss": 0.4565,
      "step": 15117
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5717458724975586,
      "learning_rate": 1.1280688625376437e-05,
      "loss": 0.4105,
      "step": 15118
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5643693208694458,
      "learning_rate": 1.1279678594582504e-05,
      "loss": 0.4778,
      "step": 15119
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5567413568496704,
      "learning_rate": 1.1278668550516259e-05,
      "loss": 0.4273,
      "step": 15120
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6075701713562012,
      "learning_rate": 1.1277658493188175e-05,
      "loss": 0.4916,
      "step": 15121
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5593551397323608,
      "learning_rate": 1.127664842260873e-05,
      "loss": 0.4295,
      "step": 15122
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5276776552200317,
      "learning_rate": 1.12756383387884e-05,
      "loss": 0.4498,
      "step": 15123
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5625264644622803,
      "learning_rate": 1.1274628241737657e-05,
      "loss": 0.4492,
      "step": 15124
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.569740653038025,
      "learning_rate": 1.1273618131466985e-05,
      "loss": 0.5069,
      "step": 15125
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.531605839729309,
      "learning_rate": 1.1272608007986855e-05,
      "loss": 0.4416,
      "step": 15126
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5143954753875732,
      "learning_rate": 1.1271597871307745e-05,
      "loss": 0.4763,
      "step": 15127
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5167521238327026,
      "learning_rate": 1.127058772144013e-05,
      "loss": 0.4399,
      "step": 15128
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5555206537246704,
      "learning_rate": 1.1269577558394488e-05,
      "loss": 0.4982,
      "step": 15129
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5613778829574585,
      "learning_rate": 1.1268567382181302e-05,
      "loss": 0.4312,
      "step": 15130
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6741360425949097,
      "learning_rate": 1.1267557192811037e-05,
      "loss": 0.4855,
      "step": 15131
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.578486442565918,
      "learning_rate": 1.1266546990294183e-05,
      "loss": 0.4201,
      "step": 15132
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.475461483001709,
      "learning_rate": 1.1265536774641208e-05,
      "loss": 0.4459,
      "step": 15133
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4993408918380737,
      "learning_rate": 1.126452654586259e-05,
      "loss": 0.4179,
      "step": 15134
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.757848858833313,
      "learning_rate": 1.1263516303968817e-05,
      "loss": 0.545,
      "step": 15135
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5647484064102173,
      "learning_rate": 1.1262506048970353e-05,
      "loss": 0.4449,
      "step": 15136
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5577884912490845,
      "learning_rate": 1.1261495780877687e-05,
      "loss": 0.5081,
      "step": 15137
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.280017852783203,
      "learning_rate": 1.126048549970129e-05,
      "loss": 0.4256,
      "step": 15138
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5845810174942017,
      "learning_rate": 1.1259475205451644e-05,
      "loss": 0.5307,
      "step": 15139
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.494413137435913,
      "learning_rate": 1.1258464898139224e-05,
      "loss": 0.394,
      "step": 15140
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6407577991485596,
      "learning_rate": 1.1257454577774513e-05,
      "loss": 0.4925,
      "step": 15141
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.605394721031189,
      "learning_rate": 1.1256444244367985e-05,
      "loss": 0.4163,
      "step": 15142
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.592612862586975,
      "learning_rate": 1.1255433897930126e-05,
      "loss": 0.4607,
      "step": 15143
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5142064094543457,
      "learning_rate": 1.1254423538471404e-05,
      "loss": 0.4144,
      "step": 15144
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5258575677871704,
      "learning_rate": 1.1253413166002307e-05,
      "loss": 0.4757,
      "step": 15145
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5373551845550537,
      "learning_rate": 1.1252402780533308e-05,
      "loss": 0.4122,
      "step": 15146
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5628801584243774,
      "learning_rate": 1.1251392382074892e-05,
      "loss": 0.4958,
      "step": 15147
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5367599725723267,
      "learning_rate": 1.125038197063754e-05,
      "loss": 0.4069,
      "step": 15148
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.660301923751831,
      "learning_rate": 1.1249371546231722e-05,
      "loss": 0.5037,
      "step": 15149
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5354845523834229,
      "learning_rate": 1.1248361108867924e-05,
      "loss": 0.4243,
      "step": 15150
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6598494052886963,
      "learning_rate": 1.1247350658556626e-05,
      "loss": 0.4784,
      "step": 15151
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.593315601348877,
      "learning_rate": 1.1246340195308305e-05,
      "loss": 0.482,
      "step": 15152
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5884487628936768,
      "learning_rate": 1.1245329719133447e-05,
      "loss": 0.4366,
      "step": 15153
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5483571290969849,
      "learning_rate": 1.1244319230042524e-05,
      "loss": 0.4498,
      "step": 15154
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8978326320648193,
      "learning_rate": 1.1243308728046023e-05,
      "loss": 0.4608,
      "step": 15155
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6500329971313477,
      "learning_rate": 1.1242298213154424e-05,
      "loss": 0.4472,
      "step": 15156
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5900474786758423,
      "learning_rate": 1.1241287685378203e-05,
      "loss": 0.4961,
      "step": 15157
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.575239896774292,
      "learning_rate": 1.1240277144727848e-05,
      "loss": 0.4491,
      "step": 15158
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.601394534111023,
      "learning_rate": 1.1239266591213832e-05,
      "loss": 0.4697,
      "step": 15159
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.512509822845459,
      "learning_rate": 1.123825602484664e-05,
      "loss": 0.4248,
      "step": 15160
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6022045612335205,
      "learning_rate": 1.1237245445636755e-05,
      "loss": 0.4986,
      "step": 15161
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.436699390411377,
      "learning_rate": 1.1236234853594651e-05,
      "loss": 0.4019,
      "step": 15162
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6051671504974365,
      "learning_rate": 1.1235224248730821e-05,
      "loss": 0.5079,
      "step": 15163
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5292093753814697,
      "learning_rate": 1.1234213631055736e-05,
      "loss": 0.417,
      "step": 15164
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6760084629058838,
      "learning_rate": 1.1233203000579887e-05,
      "loss": 0.4922,
      "step": 15165
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4804569482803345,
      "learning_rate": 1.1232192357313748e-05,
      "loss": 0.4132,
      "step": 15166
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5217565298080444,
      "learning_rate": 1.1231181701267801e-05,
      "loss": 0.424,
      "step": 15167
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6347771883010864,
      "learning_rate": 1.1230171032452535e-05,
      "loss": 0.464,
      "step": 15168
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.080759286880493,
      "learning_rate": 1.1229160350878428e-05,
      "loss": 1.1071,
      "step": 15169
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.767382264137268,
      "learning_rate": 1.122814965655596e-05,
      "loss": 1.0635,
      "step": 15170
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.667665958404541,
      "learning_rate": 1.122713894949562e-05,
      "loss": 0.4962,
      "step": 15171
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6126823425292969,
      "learning_rate": 1.1226128229707882e-05,
      "loss": 0.4271,
      "step": 15172
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.727620244026184,
      "learning_rate": 1.1225117497203238e-05,
      "loss": 0.457,
      "step": 15173
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6044198274612427,
      "learning_rate": 1.1224106751992164e-05,
      "loss": 0.4022,
      "step": 15174
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5668994188308716,
      "learning_rate": 1.1223095994085145e-05,
      "loss": 0.49,
      "step": 15175
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4769428968429565,
      "learning_rate": 1.1222085223492667e-05,
      "loss": 0.3977,
      "step": 15176
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5625497102737427,
      "learning_rate": 1.1221074440225209e-05,
      "loss": 0.4616,
      "step": 15177
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5413093566894531,
      "learning_rate": 1.1220063644293258e-05,
      "loss": 0.418,
      "step": 15178
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.622431993484497,
      "learning_rate": 1.1219052835707293e-05,
      "loss": 0.5144,
      "step": 15179
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5113216638565063,
      "learning_rate": 1.1218042014477806e-05,
      "loss": 0.4129,
      "step": 15180
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6459600925445557,
      "learning_rate": 1.1217031180615272e-05,
      "loss": 0.5942,
      "step": 15181
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5165520906448364,
      "learning_rate": 1.121602033413018e-05,
      "loss": 0.414,
      "step": 15182
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5028975009918213,
      "learning_rate": 1.1215009475033009e-05,
      "loss": 0.4523,
      "step": 15183
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5142226219177246,
      "learning_rate": 1.1213998603334253e-05,
      "loss": 0.4461,
      "step": 15184
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5866303443908691,
      "learning_rate": 1.1212987719044384e-05,
      "loss": 0.499,
      "step": 15185
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7126058340072632,
      "learning_rate": 1.1211976822173897e-05,
      "loss": 0.4474,
      "step": 15186
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5095875263214111,
      "learning_rate": 1.1210965912733269e-05,
      "loss": 0.4567,
      "step": 15187
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5376313924789429,
      "learning_rate": 1.1209954990732992e-05,
      "loss": 0.4184,
      "step": 15188
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9846336841583252,
      "learning_rate": 1.1208944056183544e-05,
      "loss": 0.4636,
      "step": 15189
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5411193370819092,
      "learning_rate": 1.1207933109095414e-05,
      "loss": 0.4315,
      "step": 15190
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5395952463150024,
      "learning_rate": 1.1206922149479087e-05,
      "loss": 0.469,
      "step": 15191
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5957962274551392,
      "learning_rate": 1.1205911177345046e-05,
      "loss": 0.4513,
      "step": 15192
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5216275453567505,
      "learning_rate": 1.1204900192703777e-05,
      "loss": 0.4525,
      "step": 15193
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5161614418029785,
      "learning_rate": 1.1203889195565768e-05,
      "loss": 0.428,
      "step": 15194
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6958727836608887,
      "learning_rate": 1.1202878185941502e-05,
      "loss": 0.4832,
      "step": 15195
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5655686855316162,
      "learning_rate": 1.1201867163841464e-05,
      "loss": 0.4117,
      "step": 15196
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6332060098648071,
      "learning_rate": 1.1200856129276145e-05,
      "loss": 0.5069,
      "step": 15197
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6385153532028198,
      "learning_rate": 1.1199845082256027e-05,
      "loss": 0.4356,
      "step": 15198
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.503817081451416,
      "learning_rate": 1.1198834022791597e-05,
      "loss": 0.4879,
      "step": 15199
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.558858871459961,
      "learning_rate": 1.119782295089334e-05,
      "loss": 0.4124,
      "step": 15200
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.534019112586975,
      "learning_rate": 1.1196811866571742e-05,
      "loss": 0.4407,
      "step": 15201
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.476204514503479,
      "learning_rate": 1.1195800769837296e-05,
      "loss": 0.413,
      "step": 15202
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6041523218154907,
      "learning_rate": 1.1194789660700478e-05,
      "loss": 0.4633,
      "step": 15203
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5303966999053955,
      "learning_rate": 1.1193778539171785e-05,
      "loss": 0.4092,
      "step": 15204
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5113853216171265,
      "learning_rate": 1.1192767405261696e-05,
      "loss": 0.4731,
      "step": 15205
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5301868915557861,
      "learning_rate": 1.1191756258980705e-05,
      "loss": 0.4369,
      "step": 15206
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5345842838287354,
      "learning_rate": 1.1190745100339293e-05,
      "loss": 0.4296,
      "step": 15207
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5462448596954346,
      "learning_rate": 1.1189733929347952e-05,
      "loss": 0.4297,
      "step": 15208
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.482351303100586,
      "learning_rate": 1.1188722746017169e-05,
      "loss": 0.4332,
      "step": 15209
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5112690925598145,
      "learning_rate": 1.1187711550357428e-05,
      "loss": 0.4175,
      "step": 15210
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6066439151763916,
      "learning_rate": 1.1186700342379217e-05,
      "loss": 0.4792,
      "step": 15211
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5012562274932861,
      "learning_rate": 1.118568912209303e-05,
      "loss": 0.4134,
      "step": 15212
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5407615900039673,
      "learning_rate": 1.118467788950935e-05,
      "loss": 0.5084,
      "step": 15213
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.487277626991272,
      "learning_rate": 1.1183666644638664e-05,
      "loss": 0.4194,
      "step": 15214
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.3466756343841553,
      "learning_rate": 1.1182655387491463e-05,
      "loss": 0.4576,
      "step": 15215
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.602294683456421,
      "learning_rate": 1.1181644118078236e-05,
      "loss": 0.4315,
      "step": 15216
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5418952703475952,
      "learning_rate": 1.1180632836409469e-05,
      "loss": 0.4895,
      "step": 15217
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.558221697807312,
      "learning_rate": 1.117962154249565e-05,
      "loss": 0.4223,
      "step": 15218
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.540426254272461,
      "learning_rate": 1.1178610236347269e-05,
      "loss": 0.4313,
      "step": 15219
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4849780797958374,
      "learning_rate": 1.1177598917974822e-05,
      "loss": 0.4236,
      "step": 15220
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6121865510940552,
      "learning_rate": 1.1176587587388784e-05,
      "loss": 0.5013,
      "step": 15221
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5487884283065796,
      "learning_rate": 1.1175576244599654e-05,
      "loss": 0.4183,
      "step": 15222
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6318522691726685,
      "learning_rate": 1.1174564889617916e-05,
      "loss": 0.4525,
      "step": 15223
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5560131072998047,
      "learning_rate": 1.1173553522454063e-05,
      "loss": 0.4275,
      "step": 15224
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6227269172668457,
      "learning_rate": 1.1172542143118583e-05,
      "loss": 0.4782,
      "step": 15225
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5797202587127686,
      "learning_rate": 1.1171530751621969e-05,
      "loss": 0.4623,
      "step": 15226
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6485896110534668,
      "learning_rate": 1.1170519347974704e-05,
      "loss": 0.4992,
      "step": 15227
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5650166273117065,
      "learning_rate": 1.1169507932187285e-05,
      "loss": 0.4036,
      "step": 15228
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5164071321487427,
      "learning_rate": 1.1168496504270195e-05,
      "loss": 0.469,
      "step": 15229
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5223381519317627,
      "learning_rate": 1.1167485064233932e-05,
      "loss": 0.4526,
      "step": 15230
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5059833526611328,
      "learning_rate": 1.1166473612088978e-05,
      "loss": 0.4497,
      "step": 15231
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4836101531982422,
      "learning_rate": 1.1165462147845828e-05,
      "loss": 0.418,
      "step": 15232
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7784422636032104,
      "learning_rate": 1.1164450671514972e-05,
      "loss": 0.5404,
      "step": 15233
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6404075622558594,
      "learning_rate": 1.1163439183106904e-05,
      "loss": 0.4142,
      "step": 15234
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.600290060043335,
      "learning_rate": 1.1162427682632106e-05,
      "loss": 0.4467,
      "step": 15235
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5664429664611816,
      "learning_rate": 1.1161416170101079e-05,
      "loss": 0.4213,
      "step": 15236
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5312050580978394,
      "learning_rate": 1.1160404645524304e-05,
      "loss": 0.4957,
      "step": 15237
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6893430948257446,
      "learning_rate": 1.1159393108912282e-05,
      "loss": 0.4443,
      "step": 15238
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5964218378067017,
      "learning_rate": 1.1158381560275496e-05,
      "loss": 0.4759,
      "step": 15239
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5928977727890015,
      "learning_rate": 1.1157369999624441e-05,
      "loss": 0.4481,
      "step": 15240
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6092352867126465,
      "learning_rate": 1.115635842696961e-05,
      "loss": 0.4687,
      "step": 15241
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6588884592056274,
      "learning_rate": 1.1155346842321493e-05,
      "loss": 0.4051,
      "step": 15242
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.612891674041748,
      "learning_rate": 1.1154335245690582e-05,
      "loss": 0.5054,
      "step": 15243
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5249052047729492,
      "learning_rate": 1.1153323637087368e-05,
      "loss": 0.4195,
      "step": 15244
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6117023229599,
      "learning_rate": 1.1152312016522344e-05,
      "loss": 0.4629,
      "step": 15245
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4346776008605957,
      "learning_rate": 1.1151300384006003e-05,
      "loss": 0.4056,
      "step": 15246
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6457492113113403,
      "learning_rate": 1.1150288739548831e-05,
      "loss": 0.5299,
      "step": 15247
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6096774339675903,
      "learning_rate": 1.1149277083161331e-05,
      "loss": 0.4194,
      "step": 15248
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8584613800048828,
      "learning_rate": 1.1148265414853987e-05,
      "loss": 0.9925,
      "step": 15249
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6229628324508667,
      "learning_rate": 1.1147253734637295e-05,
      "loss": 1.1095,
      "step": 15250
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6000336408615112,
      "learning_rate": 1.1146242042521747e-05,
      "loss": 0.5066,
      "step": 15251
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5821058750152588,
      "learning_rate": 1.1145230338517835e-05,
      "loss": 0.4484,
      "step": 15252
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5731865167617798,
      "learning_rate": 1.1144218622636056e-05,
      "loss": 0.484,
      "step": 15253
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5719486474990845,
      "learning_rate": 1.11432068948869e-05,
      "loss": 0.4228,
      "step": 15254
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6299718618392944,
      "learning_rate": 1.1142195155280858e-05,
      "loss": 0.4898,
      "step": 15255
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6181153059005737,
      "learning_rate": 1.1141183403828429e-05,
      "loss": 0.4305,
      "step": 15256
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8040944337844849,
      "learning_rate": 1.1140171640540099e-05,
      "loss": 0.4525,
      "step": 15257
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6442430019378662,
      "learning_rate": 1.1139159865426367e-05,
      "loss": 0.4523,
      "step": 15258
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.540864109992981,
      "learning_rate": 1.1138148078497729e-05,
      "loss": 0.4374,
      "step": 15259
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5788551568984985,
      "learning_rate": 1.1137136279764671e-05,
      "loss": 0.4373,
      "step": 15260
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.463711142539978,
      "learning_rate": 1.1136124469237695e-05,
      "loss": 0.4403,
      "step": 15261
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5645618438720703,
      "learning_rate": 1.1135112646927289e-05,
      "loss": 0.4512,
      "step": 15262
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5028101205825806,
      "learning_rate": 1.113410081284395e-05,
      "loss": 0.4802,
      "step": 15263
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5404765605926514,
      "learning_rate": 1.1133088966998173e-05,
      "loss": 0.4404,
      "step": 15264
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5175492763519287,
      "learning_rate": 1.113207710940045e-05,
      "loss": 0.4212,
      "step": 15265
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5550360679626465,
      "learning_rate": 1.113106524006128e-05,
      "loss": 0.459,
      "step": 15266
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7356362342834473,
      "learning_rate": 1.1130053358991154e-05,
      "loss": 0.5031,
      "step": 15267
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.502366542816162,
      "learning_rate": 1.1129041466200565e-05,
      "loss": 0.4018,
      "step": 15268
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5586576461791992,
      "learning_rate": 1.1128029561700013e-05,
      "loss": 0.4416,
      "step": 15269
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.550423502922058,
      "learning_rate": 1.1127017645499989e-05,
      "loss": 0.4313,
      "step": 15270
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5519814491271973,
      "learning_rate": 1.1126005717610992e-05,
      "loss": 0.4689,
      "step": 15271
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5668679475784302,
      "learning_rate": 1.1124993778043513e-05,
      "loss": 0.4026,
      "step": 15272
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.568274974822998,
      "learning_rate": 1.112398182680805e-05,
      "loss": 0.4354,
      "step": 15273
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6166927814483643,
      "learning_rate": 1.11229698639151e-05,
      "loss": 0.4323,
      "step": 15274
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6294454336166382,
      "learning_rate": 1.1121957889375154e-05,
      "loss": 0.469,
      "step": 15275
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5683969259262085,
      "learning_rate": 1.1120945903198713e-05,
      "loss": 0.4429,
      "step": 15276
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.664092779159546,
      "learning_rate": 1.1119933905396269e-05,
      "loss": 0.4816,
      "step": 15277
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.510196566581726,
      "learning_rate": 1.1118921895978318e-05,
      "loss": 0.3982,
      "step": 15278
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.629683256149292,
      "learning_rate": 1.1117909874955362e-05,
      "loss": 0.4759,
      "step": 15279
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.631425380706787,
      "learning_rate": 1.111689784233789e-05,
      "loss": 0.513,
      "step": 15280
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6150298118591309,
      "learning_rate": 1.1115885798136404e-05,
      "loss": 0.5175,
      "step": 15281
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5842934846878052,
      "learning_rate": 1.1114873742361393e-05,
      "loss": 0.4197,
      "step": 15282
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5764161348342896,
      "learning_rate": 1.1113861675023362e-05,
      "loss": 0.479,
      "step": 15283
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4882748126983643,
      "learning_rate": 1.1112849596132803e-05,
      "loss": 0.4167,
      "step": 15284
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6417267322540283,
      "learning_rate": 1.1111837505700215e-05,
      "loss": 0.4989,
      "step": 15285
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5898691415786743,
      "learning_rate": 1.1110825403736095e-05,
      "loss": 0.4439,
      "step": 15286
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5867760181427002,
      "learning_rate": 1.1109813290250937e-05,
      "loss": 0.4956,
      "step": 15287
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6196578741073608,
      "learning_rate": 1.1108801165255243e-05,
      "loss": 0.4334,
      "step": 15288
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6711421012878418,
      "learning_rate": 1.1107789028759505e-05,
      "loss": 0.4767,
      "step": 15289
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.456679344177246,
      "learning_rate": 1.1106776880774226e-05,
      "loss": 0.4034,
      "step": 15290
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6184121370315552,
      "learning_rate": 1.11057647213099e-05,
      "loss": 0.4929,
      "step": 15291
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.56349778175354,
      "learning_rate": 1.1104752550377026e-05,
      "loss": 0.4361,
      "step": 15292
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4650033712387085,
      "learning_rate": 1.11037403679861e-05,
      "loss": 0.4832,
      "step": 15293
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.477113962173462,
      "learning_rate": 1.110272817414762e-05,
      "loss": 0.4685,
      "step": 15294
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.645378828048706,
      "learning_rate": 1.1101715968872091e-05,
      "loss": 0.473,
      "step": 15295
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.554758071899414,
      "learning_rate": 1.110070375217e-05,
      "loss": 0.4129,
      "step": 15296
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6265878677368164,
      "learning_rate": 1.1099691524051854e-05,
      "loss": 0.5183,
      "step": 15297
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5543245077133179,
      "learning_rate": 1.1098679284528149e-05,
      "loss": 0.4679,
      "step": 15298
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6804301738739014,
      "learning_rate": 1.109766703360938e-05,
      "loss": 0.552,
      "step": 15299
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6289756298065186,
      "learning_rate": 1.1096654771306053e-05,
      "loss": 0.4149,
      "step": 15300
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6200720071792603,
      "learning_rate": 1.1095642497628657e-05,
      "loss": 0.4805,
      "step": 15301
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4904627799987793,
      "learning_rate": 1.1094630212587703e-05,
      "loss": 0.4062,
      "step": 15302
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5394717454910278,
      "learning_rate": 1.1093617916193677e-05,
      "loss": 0.4687,
      "step": 15303
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5656818151474,
      "learning_rate": 1.1092605608457088e-05,
      "loss": 0.4167,
      "step": 15304
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.615649938583374,
      "learning_rate": 1.1091593289388428e-05,
      "loss": 0.4794,
      "step": 15305
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.541045069694519,
      "learning_rate": 1.1090580958998205e-05,
      "loss": 0.4676,
      "step": 15306
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5131263732910156,
      "learning_rate": 1.1089568617296911e-05,
      "loss": 0.4708,
      "step": 15307
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4919899702072144,
      "learning_rate": 1.1088556264295048e-05,
      "loss": 0.4101,
      "step": 15308
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.548208236694336,
      "learning_rate": 1.1087543900003114e-05,
      "loss": 0.482,
      "step": 15309
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6615017652511597,
      "learning_rate": 1.1086531524431616e-05,
      "loss": 0.4381,
      "step": 15310
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7970973253250122,
      "learning_rate": 1.1085519137591044e-05,
      "loss": 0.9934,
      "step": 15311
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6064338684082031,
      "learning_rate": 1.1084506739491903e-05,
      "loss": 1.0519,
      "step": 15312
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.662420392036438,
      "learning_rate": 1.1083494330144693e-05,
      "loss": 0.4815,
      "step": 15313
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.668428897857666,
      "learning_rate": 1.1082481909559917e-05,
      "loss": 0.4548,
      "step": 15314
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5165579319000244,
      "learning_rate": 1.108146947774807e-05,
      "loss": 0.5002,
      "step": 15315
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5456993579864502,
      "learning_rate": 1.1080457034719656e-05,
      "loss": 0.4215,
      "step": 15316
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5467333793640137,
      "learning_rate": 1.1079444580485172e-05,
      "loss": 0.4575,
      "step": 15317
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4906375408172607,
      "learning_rate": 1.107843211505513e-05,
      "loss": 0.4507,
      "step": 15318
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5746263265609741,
      "learning_rate": 1.1077419638440012e-05,
      "loss": 0.4656,
      "step": 15319
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5388576984405518,
      "learning_rate": 1.1076407150650336e-05,
      "loss": 0.4395,
      "step": 15320
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6156930923461914,
      "learning_rate": 1.1075394651696593e-05,
      "loss": 0.4496,
      "step": 15321
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5647929906845093,
      "learning_rate": 1.107438214158929e-05,
      "loss": 0.449,
      "step": 15322
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5646884441375732,
      "learning_rate": 1.1073369620338928e-05,
      "loss": 0.4523,
      "step": 15323
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7490235567092896,
      "learning_rate": 1.1072357087956005e-05,
      "loss": 0.4415,
      "step": 15324
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6586120128631592,
      "learning_rate": 1.1071344544451024e-05,
      "loss": 0.4736,
      "step": 15325
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.567895770072937,
      "learning_rate": 1.1070331989834487e-05,
      "loss": 0.4404,
      "step": 15326
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5956649780273438,
      "learning_rate": 1.106931942411689e-05,
      "loss": 0.4536,
      "step": 15327
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4985922574996948,
      "learning_rate": 1.106830684730875e-05,
      "loss": 0.4091,
      "step": 15328
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9729143381118774,
      "learning_rate": 1.1067294259420554e-05,
      "loss": 0.4726,
      "step": 15329
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5741426944732666,
      "learning_rate": 1.1066281660462811e-05,
      "loss": 0.4283,
      "step": 15330
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5872820615768433,
      "learning_rate": 1.1065269050446024e-05,
      "loss": 0.5064,
      "step": 15331
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6408511400222778,
      "learning_rate": 1.1064256429380693e-05,
      "loss": 0.4568,
      "step": 15332
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5272315740585327,
      "learning_rate": 1.106324379727732e-05,
      "loss": 0.4576,
      "step": 15333
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5673487186431885,
      "learning_rate": 1.1062231154146407e-05,
      "loss": 0.4604,
      "step": 15334
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6722947359085083,
      "learning_rate": 1.1061218499998459e-05,
      "loss": 0.4263,
      "step": 15335
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.528327465057373,
      "learning_rate": 1.1060205834843982e-05,
      "loss": 0.4161,
      "step": 15336
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.576492428779602,
      "learning_rate": 1.105919315869347e-05,
      "loss": 0.5472,
      "step": 15337
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4976609945297241,
      "learning_rate": 1.1058180471557435e-05,
      "loss": 0.4267,
      "step": 15338
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6342602968215942,
      "learning_rate": 1.1057167773446373e-05,
      "loss": 0.4875,
      "step": 15339
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5686308145523071,
      "learning_rate": 1.1056155064370795e-05,
      "loss": 0.4256,
      "step": 15340
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6660798788070679,
      "learning_rate": 1.1055142344341197e-05,
      "loss": 0.5018,
      "step": 15341
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5178622007369995,
      "learning_rate": 1.1054129613368088e-05,
      "loss": 0.4537,
      "step": 15342
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4767534732818604,
      "learning_rate": 1.1053116871461967e-05,
      "loss": 0.4416,
      "step": 15343
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5543174743652344,
      "learning_rate": 1.105210411863334e-05,
      "loss": 0.4342,
      "step": 15344
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5379714965820312,
      "learning_rate": 1.1051091354892709e-05,
      "loss": 0.4653,
      "step": 15345
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4836914539337158,
      "learning_rate": 1.1050078580250585e-05,
      "loss": 0.4301,
      "step": 15346
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9390147924423218,
      "learning_rate": 1.1049065794717463e-05,
      "loss": 1.0344,
      "step": 15347
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7561829090118408,
      "learning_rate": 1.1048052998303854e-05,
      "loss": 1.073,
      "step": 15348
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6060245037078857,
      "learning_rate": 1.1047040191020257e-05,
      "loss": 0.4375,
      "step": 15349
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6409398317337036,
      "learning_rate": 1.104602737287718e-05,
      "loss": 0.4146,
      "step": 15350
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5210158824920654,
      "learning_rate": 1.1045014543885126e-05,
      "loss": 0.4368,
      "step": 15351
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.518302321434021,
      "learning_rate": 1.1044001704054601e-05,
      "loss": 0.432,
      "step": 15352
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7243940830230713,
      "learning_rate": 1.1042988853396106e-05,
      "loss": 0.5126,
      "step": 15353
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5224205255508423,
      "learning_rate": 1.1041975991920156e-05,
      "loss": 0.4246,
      "step": 15354
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6339455842971802,
      "learning_rate": 1.104096311963724e-05,
      "loss": 0.4788,
      "step": 15355
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5107640027999878,
      "learning_rate": 1.1039950236557877e-05,
      "loss": 0.3873,
      "step": 15356
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6421922445297241,
      "learning_rate": 1.1038937342692566e-05,
      "loss": 0.4574,
      "step": 15357
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5541980266571045,
      "learning_rate": 1.1037924438051812e-05,
      "loss": 0.4056,
      "step": 15358
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.550424337387085,
      "learning_rate": 1.1036911522646125e-05,
      "loss": 0.4934,
      "step": 15359
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.661142349243164,
      "learning_rate": 1.1035898596486006e-05,
      "loss": 0.4531,
      "step": 15360
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6242696046829224,
      "learning_rate": 1.1034885659581961e-05,
      "loss": 0.4755,
      "step": 15361
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6071707010269165,
      "learning_rate": 1.1033872711944497e-05,
      "loss": 0.4503,
      "step": 15362
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6161147356033325,
      "learning_rate": 1.103285975358412e-05,
      "loss": 0.5482,
      "step": 15363
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5606454610824585,
      "learning_rate": 1.1031846784511337e-05,
      "loss": 0.4164,
      "step": 15364
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6319208145141602,
      "learning_rate": 1.1030833804736652e-05,
      "loss": 0.4948,
      "step": 15365
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5370755195617676,
      "learning_rate": 1.102982081427057e-05,
      "loss": 0.3808,
      "step": 15366
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6482709646224976,
      "learning_rate": 1.10288078131236e-05,
      "loss": 0.4677,
      "step": 15367
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.510772466659546,
      "learning_rate": 1.1027794801306251e-05,
      "loss": 0.4215,
      "step": 15368
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.606465220451355,
      "learning_rate": 1.1026781778829025e-05,
      "loss": 0.4655,
      "step": 15369
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.507977843284607,
      "learning_rate": 1.102576874570243e-05,
      "loss": 0.4095,
      "step": 15370
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4918180704116821,
      "learning_rate": 1.102475570193697e-05,
      "loss": 0.4273,
      "step": 15371
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6358729600906372,
      "learning_rate": 1.1023742647543158e-05,
      "loss": 0.4595,
      "step": 15372
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5800318717956543,
      "learning_rate": 1.1022729582531495e-05,
      "loss": 0.4671,
      "step": 15373
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.504196286201477,
      "learning_rate": 1.1021716506912494e-05,
      "loss": 0.4505,
      "step": 15374
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5970044136047363,
      "learning_rate": 1.1020703420696656e-05,
      "loss": 0.474,
      "step": 15375
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5604859590530396,
      "learning_rate": 1.1019690323894494e-05,
      "loss": 0.4121,
      "step": 15376
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6234261989593506,
      "learning_rate": 1.1018677216516513e-05,
      "loss": 0.5103,
      "step": 15377
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5685733556747437,
      "learning_rate": 1.101766409857322e-05,
      "loss": 0.4223,
      "step": 15378
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6193314790725708,
      "learning_rate": 1.101665097007512e-05,
      "loss": 0.4668,
      "step": 15379
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5216830968856812,
      "learning_rate": 1.1015637831032727e-05,
      "loss": 0.414,
      "step": 15380
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5728294849395752,
      "learning_rate": 1.1014624681456543e-05,
      "loss": 0.4663,
      "step": 15381
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.598861575126648,
      "learning_rate": 1.101361152135708e-05,
      "loss": 0.4311,
      "step": 15382
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5130618810653687,
      "learning_rate": 1.1012598350744844e-05,
      "loss": 0.4395,
      "step": 15383
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4414039850234985,
      "learning_rate": 1.1011585169630345e-05,
      "loss": 0.4065,
      "step": 15384
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5876883268356323,
      "learning_rate": 1.1010571978024089e-05,
      "loss": 0.5104,
      "step": 15385
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6298086643218994,
      "learning_rate": 1.1009558775936587e-05,
      "loss": 0.415,
      "step": 15386
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5476233959197998,
      "learning_rate": 1.1008545563378345e-05,
      "loss": 0.4393,
      "step": 15387
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5028480291366577,
      "learning_rate": 1.1007532340359875e-05,
      "loss": 0.4295,
      "step": 15388
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4991446733474731,
      "learning_rate": 1.1006519106891678e-05,
      "loss": 0.4447,
      "step": 15389
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5920791625976562,
      "learning_rate": 1.1005505862984274e-05,
      "loss": 0.4091,
      "step": 15390
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6224100589752197,
      "learning_rate": 1.1004492608648163e-05,
      "loss": 0.467,
      "step": 15391
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5972363948822021,
      "learning_rate": 1.1003479343893859e-05,
      "loss": 0.4856,
      "step": 15392
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6586052179336548,
      "learning_rate": 1.1002466068731869e-05,
      "loss": 0.4732,
      "step": 15393
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4611499309539795,
      "learning_rate": 1.1001452783172704e-05,
      "loss": 0.4039,
      "step": 15394
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5502207279205322,
      "learning_rate": 1.100043948722687e-05,
      "loss": 0.4887,
      "step": 15395
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.483162522315979,
      "learning_rate": 1.0999426180904882e-05,
      "loss": 0.4214,
      "step": 15396
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6134339570999146,
      "learning_rate": 1.0998412864217242e-05,
      "loss": 0.4949,
      "step": 15397
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6208961009979248,
      "learning_rate": 1.099739953717447e-05,
      "loss": 0.4246,
      "step": 15398
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5778577327728271,
      "learning_rate": 1.0996386199787063e-05,
      "loss": 0.5049,
      "step": 15399
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4570423364639282,
      "learning_rate": 1.0995372852065543e-05,
      "loss": 0.413,
      "step": 15400
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6571853160858154,
      "learning_rate": 1.099435949402041e-05,
      "loss": 0.5172,
      "step": 15401
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5536857843399048,
      "learning_rate": 1.0993346125662184e-05,
      "loss": 0.4363,
      "step": 15402
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5308177471160889,
      "learning_rate": 1.0992332747001369e-05,
      "loss": 0.4672,
      "step": 15403
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5312280654907227,
      "learning_rate": 1.0991319358048473e-05,
      "loss": 0.4333,
      "step": 15404
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5324907302856445,
      "learning_rate": 1.0990305958814014e-05,
      "loss": 0.5144,
      "step": 15405
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6375335454940796,
      "learning_rate": 1.09892925493085e-05,
      "loss": 0.4562,
      "step": 15406
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5372204780578613,
      "learning_rate": 1.0988279129542434e-05,
      "loss": 0.5069,
      "step": 15407
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6419398784637451,
      "learning_rate": 1.0987265699526339e-05,
      "loss": 0.4654,
      "step": 15408
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5644530057907104,
      "learning_rate": 1.0986252259270716e-05,
      "loss": 0.5039,
      "step": 15409
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5711358785629272,
      "learning_rate": 1.0985238808786081e-05,
      "loss": 0.4402,
      "step": 15410
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5667625665664673,
      "learning_rate": 1.0984225348082946e-05,
      "loss": 0.5078,
      "step": 15411
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6285133361816406,
      "learning_rate": 1.098321187717182e-05,
      "loss": 0.4523,
      "step": 15412
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6270577907562256,
      "learning_rate": 1.0982198396063213e-05,
      "loss": 0.4726,
      "step": 15413
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4650911092758179,
      "learning_rate": 1.0981184904767638e-05,
      "loss": 0.4103,
      "step": 15414
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6982933282852173,
      "learning_rate": 1.098017140329561e-05,
      "loss": 0.5168,
      "step": 15415
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4961700439453125,
      "learning_rate": 1.097915789165763e-05,
      "loss": 0.3897,
      "step": 15416
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5967501401901245,
      "learning_rate": 1.0978144369864224e-05,
      "loss": 0.4877,
      "step": 15417
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4307981729507446,
      "learning_rate": 1.0977130837925894e-05,
      "loss": 0.3936,
      "step": 15418
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5339715480804443,
      "learning_rate": 1.0976117295853155e-05,
      "loss": 0.4816,
      "step": 15419
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5811994075775146,
      "learning_rate": 1.0975103743656519e-05,
      "loss": 0.4368,
      "step": 15420
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.553521990776062,
      "learning_rate": 1.0974090181346498e-05,
      "loss": 0.4423,
      "step": 15421
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6483713388442993,
      "learning_rate": 1.09730766089336e-05,
      "loss": 0.4271,
      "step": 15422
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5902717113494873,
      "learning_rate": 1.0972063026428348e-05,
      "loss": 0.5049,
      "step": 15423
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5122926235198975,
      "learning_rate": 1.0971049433841244e-05,
      "loss": 0.4142,
      "step": 15424
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5133354663848877,
      "learning_rate": 1.0970035831182807e-05,
      "loss": 0.4497,
      "step": 15425
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5213737487792969,
      "learning_rate": 1.0969022218463546e-05,
      "loss": 0.4578,
      "step": 15426
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4974769353866577,
      "learning_rate": 1.0968008595693973e-05,
      "loss": 0.4347,
      "step": 15427
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.623315691947937,
      "learning_rate": 1.0966994962884605e-05,
      "loss": 0.4331,
      "step": 15428
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7990810871124268,
      "learning_rate": 1.0965981320045954e-05,
      "loss": 1.0709,
      "step": 15429
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6304339170455933,
      "learning_rate": 1.0964967667188529e-05,
      "loss": 1.0761,
      "step": 15430
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.927649736404419,
      "learning_rate": 1.0963954004322849e-05,
      "loss": 0.9901,
      "step": 15431
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5069419145584106,
      "learning_rate": 1.0962940331459424e-05,
      "loss": 0.9567,
      "step": 15432
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7786264419555664,
      "learning_rate": 1.0961926648608768e-05,
      "loss": 0.5093,
      "step": 15433
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.692500114440918,
      "learning_rate": 1.0960912955781393e-05,
      "loss": 0.4423,
      "step": 15434
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6401928663253784,
      "learning_rate": 1.0959899252987815e-05,
      "loss": 0.4437,
      "step": 15435
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.556803584098816,
      "learning_rate": 1.0958885540238549e-05,
      "loss": 0.41,
      "step": 15436
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.541294813156128,
      "learning_rate": 1.0957871817544103e-05,
      "loss": 0.5225,
      "step": 15437
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6226768493652344,
      "learning_rate": 1.0956858084914999e-05,
      "loss": 0.4883,
      "step": 15438
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5996013879776,
      "learning_rate": 1.0955844342361744e-05,
      "loss": 0.4922,
      "step": 15439
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5498021841049194,
      "learning_rate": 1.0954830589894852e-05,
      "loss": 0.4577,
      "step": 15440
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6046663522720337,
      "learning_rate": 1.0953816827524844e-05,
      "loss": 0.5553,
      "step": 15441
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5675792694091797,
      "learning_rate": 1.0952803055262226e-05,
      "loss": 0.5293,
      "step": 15442
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5695735216140747,
      "learning_rate": 1.0951789273117521e-05,
      "loss": 0.5328,
      "step": 15443
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.534369707107544,
      "learning_rate": 1.0950775481101236e-05,
      "loss": 0.4125,
      "step": 15444
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5080349445343018,
      "learning_rate": 1.0949761679223892e-05,
      "loss": 0.4425,
      "step": 15445
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5975607633590698,
      "learning_rate": 1.0948747867495999e-05,
      "loss": 0.4389,
      "step": 15446
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5462008714675903,
      "learning_rate": 1.0947734045928073e-05,
      "loss": 0.4678,
      "step": 15447
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.513667345046997,
      "learning_rate": 1.0946720214530628e-05,
      "loss": 0.4306,
      "step": 15448
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5023090839385986,
      "learning_rate": 1.0945706373314187e-05,
      "loss": 0.4217,
      "step": 15449
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5410360097885132,
      "learning_rate": 1.094469252228925e-05,
      "loss": 0.4338,
      "step": 15450
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6343499422073364,
      "learning_rate": 1.0943678661466346e-05,
      "loss": 0.4889,
      "step": 15451
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7184062004089355,
      "learning_rate": 1.0942664790855985e-05,
      "loss": 0.4551,
      "step": 15452
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5128728151321411,
      "learning_rate": 1.0941650910468682e-05,
      "loss": 0.4263,
      "step": 15453
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5028762817382812,
      "learning_rate": 1.0940637020314955e-05,
      "loss": 0.3961,
      "step": 15454
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5165297985076904,
      "learning_rate": 1.0939623120405316e-05,
      "loss": 0.4721,
      "step": 15455
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4971513748168945,
      "learning_rate": 1.0938609210750281e-05,
      "loss": 0.4323,
      "step": 15456
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5386037826538086,
      "learning_rate": 1.0937595291360372e-05,
      "loss": 0.4757,
      "step": 15457
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5646851062774658,
      "learning_rate": 1.0936581362246094e-05,
      "loss": 0.4112,
      "step": 15458
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5937303304672241,
      "learning_rate": 1.093556742341798e-05,
      "loss": 0.466,
      "step": 15459
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5907981395721436,
      "learning_rate": 1.0934553474886525e-05,
      "loss": 0.4085,
      "step": 15460
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6292083263397217,
      "learning_rate": 1.0933539516662262e-05,
      "loss": 0.4706,
      "step": 15461
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.432837724685669,
      "learning_rate": 1.09325255487557e-05,
      "loss": 0.3897,
      "step": 15462
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.49920654296875,
      "learning_rate": 1.0931511571177356e-05,
      "loss": 0.5105,
      "step": 15463
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.604689121246338,
      "learning_rate": 1.093049758393775e-05,
      "loss": 0.4453,
      "step": 15464
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6507794857025146,
      "learning_rate": 1.0929483587047393e-05,
      "loss": 0.4932,
      "step": 15465
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.5911543369293213,
      "learning_rate": 1.0928469580516804e-05,
      "loss": 0.4354,
      "step": 15466
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.611855149269104,
      "learning_rate": 1.0927455564356506e-05,
      "loss": 0.519,
      "step": 15467
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6025004386901855,
      "learning_rate": 1.0926441538577005e-05,
      "loss": 0.4363,
      "step": 15468
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.570593237876892,
      "learning_rate": 1.0925427503188827e-05,
      "loss": 0.4587,
      "step": 15469
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.517962098121643,
      "learning_rate": 1.0924413458202485e-05,
      "loss": 0.4273,
      "step": 15470
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5437629222869873,
      "learning_rate": 1.09233994036285e-05,
      "loss": 0.585,
      "step": 15471
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4994012117385864,
      "learning_rate": 1.0922385339477384e-05,
      "loss": 0.4232,
      "step": 15472
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7424670457839966,
      "learning_rate": 1.0921371265759658e-05,
      "loss": 0.5414,
      "step": 15473
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.464342713356018,
      "learning_rate": 1.0920357182485837e-05,
      "loss": 0.3759,
      "step": 15474
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5837551355361938,
      "learning_rate": 1.091934308966644e-05,
      "loss": 0.4745,
      "step": 15475
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5687787532806396,
      "learning_rate": 1.0918328987311986e-05,
      "loss": 0.4424,
      "step": 15476
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0208911895751953,
      "learning_rate": 1.0917314875432996e-05,
      "loss": 0.9754,
      "step": 15477
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7382372617721558,
      "learning_rate": 1.0916300754039977e-05,
      "loss": 1.1125,
      "step": 15478
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5802323818206787,
      "learning_rate": 1.0915286623143458e-05,
      "loss": 0.5059,
      "step": 15479
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.516276240348816,
      "learning_rate": 1.0914272482753951e-05,
      "loss": 0.4066,
      "step": 15480
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6667351722717285,
      "learning_rate": 1.091325833288198e-05,
      "loss": 0.5207,
      "step": 15481
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6404200792312622,
      "learning_rate": 1.0912244173538056e-05,
      "loss": 0.4279,
      "step": 15482
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5719287395477295,
      "learning_rate": 1.0911230004732702e-05,
      "loss": 0.5217,
      "step": 15483
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5930899381637573,
      "learning_rate": 1.0910215826476435e-05,
      "loss": 0.4522,
      "step": 15484
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.587765097618103,
      "learning_rate": 1.0909201638779778e-05,
      "loss": 0.5058,
      "step": 15485
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5729402303695679,
      "learning_rate": 1.0908187441653242e-05,
      "loss": 0.4275,
      "step": 15486
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.673689365386963,
      "learning_rate": 1.0907173235107351e-05,
      "loss": 0.5095,
      "step": 15487
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4697201251983643,
      "learning_rate": 1.0906159019152625e-05,
      "loss": 0.4115,
      "step": 15488
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.560308814048767,
      "learning_rate": 1.0905144793799578e-05,
      "loss": 0.4461,
      "step": 15489
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.462547779083252,
      "learning_rate": 1.0904130559058735e-05,
      "loss": 0.3974,
      "step": 15490
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5438560247421265,
      "learning_rate": 1.0903116314940612e-05,
      "loss": 0.4589,
      "step": 15491
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5585566759109497,
      "learning_rate": 1.0902102061455724e-05,
      "loss": 0.4118,
      "step": 15492
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6332165002822876,
      "learning_rate": 1.09010877986146e-05,
      "loss": 0.4715,
      "step": 15493
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5907435417175293,
      "learning_rate": 1.0900073526427752e-05,
      "loss": 0.4259,
      "step": 15494
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5847492218017578,
      "learning_rate": 1.0899059244905706e-05,
      "loss": 0.4892,
      "step": 15495
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5989279747009277,
      "learning_rate": 1.0898044954058972e-05,
      "loss": 0.4346,
      "step": 15496
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.861777901649475,
      "learning_rate": 1.0897030653898081e-05,
      "loss": 0.9866,
      "step": 15497
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5414267778396606,
      "learning_rate": 1.0896016344433547e-05,
      "loss": 1.0823,
      "step": 15498
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6542632579803467,
      "learning_rate": 1.089500202567589e-05,
      "loss": 0.4334,
      "step": 15499
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6170923709869385,
      "learning_rate": 1.0893987697635631e-05,
      "loss": 0.422,
      "step": 15500
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7242109775543213,
      "learning_rate": 1.089297336032329e-05,
      "loss": 0.4746,
      "step": 15501
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5280096530914307,
      "learning_rate": 1.0891959013749382e-05,
      "loss": 0.3995,
      "step": 15502
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5342756509780884,
      "learning_rate": 1.089094465792444e-05,
      "loss": 0.4583,
      "step": 15503
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5709558725357056,
      "learning_rate": 1.0889930292858974e-05,
      "loss": 0.4347,
      "step": 15504
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5961649417877197,
      "learning_rate": 1.088891591856351e-05,
      "loss": 0.472,
      "step": 15505
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.576315999031067,
      "learning_rate": 1.0887901535048564e-05,
      "loss": 0.448,
      "step": 15506
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5091216564178467,
      "learning_rate": 1.0886887142324661e-05,
      "loss": 0.4547,
      "step": 15507
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4879951477050781,
      "learning_rate": 1.088587274040232e-05,
      "loss": 0.433,
      "step": 15508
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5679478645324707,
      "learning_rate": 1.0884858329292061e-05,
      "loss": 0.4419,
      "step": 15509
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6470506191253662,
      "learning_rate": 1.0883843909004404e-05,
      "loss": 0.4743,
      "step": 15510
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5772876739501953,
      "learning_rate": 1.0882829479549876e-05,
      "loss": 0.469,
      "step": 15511
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5918612480163574,
      "learning_rate": 1.0881815040938992e-05,
      "loss": 0.4244,
      "step": 15512
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.596053123474121,
      "learning_rate": 1.0880800593182281e-05,
      "loss": 0.4494,
      "step": 15513
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.606847882270813,
      "learning_rate": 1.0879786136290253e-05,
      "loss": 0.4279,
      "step": 15514
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6076087951660156,
      "learning_rate": 1.087877167027344e-05,
      "loss": 0.4723,
      "step": 15515
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.518521785736084,
      "learning_rate": 1.0877757195142358e-05,
      "loss": 0.4357,
      "step": 15516
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5380439758300781,
      "learning_rate": 1.0876742710907531e-05,
      "loss": 0.4918,
      "step": 15517
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5080678462982178,
      "learning_rate": 1.0875728217579479e-05,
      "loss": 0.4265,
      "step": 15518
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5078991651535034,
      "learning_rate": 1.0874713715168728e-05,
      "loss": 0.4491,
      "step": 15519
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5244007110595703,
      "learning_rate": 1.087369920368579e-05,
      "loss": 0.442,
      "step": 15520
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6012108325958252,
      "learning_rate": 1.0872684683141204e-05,
      "loss": 0.4706,
      "step": 15521
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5264543294906616,
      "learning_rate": 1.0871670153545473e-05,
      "loss": 0.4041,
      "step": 15522
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5426872968673706,
      "learning_rate": 1.0870655614909134e-05,
      "loss": 0.4861,
      "step": 15523
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5373384952545166,
      "learning_rate": 1.0869641067242702e-05,
      "loss": 0.4115,
      "step": 15524
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6852792501449585,
      "learning_rate": 1.0868626510556704e-05,
      "loss": 0.4815,
      "step": 15525
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.546536922454834,
      "learning_rate": 1.0867611944861659e-05,
      "loss": 0.4195,
      "step": 15526
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5526376962661743,
      "learning_rate": 1.0866597370168091e-05,
      "loss": 0.4616,
      "step": 15527
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.573585033416748,
      "learning_rate": 1.086558278648652e-05,
      "loss": 0.4397,
      "step": 15528
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5158740282058716,
      "learning_rate": 1.0864568193827474e-05,
      "loss": 0.4251,
      "step": 15529
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5638097524642944,
      "learning_rate": 1.0863553592201471e-05,
      "loss": 0.4282,
      "step": 15530
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5443731546401978,
      "learning_rate": 1.086253898161904e-05,
      "loss": 0.4573,
      "step": 15531
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.589823842048645,
      "learning_rate": 1.0861524362090696e-05,
      "loss": 0.4311,
      "step": 15532
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6252515316009521,
      "learning_rate": 1.0860509733626969e-05,
      "loss": 0.5667,
      "step": 15533
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5467585325241089,
      "learning_rate": 1.0859495096238379e-05,
      "loss": 0.4154,
      "step": 15534
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5960259437561035,
      "learning_rate": 1.0858480449935453e-05,
      "loss": 0.4858,
      "step": 15535
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.639695644378662,
      "learning_rate": 1.085746579472871e-05,
      "loss": 0.4308,
      "step": 15536
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5160369873046875,
      "learning_rate": 1.0856451130628675e-05,
      "loss": 0.451,
      "step": 15537
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6060748100280762,
      "learning_rate": 1.0855436457645874e-05,
      "loss": 0.4449,
      "step": 15538
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8278800249099731,
      "learning_rate": 1.0854421775790829e-05,
      "loss": 1.0562,
      "step": 15539
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.61802339553833,
      "learning_rate": 1.085340708507406e-05,
      "loss": 1.0463,
      "step": 15540
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5773918628692627,
      "learning_rate": 1.0852392385506102e-05,
      "loss": 0.4397,
      "step": 15541
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6564334630966187,
      "learning_rate": 1.0851377677097465e-05,
      "loss": 0.4765,
      "step": 15542
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.568671703338623,
      "learning_rate": 1.0850362959858684e-05,
      "loss": 0.4855,
      "step": 15543
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5038626194000244,
      "learning_rate": 1.0849348233800278e-05,
      "loss": 0.4127,
      "step": 15544
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5112180709838867,
      "learning_rate": 1.0848333498932773e-05,
      "loss": 0.4853,
      "step": 15545
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5004664659500122,
      "learning_rate": 1.0847318755266692e-05,
      "loss": 0.4314,
      "step": 15546
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.640397071838379,
      "learning_rate": 1.0846304002812564e-05,
      "loss": 0.4988,
      "step": 15547
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6242517232894897,
      "learning_rate": 1.0845289241580906e-05,
      "loss": 0.4491,
      "step": 15548
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.538110613822937,
      "learning_rate": 1.0844274471582249e-05,
      "loss": 0.4754,
      "step": 15549
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5997065305709839,
      "learning_rate": 1.0843259692827116e-05,
      "loss": 0.4195,
      "step": 15550
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5235296487808228,
      "learning_rate": 1.0842244905326029e-05,
      "loss": 0.4553,
      "step": 15551
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.610683560371399,
      "learning_rate": 1.0841230109089518e-05,
      "loss": 0.4522,
      "step": 15552
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.524176836013794,
      "learning_rate": 1.0840215304128106e-05,
      "loss": 0.4691,
      "step": 15553
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4221421480178833,
      "learning_rate": 1.0839200490452317e-05,
      "loss": 0.4034,
      "step": 15554
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5069987773895264,
      "learning_rate": 1.0838185668072676e-05,
      "loss": 0.5186,
      "step": 15555
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.530654788017273,
      "learning_rate": 1.0837170836999708e-05,
      "loss": 0.413,
      "step": 15556
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5835213661193848,
      "learning_rate": 1.0836155997243944e-05,
      "loss": 0.4894,
      "step": 15557
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.485936164855957,
      "learning_rate": 1.0835141148815902e-05,
      "loss": 0.4135,
      "step": 15558
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5771247148513794,
      "learning_rate": 1.083412629172611e-05,
      "loss": 0.5177,
      "step": 15559
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5963026285171509,
      "learning_rate": 1.0833111425985097e-05,
      "loss": 0.4085,
      "step": 15560
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5532422065734863,
      "learning_rate": 1.0832096551603386e-05,
      "loss": 0.4802,
      "step": 15561
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.51571786403656,
      "learning_rate": 1.0831081668591503e-05,
      "loss": 0.4073,
      "step": 15562
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5656191110610962,
      "learning_rate": 1.0830066776959973e-05,
      "loss": 0.4803,
      "step": 15563
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5199788808822632,
      "learning_rate": 1.0829051876719323e-05,
      "loss": 0.4153,
      "step": 15564
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7315526008605957,
      "learning_rate": 1.0828036967880082e-05,
      "loss": 0.9834,
      "step": 15565
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.615930438041687,
      "learning_rate": 1.082702205045277e-05,
      "loss": 1.048,
      "step": 15566
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6835072040557861,
      "learning_rate": 1.0826007124447919e-05,
      "loss": 0.4731,
      "step": 15567
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6044725179672241,
      "learning_rate": 1.0824992189876054e-05,
      "loss": 0.4377,
      "step": 15568
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6718941926956177,
      "learning_rate": 1.08239772467477e-05,
      "loss": 0.4843,
      "step": 15569
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5725862979888916,
      "learning_rate": 1.0822962295073382e-05,
      "loss": 0.4614,
      "step": 15570
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.57500159740448,
      "learning_rate": 1.0821947334863634e-05,
      "loss": 0.4602,
      "step": 15571
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4830965995788574,
      "learning_rate": 1.0820932366128973e-05,
      "loss": 0.4096,
      "step": 15572
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.58882737159729,
      "learning_rate": 1.0819917388879935e-05,
      "loss": 0.4634,
      "step": 15573
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6620663404464722,
      "learning_rate": 1.0818902403127037e-05,
      "loss": 0.437,
      "step": 15574
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5509167909622192,
      "learning_rate": 1.0817887408880816e-05,
      "loss": 0.4713,
      "step": 15575
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5175776481628418,
      "learning_rate": 1.0816872406151793e-05,
      "loss": 0.4397,
      "step": 15576
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.702280879020691,
      "learning_rate": 1.0815857394950497e-05,
      "loss": 0.5661,
      "step": 15577
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5379823446273804,
      "learning_rate": 1.0814842375287456e-05,
      "loss": 0.4196,
      "step": 15578
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5854740142822266,
      "learning_rate": 1.0813827347173196e-05,
      "loss": 0.4543,
      "step": 15579
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5303317308425903,
      "learning_rate": 1.0812812310618244e-05,
      "loss": 0.4188,
      "step": 15580
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5892337560653687,
      "learning_rate": 1.0811797265633131e-05,
      "loss": 0.4731,
      "step": 15581
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5158571004867554,
      "learning_rate": 1.081078221222838e-05,
      "loss": 0.4323,
      "step": 15582
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5118720531463623,
      "learning_rate": 1.0809767150414523e-05,
      "loss": 0.4574,
      "step": 15583
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.516631007194519,
      "learning_rate": 1.0808752080202083e-05,
      "loss": 0.4326,
      "step": 15584
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5674800872802734,
      "learning_rate": 1.0807737001601593e-05,
      "loss": 0.4879,
      "step": 15585
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5937458276748657,
      "learning_rate": 1.0806721914623576e-05,
      "loss": 0.4039,
      "step": 15586
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5373504161834717,
      "learning_rate": 1.0805706819278565e-05,
      "loss": 0.481,
      "step": 15587
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5893725156784058,
      "learning_rate": 1.0804691715577087e-05,
      "loss": 0.4523,
      "step": 15588
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6795082092285156,
      "learning_rate": 1.0803676603529666e-05,
      "loss": 0.4609,
      "step": 15589
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5240200757980347,
      "learning_rate": 1.0802661483146837e-05,
      "loss": 0.3835,
      "step": 15590
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5260484218597412,
      "learning_rate": 1.0801646354439123e-05,
      "loss": 0.4434,
      "step": 15591
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4504019021987915,
      "learning_rate": 1.0800631217417053e-05,
      "loss": 0.4093,
      "step": 15592
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.583341121673584,
      "learning_rate": 1.079961607209116e-05,
      "loss": 0.4647,
      "step": 15593
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5157482624053955,
      "learning_rate": 1.0798600918471967e-05,
      "loss": 0.4242,
      "step": 15594
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5892833471298218,
      "learning_rate": 1.0797585756570007e-05,
      "loss": 0.4917,
      "step": 15595
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5472490787506104,
      "learning_rate": 1.0796570586395807e-05,
      "loss": 0.4373,
      "step": 15596
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6825194358825684,
      "learning_rate": 1.0795555407959894e-05,
      "loss": 0.5408,
      "step": 15597
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5069050788879395,
      "learning_rate": 1.0794540221272803e-05,
      "loss": 0.4026,
      "step": 15598
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.593915343284607,
      "learning_rate": 1.0793525026345057e-05,
      "loss": 0.5118,
      "step": 15599
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5263113975524902,
      "learning_rate": 1.0792509823187188e-05,
      "loss": 0.3843,
      "step": 15600
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.53522527217865,
      "learning_rate": 1.0791494611809728e-05,
      "loss": 0.4571,
      "step": 15601
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4925371408462524,
      "learning_rate": 1.0790479392223196e-05,
      "loss": 0.4173,
      "step": 15602
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5446264743804932,
      "learning_rate": 1.0789464164438135e-05,
      "loss": 0.454,
      "step": 15603
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5245660543441772,
      "learning_rate": 1.0788448928465065e-05,
      "loss": 0.4371,
      "step": 15604
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5194190740585327,
      "learning_rate": 1.0787433684314521e-05,
      "loss": 0.4717,
      "step": 15605
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4721201658248901,
      "learning_rate": 1.078641843199703e-05,
      "loss": 0.4012,
      "step": 15606
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7846966981887817,
      "learning_rate": 1.0785403171523121e-05,
      "loss": 0.4962,
      "step": 15607
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5344382524490356,
      "learning_rate": 1.0784387902903326e-05,
      "loss": 0.4591,
      "step": 15608
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5504789352416992,
      "learning_rate": 1.0783372626148177e-05,
      "loss": 0.48,
      "step": 15609
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.533539891242981,
      "learning_rate": 1.0782357341268195e-05,
      "loss": 0.4209,
      "step": 15610
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0719642639160156,
      "learning_rate": 1.0781342048273922e-05,
      "loss": 0.9883,
      "step": 15611
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6911187171936035,
      "learning_rate": 1.0780326747175878e-05,
      "loss": 1.0359,
      "step": 15612
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6189751625061035,
      "learning_rate": 1.0779311437984601e-05,
      "loss": 0.4978,
      "step": 15613
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.031748056411743,
      "learning_rate": 1.0778296120710618e-05,
      "loss": 0.4081,
      "step": 15614
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5964059829711914,
      "learning_rate": 1.0777280795364457e-05,
      "loss": 0.528,
      "step": 15615
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.4821836948394775,
      "learning_rate": 1.0776265461956655e-05,
      "loss": 0.4113,
      "step": 15616
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6536730527877808,
      "learning_rate": 1.0775250120497738e-05,
      "loss": 0.4734,
      "step": 15617
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4387139081954956,
      "learning_rate": 1.0774234770998232e-05,
      "loss": 0.4136,
      "step": 15618
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6833648681640625,
      "learning_rate": 1.0773219413468683e-05,
      "loss": 0.5052,
      "step": 15619
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5249086618423462,
      "learning_rate": 1.0772204047919606e-05,
      "loss": 0.4267,
      "step": 15620
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5430010557174683,
      "learning_rate": 1.0771188674361537e-05,
      "loss": 0.4467,
      "step": 15621
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5501034259796143,
      "learning_rate": 1.0770173292805012e-05,
      "loss": 0.4264,
      "step": 15622
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.851144552230835,
      "learning_rate": 1.0769157903260557e-05,
      "loss": 0.5138,
      "step": 15623
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5349383354187012,
      "learning_rate": 1.0768142505738705e-05,
      "loss": 0.408,
      "step": 15624
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5492595434188843,
      "learning_rate": 1.0767127100249986e-05,
      "loss": 0.4732,
      "step": 15625
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5309443473815918,
      "learning_rate": 1.0766111686804932e-05,
      "loss": 0.4349,
      "step": 15626
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5821974277496338,
      "learning_rate": 1.0765096265414076e-05,
      "loss": 0.4913,
      "step": 15627
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.539801001548767,
      "learning_rate": 1.0764080836087946e-05,
      "loss": 0.453,
      "step": 15628
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6096500158309937,
      "learning_rate": 1.076306539883708e-05,
      "loss": 0.5291,
      "step": 15629
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5670195817947388,
      "learning_rate": 1.0762049953672002e-05,
      "loss": 0.4544,
      "step": 15630
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6109131574630737,
      "learning_rate": 1.0761034500603249e-05,
      "loss": 0.4683,
      "step": 15631
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4253119230270386,
      "learning_rate": 1.0760019039641351e-05,
      "loss": 0.3898,
      "step": 15632
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5413812398910522,
      "learning_rate": 1.0759003570796843e-05,
      "loss": 0.4563,
      "step": 15633
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5066889524459839,
      "learning_rate": 1.0757988094080249e-05,
      "loss": 0.4546,
      "step": 15634
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.591265082359314,
      "learning_rate": 1.075697260950211e-05,
      "loss": 0.4791,
      "step": 15635
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5318784713745117,
      "learning_rate": 1.0755957117072951e-05,
      "loss": 0.4083,
      "step": 15636
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5814790725708008,
      "learning_rate": 1.0754941616803314e-05,
      "loss": 0.5228,
      "step": 15637
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5965933799743652,
      "learning_rate": 1.0753926108703719e-05,
      "loss": 0.4435,
      "step": 15638
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5373404026031494,
      "learning_rate": 1.0752910592784706e-05,
      "loss": 0.4739,
      "step": 15639
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5653754472732544,
      "learning_rate": 1.0751895069056808e-05,
      "loss": 0.449,
      "step": 15640
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6192156076431274,
      "learning_rate": 1.0750879537530556e-05,
      "loss": 0.4527,
      "step": 15641
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5355286598205566,
      "learning_rate": 1.074986399821648e-05,
      "loss": 0.414,
      "step": 15642
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6230182647705078,
      "learning_rate": 1.074884845112512e-05,
      "loss": 0.4844,
      "step": 15643
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5564076900482178,
      "learning_rate": 1.0747832896266998e-05,
      "loss": 0.4144,
      "step": 15644
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6359573602676392,
      "learning_rate": 1.074681733365266e-05,
      "loss": 0.4506,
      "step": 15645
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4685022830963135,
      "learning_rate": 1.0745801763292625e-05,
      "loss": 0.429,
      "step": 15646
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.628694772720337,
      "learning_rate": 1.0744786185197438e-05,
      "loss": 0.4684,
      "step": 15647
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.524633765220642,
      "learning_rate": 1.0743770599377624e-05,
      "loss": 0.4339,
      "step": 15648
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6319293975830078,
      "learning_rate": 1.0742755005843721e-05,
      "loss": 0.5439,
      "step": 15649
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6182788610458374,
      "learning_rate": 1.0741739404606262e-05,
      "loss": 0.4386,
      "step": 15650
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6207548379898071,
      "learning_rate": 1.0740723795675776e-05,
      "loss": 0.4452,
      "step": 15651
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.58534836769104,
      "learning_rate": 1.0739708179062802e-05,
      "loss": 0.4563,
      "step": 15652
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6202921867370605,
      "learning_rate": 1.0738692554777872e-05,
      "loss": 0.4699,
      "step": 15653
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5403461456298828,
      "learning_rate": 1.0737676922831517e-05,
      "loss": 0.4203,
      "step": 15654
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5599316358566284,
      "learning_rate": 1.0736661283234275e-05,
      "loss": 0.4653,
      "step": 15655
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6262556314468384,
      "learning_rate": 1.0735645635996676e-05,
      "loss": 0.3929,
      "step": 15656
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0692849159240723,
      "learning_rate": 1.0734629981129254e-05,
      "loss": 0.487,
      "step": 15657
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.500624179840088,
      "learning_rate": 1.0733614318642546e-05,
      "loss": 0.4102,
      "step": 15658
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6496037244796753,
      "learning_rate": 1.0732598648547084e-05,
      "loss": 0.4823,
      "step": 15659
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4983704090118408,
      "learning_rate": 1.0731582970853407e-05,
      "loss": 0.4106,
      "step": 15660
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5992109775543213,
      "learning_rate": 1.0730567285572037e-05,
      "loss": 0.5241,
      "step": 15661
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5483088493347168,
      "learning_rate": 1.0729551592713521e-05,
      "loss": 0.4645,
      "step": 15662
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5723576545715332,
      "learning_rate": 1.0728535892288388e-05,
      "loss": 0.454,
      "step": 15663
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5905160903930664,
      "learning_rate": 1.0727520184307171e-05,
      "loss": 0.4205,
      "step": 15664
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6250687837600708,
      "learning_rate": 1.0726504468780407e-05,
      "loss": 0.4908,
      "step": 15665
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5240919589996338,
      "learning_rate": 1.0725488745718631e-05,
      "loss": 0.4171,
      "step": 15666
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.911743402481079,
      "learning_rate": 1.0724473015132377e-05,
      "loss": 1.0603,
      "step": 15667
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.502052903175354,
      "learning_rate": 1.0723457277032179e-05,
      "loss": 0.9571,
      "step": 15668
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.617232084274292,
      "learning_rate": 1.0722441531428569e-05,
      "loss": 0.4431,
      "step": 15669
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.648376703262329,
      "learning_rate": 1.0721425778332089e-05,
      "loss": 0.4058,
      "step": 15670
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6005592346191406,
      "learning_rate": 1.0720410017753267e-05,
      "loss": 0.4341,
      "step": 15671
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5505508184432983,
      "learning_rate": 1.0719394249702641e-05,
      "loss": 0.418,
      "step": 15672
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.569543719291687,
      "learning_rate": 1.0718378474190749e-05,
      "loss": 0.4964,
      "step": 15673
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5576221942901611,
      "learning_rate": 1.0717362691228124e-05,
      "loss": 0.4063,
      "step": 15674
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.533218502998352,
      "learning_rate": 1.0716346900825298e-05,
      "loss": 0.4318,
      "step": 15675
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6312083005905151,
      "learning_rate": 1.071533110299281e-05,
      "loss": 0.4265,
      "step": 15676
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5190393924713135,
      "learning_rate": 1.0714315297741193e-05,
      "loss": 0.4418,
      "step": 15677
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4842627048492432,
      "learning_rate": 1.071329948508099e-05,
      "loss": 0.4358,
      "step": 15678
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5837366580963135,
      "learning_rate": 1.0712283665022725e-05,
      "loss": 0.4182,
      "step": 15679
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5240384340286255,
      "learning_rate": 1.0711267837576939e-05,
      "loss": 0.4366,
      "step": 15680
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.517693042755127,
      "learning_rate": 1.0710252002754172e-05,
      "loss": 0.4551,
      "step": 15681
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.606766939163208,
      "learning_rate": 1.0709236160564954e-05,
      "loss": 0.4824,
      "step": 15682
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6334593296051025,
      "learning_rate": 1.0708220311019824e-05,
      "loss": 0.5151,
      "step": 15683
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4927971363067627,
      "learning_rate": 1.0707204454129316e-05,
      "loss": 0.3905,
      "step": 15684
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6883351802825928,
      "learning_rate": 1.0706188589903965e-05,
      "loss": 0.5187,
      "step": 15685
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5971922874450684,
      "learning_rate": 1.0705172718354313e-05,
      "loss": 0.4252,
      "step": 15686
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.517020583152771,
      "learning_rate": 1.070415683949089e-05,
      "loss": 0.4475,
      "step": 15687
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5574028491973877,
      "learning_rate": 1.0703140953324236e-05,
      "loss": 0.4444,
      "step": 15688
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5582928657531738,
      "learning_rate": 1.0702125059864882e-05,
      "loss": 0.4367,
      "step": 15689
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5730655193328857,
      "learning_rate": 1.070110915912337e-05,
      "loss": 0.3917,
      "step": 15690
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6708816289901733,
      "learning_rate": 1.0700093251110238e-05,
      "loss": 0.5113,
      "step": 15691
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4733303785324097,
      "learning_rate": 1.0699077335836017e-05,
      "loss": 0.404,
      "step": 15692
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.524341106414795,
      "learning_rate": 1.0698061413311246e-05,
      "loss": 0.5221,
      "step": 15693
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5372039079666138,
      "learning_rate": 1.0697045483546465e-05,
      "loss": 0.4405,
      "step": 15694
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5501115322113037,
      "learning_rate": 1.0696029546552202e-05,
      "loss": 0.4645,
      "step": 15695
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6625289916992188,
      "learning_rate": 1.0695013602339006e-05,
      "loss": 0.4538,
      "step": 15696
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.68917715549469,
      "learning_rate": 1.0693997650917403e-05,
      "loss": 0.4534,
      "step": 15697
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.533739447593689,
      "learning_rate": 1.0692981692297935e-05,
      "loss": 0.4081,
      "step": 15698
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.837631344795227,
      "learning_rate": 1.0691965726491141e-05,
      "loss": 0.4344,
      "step": 15699
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6130521297454834,
      "learning_rate": 1.0690949753507554e-05,
      "loss": 0.4417,
      "step": 15700
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4637610912322998,
      "learning_rate": 1.0689933773357714e-05,
      "loss": 0.4612,
      "step": 15701
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5427231788635254,
      "learning_rate": 1.0688917786052157e-05,
      "loss": 0.426,
      "step": 15702
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.592361330986023,
      "learning_rate": 1.068790179160142e-05,
      "loss": 0.4916,
      "step": 15703
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5092397928237915,
      "learning_rate": 1.0686885790016044e-05,
      "loss": 0.4225,
      "step": 15704
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4975985288619995,
      "learning_rate": 1.068586978130656e-05,
      "loss": 0.44,
      "step": 15705
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5267221927642822,
      "learning_rate": 1.0684853765483517e-05,
      "loss": 0.4018,
      "step": 15706
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.314159870147705,
      "learning_rate": 1.0683837742557437e-05,
      "loss": 0.5055,
      "step": 15707
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5568158626556396,
      "learning_rate": 1.0682821712538872e-05,
      "loss": 0.4666,
      "step": 15708
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6249479055404663,
      "learning_rate": 1.068180567543835e-05,
      "loss": 0.4742,
      "step": 15709
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.359649419784546,
      "learning_rate": 1.0680789631266414e-05,
      "loss": 0.4275,
      "step": 15710
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5768433809280396,
      "learning_rate": 1.0679773580033604e-05,
      "loss": 0.4732,
      "step": 15711
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5499317646026611,
      "learning_rate": 1.0678757521750451e-05,
      "loss": 0.4223,
      "step": 15712
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6648645401000977,
      "learning_rate": 1.0677741456427497e-05,
      "loss": 0.4796,
      "step": 15713
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4629242420196533,
      "learning_rate": 1.0676725384075283e-05,
      "loss": 0.3964,
      "step": 15714
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5558786392211914,
      "learning_rate": 1.0675709304704343e-05,
      "loss": 0.473,
      "step": 15715
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4702755212783813,
      "learning_rate": 1.0674693218325216e-05,
      "loss": 0.4283,
      "step": 15716
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.773866653442383,
      "learning_rate": 1.0673677124948445e-05,
      "loss": 0.4748,
      "step": 15717
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5167146921157837,
      "learning_rate": 1.067266102458456e-05,
      "loss": 0.4219,
      "step": 15718
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.628566026687622,
      "learning_rate": 1.0671644917244111e-05,
      "loss": 0.4702,
      "step": 15719
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4954665899276733,
      "learning_rate": 1.0670628802937626e-05,
      "loss": 0.4098,
      "step": 15720
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5629338026046753,
      "learning_rate": 1.0669612681675644e-05,
      "loss": 0.4535,
      "step": 15721
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.60013747215271,
      "learning_rate": 1.0668596553468714e-05,
      "loss": 0.4332,
      "step": 15722
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.539759874343872,
      "learning_rate": 1.0667580418327368e-05,
      "loss": 0.4074,
      "step": 15723
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5559704303741455,
      "learning_rate": 1.0666564276262146e-05,
      "loss": 0.4097,
      "step": 15724
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.52420973777771,
      "learning_rate": 1.0665548127283583e-05,
      "loss": 0.5007,
      "step": 15725
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5427660942077637,
      "learning_rate": 1.0664531971402223e-05,
      "loss": 0.458,
      "step": 15726
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6234126091003418,
      "learning_rate": 1.0663515808628604e-05,
      "loss": 0.5502,
      "step": 15727
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4689202308654785,
      "learning_rate": 1.0662499638973267e-05,
      "loss": 0.4009,
      "step": 15728
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5817601680755615,
      "learning_rate": 1.0661483462446746e-05,
      "loss": 0.4485,
      "step": 15729
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4946976900100708,
      "learning_rate": 1.0660467279059587e-05,
      "loss": 0.4194,
      "step": 15730
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5904310941696167,
      "learning_rate": 1.0659451088822324e-05,
      "loss": 0.4934,
      "step": 15731
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.653252363204956,
      "learning_rate": 1.06584348917455e-05,
      "loss": 0.4215,
      "step": 15732
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6386882066726685,
      "learning_rate": 1.065741868783965e-05,
      "loss": 0.5735,
      "step": 15733
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.545034646987915,
      "learning_rate": 1.0656402477115322e-05,
      "loss": 0.4063,
      "step": 15734
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5641382932662964,
      "learning_rate": 1.0655386259583047e-05,
      "loss": 0.4821,
      "step": 15735
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5944232940673828,
      "learning_rate": 1.065437003525337e-05,
      "loss": 0.4293,
      "step": 15736
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6099425554275513,
      "learning_rate": 1.065335380413683e-05,
      "loss": 0.4526,
      "step": 15737
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5961365699768066,
      "learning_rate": 1.0652337566243964e-05,
      "loss": 0.4168,
      "step": 15738
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4462095499038696,
      "learning_rate": 1.0651321321585315e-05,
      "loss": 0.4305,
      "step": 15739
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5570123195648193,
      "learning_rate": 1.0650305070171425e-05,
      "loss": 0.4401,
      "step": 15740
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5578594207763672,
      "learning_rate": 1.0649288812012828e-05,
      "loss": 0.5072,
      "step": 15741
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6370329856872559,
      "learning_rate": 1.064827254712007e-05,
      "loss": 0.4658,
      "step": 15742
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5660653114318848,
      "learning_rate": 1.0647256275503689e-05,
      "loss": 0.491,
      "step": 15743
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4816186428070068,
      "learning_rate": 1.0646239997174224e-05,
      "loss": 0.4083,
      "step": 15744
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5769867897033691,
      "learning_rate": 1.064522371214222e-05,
      "loss": 0.4772,
      "step": 15745
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4739378690719604,
      "learning_rate": 1.064420742041821e-05,
      "loss": 0.4071,
      "step": 15746
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6349118947982788,
      "learning_rate": 1.0643191122012742e-05,
      "loss": 0.5466,
      "step": 15747
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.510880947113037,
      "learning_rate": 1.064217481693635e-05,
      "loss": 0.3935,
      "step": 15748
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5751665830612183,
      "learning_rate": 1.064115850519958e-05,
      "loss": 0.4727,
      "step": 15749
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.498621940612793,
      "learning_rate": 1.0640142186812974e-05,
      "loss": 0.4156,
      "step": 15750
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5827460289001465,
      "learning_rate": 1.0639125861787065e-05,
      "loss": 0.4837,
      "step": 15751
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5753731727600098,
      "learning_rate": 1.0638109530132401e-05,
      "loss": 0.4335,
      "step": 15752
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6812154054641724,
      "learning_rate": 1.0637093191859521e-05,
      "loss": 0.5197,
      "step": 15753
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4659656286239624,
      "learning_rate": 1.0636076846978968e-05,
      "loss": 0.3857,
      "step": 15754
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5454877614974976,
      "learning_rate": 1.0635060495501279e-05,
      "loss": 0.4768,
      "step": 15755
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6482511758804321,
      "learning_rate": 1.0634044137436997e-05,
      "loss": 0.431,
      "step": 15756
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6228240728378296,
      "learning_rate": 1.0633027772796662e-05,
      "loss": 0.4879,
      "step": 15757
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5365824699401855,
      "learning_rate": 1.063201140159082e-05,
      "loss": 0.4592,
      "step": 15758
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5091111660003662,
      "learning_rate": 1.0630995023830007e-05,
      "loss": 0.4907,
      "step": 15759
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6116349697113037,
      "learning_rate": 1.0629978639524766e-05,
      "loss": 0.4534,
      "step": 15760
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.570784330368042,
      "learning_rate": 1.0628962248685641e-05,
      "loss": 0.4792,
      "step": 15761
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5571863651275635,
      "learning_rate": 1.0627945851323171e-05,
      "loss": 0.4659,
      "step": 15762
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5089083909988403,
      "learning_rate": 1.0626929447447898e-05,
      "loss": 0.4754,
      "step": 15763
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.536550521850586,
      "learning_rate": 1.0625913037070363e-05,
      "loss": 0.4033,
      "step": 15764
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5662966966629028,
      "learning_rate": 1.062489662020111e-05,
      "loss": 0.4498,
      "step": 15765
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4491980075836182,
      "learning_rate": 1.0623880196850683e-05,
      "loss": 0.3924,
      "step": 15766
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7387340068817139,
      "learning_rate": 1.0622863767029615e-05,
      "loss": 0.5056,
      "step": 15767
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5566236972808838,
      "learning_rate": 1.0621847330748459e-05,
      "loss": 0.3791,
      "step": 15768
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5479109287261963,
      "learning_rate": 1.0620830888017746e-05,
      "loss": 0.5025,
      "step": 15769
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4822189807891846,
      "learning_rate": 1.0619814438848029e-05,
      "loss": 0.4125,
      "step": 15770
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.645067811012268,
      "learning_rate": 1.0618797983249842e-05,
      "loss": 0.4688,
      "step": 15771
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5360329151153564,
      "learning_rate": 1.0617781521233732e-05,
      "loss": 0.4209,
      "step": 15772
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.594296932220459,
      "learning_rate": 1.061676505281024e-05,
      "loss": 0.4614,
      "step": 15773
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4358795881271362,
      "learning_rate": 1.0615748577989908e-05,
      "loss": 0.4065,
      "step": 15774
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5593253374099731,
      "learning_rate": 1.0614732096783275e-05,
      "loss": 0.464,
      "step": 15775
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6030467748641968,
      "learning_rate": 1.0613715609200894e-05,
      "loss": 0.4359,
      "step": 15776
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6165473461151123,
      "learning_rate": 1.0612699115253296e-05,
      "loss": 0.5465,
      "step": 15777
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.497185230255127,
      "learning_rate": 1.0611682614951028e-05,
      "loss": 0.4142,
      "step": 15778
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.583884358406067,
      "learning_rate": 1.0610666108304636e-05,
      "loss": 0.4886,
      "step": 15779
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5797417163848877,
      "learning_rate": 1.0609649595324659e-05,
      "loss": 0.4401,
      "step": 15780
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5139669179916382,
      "learning_rate": 1.0608633076021641e-05,
      "loss": 0.4571,
      "step": 15781
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.521633267402649,
      "learning_rate": 1.0607616550406123e-05,
      "loss": 0.4353,
      "step": 15782
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5985198020935059,
      "learning_rate": 1.0606600018488653e-05,
      "loss": 0.4658,
      "step": 15783
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6262933015823364,
      "learning_rate": 1.0605583480279769e-05,
      "loss": 0.4376,
      "step": 15784
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6043956279754639,
      "learning_rate": 1.0604566935790013e-05,
      "loss": 0.5963,
      "step": 15785
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4922595024108887,
      "learning_rate": 1.0603550385029937e-05,
      "loss": 0.4175,
      "step": 15786
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5792107582092285,
      "learning_rate": 1.0602533828010074e-05,
      "loss": 0.4873,
      "step": 15787
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5344432592391968,
      "learning_rate": 1.0601517264740973e-05,
      "loss": 0.4308,
      "step": 15788
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7564749717712402,
      "learning_rate": 1.0600500695233176e-05,
      "loss": 0.9992,
      "step": 15789
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.527158498764038,
      "learning_rate": 1.0599484119497225e-05,
      "loss": 1.0166,
      "step": 15790
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5360651016235352,
      "learning_rate": 1.0598467537543666e-05,
      "loss": 0.4286,
      "step": 15791
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4927504062652588,
      "learning_rate": 1.0597450949383043e-05,
      "loss": 0.4256,
      "step": 15792
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6391112804412842,
      "learning_rate": 1.0596434355025895e-05,
      "loss": 0.4578,
      "step": 15793
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5554951429367065,
      "learning_rate": 1.0595417754482775e-05,
      "loss": 0.4169,
      "step": 15794
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.52668035030365,
      "learning_rate": 1.0594401147764216e-05,
      "loss": 0.4745,
      "step": 15795
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5168774127960205,
      "learning_rate": 1.0593384534880768e-05,
      "loss": 0.411,
      "step": 15796
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.540008783340454,
      "learning_rate": 1.0592367915842973e-05,
      "loss": 0.4474,
      "step": 15797
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5254583358764648,
      "learning_rate": 1.0591351290661376e-05,
      "loss": 0.4312,
      "step": 15798
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9243896007537842,
      "learning_rate": 1.0590334659346522e-05,
      "loss": 0.4902,
      "step": 15799
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5460155010223389,
      "learning_rate": 1.0589318021908952e-05,
      "loss": 0.4255,
      "step": 15800
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4883172512054443,
      "learning_rate": 1.0588301378359209e-05,
      "loss": 0.4811,
      "step": 15801
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.68085515499115,
      "learning_rate": 1.0587284728707846e-05,
      "loss": 0.4416,
      "step": 15802
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5426863431930542,
      "learning_rate": 1.0586268072965395e-05,
      "loss": 0.4666,
      "step": 15803
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5743659734725952,
      "learning_rate": 1.0585251411142413e-05,
      "loss": 0.4229,
      "step": 15804
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7817072868347168,
      "learning_rate": 1.0584234743249431e-05,
      "loss": 0.5121,
      "step": 15805
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5866862535476685,
      "learning_rate": 1.0583218069297004e-05,
      "loss": 0.4226,
      "step": 15806
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0139858722686768,
      "learning_rate": 1.0582201389295675e-05,
      "loss": 0.9178,
      "step": 15807
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5754886865615845,
      "learning_rate": 1.0581184703255982e-05,
      "loss": 1.0421,
      "step": 15808
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5349633693695068,
      "learning_rate": 1.0580168011188478e-05,
      "loss": 0.4333,
      "step": 15809
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.601535439491272,
      "learning_rate": 1.0579151313103704e-05,
      "loss": 0.4497,
      "step": 15810
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5758370161056519,
      "learning_rate": 1.05781346090122e-05,
      "loss": 0.4548,
      "step": 15811
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6915546655654907,
      "learning_rate": 1.0577117898924522e-05,
      "loss": 0.4266,
      "step": 15812
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5556435585021973,
      "learning_rate": 1.0576101182851204e-05,
      "loss": 0.4555,
      "step": 15813
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5796089172363281,
      "learning_rate": 1.0575084460802797e-05,
      "loss": 0.4217,
      "step": 15814
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7382959127426147,
      "learning_rate": 1.0574067732789845e-05,
      "loss": 0.9191,
      "step": 15815
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.607834815979004,
      "learning_rate": 1.0573050998822893e-05,
      "loss": 1.0821,
      "step": 15816
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.760902762413025,
      "learning_rate": 1.0572034258912484e-05,
      "loss": 0.5204,
      "step": 15817
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6564360857009888,
      "learning_rate": 1.0571017513069168e-05,
      "loss": 0.4305,
      "step": 15818
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7009929418563843,
      "learning_rate": 1.0570000761303483e-05,
      "loss": 0.481,
      "step": 15819
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5243951082229614,
      "learning_rate": 1.0568984003625983e-05,
      "loss": 0.4469,
      "step": 15820
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.588045597076416,
      "learning_rate": 1.0567967240047205e-05,
      "loss": 0.4868,
      "step": 15821
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.527193307876587,
      "learning_rate": 1.0566950470577704e-05,
      "loss": 0.4377,
      "step": 15822
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6663929224014282,
      "learning_rate": 1.0565933695228014e-05,
      "loss": 0.4792,
      "step": 15823
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6025934219360352,
      "learning_rate": 1.0564916914008691e-05,
      "loss": 0.4176,
      "step": 15824
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6672695875167847,
      "learning_rate": 1.0563900126930275e-05,
      "loss": 0.4829,
      "step": 15825
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5465463399887085,
      "learning_rate": 1.0562883334003313e-05,
      "loss": 0.418,
      "step": 15826
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9102672338485718,
      "learning_rate": 1.0561866535238349e-05,
      "loss": 1.0745,
      "step": 15827
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4410970211029053,
      "learning_rate": 1.0560849730645932e-05,
      "loss": 0.9901,
      "step": 15828
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6255927085876465,
      "learning_rate": 1.0559832920236605e-05,
      "loss": 0.4486,
      "step": 15829
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6661823987960815,
      "learning_rate": 1.055881610402092e-05,
      "loss": 0.4144,
      "step": 15830
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6864086389541626,
      "learning_rate": 1.0557799282009411e-05,
      "loss": 0.4832,
      "step": 15831
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5428320169448853,
      "learning_rate": 1.0556782454212637e-05,
      "loss": 0.4409,
      "step": 15832
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5697683095932007,
      "learning_rate": 1.0555765620641136e-05,
      "loss": 0.4579,
      "step": 15833
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5775271654129028,
      "learning_rate": 1.055474878130546e-05,
      "loss": 0.4425,
      "step": 15834
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6597542762756348,
      "learning_rate": 1.055373193621615e-05,
      "loss": 0.4506,
      "step": 15835
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5447336435317993,
      "learning_rate": 1.0552715085383753e-05,
      "loss": 0.4337,
      "step": 15836
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4982917308807373,
      "learning_rate": 1.0551698228818816e-05,
      "loss": 0.4872,
      "step": 15837
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5057923793792725,
      "learning_rate": 1.0550681366531892e-05,
      "loss": 0.4379,
      "step": 15838
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6401948928833008,
      "learning_rate": 1.0549664498533516e-05,
      "loss": 0.5021,
      "step": 15839
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5049041509628296,
      "learning_rate": 1.0548647624834245e-05,
      "loss": 0.3779,
      "step": 15840
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5938278436660767,
      "learning_rate": 1.0547630745444614e-05,
      "loss": 0.4546,
      "step": 15841
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4997649192810059,
      "learning_rate": 1.054661386037518e-05,
      "loss": 0.403,
      "step": 15842
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5511727333068848,
      "learning_rate": 1.0545596969636489e-05,
      "loss": 0.453,
      "step": 15843
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5207765102386475,
      "learning_rate": 1.054458007323908e-05,
      "loss": 0.423,
      "step": 15844
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5469961166381836,
      "learning_rate": 1.054356317119351e-05,
      "loss": 0.4547,
      "step": 15845
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5512477159500122,
      "learning_rate": 1.0542546263510319e-05,
      "loss": 0.4001,
      "step": 15846
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5418275594711304,
      "learning_rate": 1.0541529350200051e-05,
      "loss": 0.5038,
      "step": 15847
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5340397357940674,
      "learning_rate": 1.0540512431273264e-05,
      "loss": 0.4315,
      "step": 15848
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.521270990371704,
      "learning_rate": 1.0539495506740494e-05,
      "loss": 0.4428,
      "step": 15849
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5586551427841187,
      "learning_rate": 1.0538478576612296e-05,
      "loss": 0.4515,
      "step": 15850
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.617961049079895,
      "learning_rate": 1.0537461640899213e-05,
      "loss": 0.5263,
      "step": 15851
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.582139492034912,
      "learning_rate": 1.0536444699611794e-05,
      "loss": 0.3961,
      "step": 15852
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6008402109146118,
      "learning_rate": 1.0535427752760585e-05,
      "loss": 0.5265,
      "step": 15853
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6899077892303467,
      "learning_rate": 1.0534410800356134e-05,
      "loss": 0.4091,
      "step": 15854
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5648889541625977,
      "learning_rate": 1.0533393842408988e-05,
      "loss": 0.4803,
      "step": 15855
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5962105989456177,
      "learning_rate": 1.0532376878929698e-05,
      "loss": 0.4572,
      "step": 15856
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5200899839401245,
      "learning_rate": 1.0531359909928803e-05,
      "loss": 0.4403,
      "step": 15857
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4769470691680908,
      "learning_rate": 1.0530342935416862e-05,
      "loss": 0.412,
      "step": 15858
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5492502450942993,
      "learning_rate": 1.052932595540441e-05,
      "loss": 0.5012,
      "step": 15859
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6019936800003052,
      "learning_rate": 1.0528308969902006e-05,
      "loss": 0.4298,
      "step": 15860
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6545660495758057,
      "learning_rate": 1.0527291978920194e-05,
      "loss": 0.5089,
      "step": 15861
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6377211809158325,
      "learning_rate": 1.0526274982469518e-05,
      "loss": 0.4596,
      "step": 15862
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.566298484802246,
      "learning_rate": 1.052525798056053e-05,
      "loss": 0.5292,
      "step": 15863
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5911235809326172,
      "learning_rate": 1.0524240973203776e-05,
      "loss": 0.4556,
      "step": 15864
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5378130674362183,
      "learning_rate": 1.0523223960409806e-05,
      "loss": 0.4539,
      "step": 15865
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.602730393409729,
      "learning_rate": 1.0522206942189168e-05,
      "loss": 0.4421,
      "step": 15866
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6073188781738281,
      "learning_rate": 1.0521189918552406e-05,
      "loss": 0.4577,
      "step": 15867
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5352239608764648,
      "learning_rate": 1.0520172889510072e-05,
      "loss": 0.4795,
      "step": 15868
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.584794044494629,
      "learning_rate": 1.0519155855072716e-05,
      "loss": 0.4879,
      "step": 15869
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5409811735153198,
      "learning_rate": 1.051813881525088e-05,
      "loss": 0.4232,
      "step": 15870
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5463716983795166,
      "learning_rate": 1.0517121770055119e-05,
      "loss": 0.4752,
      "step": 15871
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.494401454925537,
      "learning_rate": 1.0516104719495976e-05,
      "loss": 0.4444,
      "step": 15872
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5312438011169434,
      "learning_rate": 1.0515087663584e-05,
      "loss": 0.4541,
      "step": 15873
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.58082914352417,
      "learning_rate": 1.0514070602329746e-05,
      "loss": 0.4742,
      "step": 15874
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.51201331615448,
      "learning_rate": 1.0513053535743756e-05,
      "loss": 0.4979,
      "step": 15875
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.576432228088379,
      "learning_rate": 1.0512036463836579e-05,
      "loss": 0.4416,
      "step": 15876
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.591881513595581,
      "learning_rate": 1.0511019386618767e-05,
      "loss": 0.472,
      "step": 15877
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.487153172492981,
      "learning_rate": 1.0510002304100866e-05,
      "loss": 0.4118,
      "step": 15878
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6241384744644165,
      "learning_rate": 1.0508985216293425e-05,
      "loss": 0.4908,
      "step": 15879
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5822004079818726,
      "learning_rate": 1.0507968123206996e-05,
      "loss": 0.4209,
      "step": 15880
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.555496335029602,
      "learning_rate": 1.0506951024852123e-05,
      "loss": 0.4877,
      "step": 15881
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5278563499450684,
      "learning_rate": 1.0505933921239358e-05,
      "loss": 0.4174,
      "step": 15882
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.548699140548706,
      "learning_rate": 1.0504916812379246e-05,
      "loss": 0.4701,
      "step": 15883
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.512041449546814,
      "learning_rate": 1.0503899698282346e-05,
      "loss": 0.4352,
      "step": 15884
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5342323780059814,
      "learning_rate": 1.0502882578959193e-05,
      "loss": 0.493,
      "step": 15885
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6009178161621094,
      "learning_rate": 1.0501865454420346e-05,
      "loss": 0.451,
      "step": 15886
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.506298542022705,
      "learning_rate": 1.0500848324676354e-05,
      "loss": 0.4857,
      "step": 15887
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4959311485290527,
      "learning_rate": 1.049983118973776e-05,
      "loss": 0.4233,
      "step": 15888
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5468239784240723,
      "learning_rate": 1.0498814049615123e-05,
      "loss": 0.4407,
      "step": 15889
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6003786325454712,
      "learning_rate": 1.0497796904318982e-05,
      "loss": 0.4774,
      "step": 15890
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5232141017913818,
      "learning_rate": 1.049677975385989e-05,
      "loss": 0.4772,
      "step": 15891
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4829388856887817,
      "learning_rate": 1.0495762598248405e-05,
      "loss": 0.4088,
      "step": 15892
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4924311637878418,
      "learning_rate": 1.0494745437495061e-05,
      "loss": 0.4578,
      "step": 15893
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5059289932250977,
      "learning_rate": 1.0493728271610419e-05,
      "loss": 0.4249,
      "step": 15894
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5984519720077515,
      "learning_rate": 1.0492711100605026e-05,
      "loss": 0.4956,
      "step": 15895
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4958088397979736,
      "learning_rate": 1.049169392448943e-05,
      "loss": 0.4109,
      "step": 15896
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6224873065948486,
      "learning_rate": 1.0490676743274181e-05,
      "loss": 0.4738,
      "step": 15897
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6287997961044312,
      "learning_rate": 1.0489659556969828e-05,
      "loss": 0.4711,
      "step": 15898
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.531955599784851,
      "learning_rate": 1.048864236558693e-05,
      "loss": 0.4348,
      "step": 15899
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.591198205947876,
      "learning_rate": 1.0487625169136023e-05,
      "loss": 0.4469,
      "step": 15900
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5845839977264404,
      "learning_rate": 1.0486607967627662e-05,
      "loss": 0.4808,
      "step": 15901
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5882718563079834,
      "learning_rate": 1.0485590761072401e-05,
      "loss": 0.414,
      "step": 15902
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5832273960113525,
      "learning_rate": 1.0484573549480787e-05,
      "loss": 0.4607,
      "step": 15903
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6849005222320557,
      "learning_rate": 1.048355633286337e-05,
      "loss": 0.4594,
      "step": 15904
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5299270153045654,
      "learning_rate": 1.0482539111230703e-05,
      "loss": 0.4351,
      "step": 15905
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5710384845733643,
      "learning_rate": 1.048152188459333e-05,
      "loss": 0.3929,
      "step": 15906
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.613855004310608,
      "learning_rate": 1.0480504652961809e-05,
      "loss": 0.5147,
      "step": 15907
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4525697231292725,
      "learning_rate": 1.0479487416346682e-05,
      "loss": 0.394,
      "step": 15908
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4561361074447632,
      "learning_rate": 1.0478470174758505e-05,
      "loss": 0.4469,
      "step": 15909
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.530693769454956,
      "learning_rate": 1.047745292820783e-05,
      "loss": 0.42,
      "step": 15910
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4959088563919067,
      "learning_rate": 1.0476435676705202e-05,
      "loss": 0.4653,
      "step": 15911
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4583529233932495,
      "learning_rate": 1.0475418420261174e-05,
      "loss": 0.4012,
      "step": 15912
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7186535596847534,
      "learning_rate": 1.0474401158886298e-05,
      "loss": 1.0422,
      "step": 15913
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5758978128433228,
      "learning_rate": 1.0473383892591123e-05,
      "loss": 1.0796,
      "step": 15914
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.686832308769226,
      "learning_rate": 1.0472366621386201e-05,
      "loss": 0.4659,
      "step": 15915
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.714810848236084,
      "learning_rate": 1.047134934528208e-05,
      "loss": 0.4441,
      "step": 15916
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6735360622406006,
      "learning_rate": 1.0470332064289314e-05,
      "loss": 0.4935,
      "step": 15917
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5029553174972534,
      "learning_rate": 1.0469314778418452e-05,
      "loss": 0.3961,
      "step": 15918
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8000386953353882,
      "learning_rate": 1.0468297487680044e-05,
      "loss": 0.4641,
      "step": 15919
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5606859922409058,
      "learning_rate": 1.0467280192084644e-05,
      "loss": 0.437,
      "step": 15920
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5333648920059204,
      "learning_rate": 1.0466262891642799e-05,
      "loss": 0.4307,
      "step": 15921
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5172724723815918,
      "learning_rate": 1.0465245586365067e-05,
      "loss": 0.4455,
      "step": 15922
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6481224298477173,
      "learning_rate": 1.046422827626199e-05,
      "loss": 0.5184,
      "step": 15923
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5477570295333862,
      "learning_rate": 1.0463210961344124e-05,
      "loss": 0.4239,
      "step": 15924
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5915429592132568,
      "learning_rate": 1.046219364162202e-05,
      "loss": 0.4551,
      "step": 15925
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5231471061706543,
      "learning_rate": 1.0461176317106227e-05,
      "loss": 0.4198,
      "step": 15926
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6255284547805786,
      "learning_rate": 1.0460158987807301e-05,
      "loss": 0.4829,
      "step": 15927
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5564137697219849,
      "learning_rate": 1.0459141653735792e-05,
      "loss": 0.4299,
      "step": 15928
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5926308631896973,
      "learning_rate": 1.0458124314902247e-05,
      "loss": 0.4802,
      "step": 15929
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5300832986831665,
      "learning_rate": 1.0457106971317219e-05,
      "loss": 0.4177,
      "step": 15930
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6337647438049316,
      "learning_rate": 1.0456089622991264e-05,
      "loss": 0.4577,
      "step": 15931
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4798057079315186,
      "learning_rate": 1.0455072269934927e-05,
      "loss": 0.4275,
      "step": 15932
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.587022066116333,
      "learning_rate": 1.0454054912158766e-05,
      "loss": 0.4579,
      "step": 15933
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5839414596557617,
      "learning_rate": 1.0453037549673326e-05,
      "loss": 0.4496,
      "step": 15934
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6195685863494873,
      "learning_rate": 1.0452020182489165e-05,
      "loss": 0.5033,
      "step": 15935
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4839006662368774,
      "learning_rate": 1.045100281061683e-05,
      "loss": 0.4071,
      "step": 15936
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5220924615859985,
      "learning_rate": 1.0449985434066878e-05,
      "loss": 0.4558,
      "step": 15937
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5034235715866089,
      "learning_rate": 1.0448968052849855e-05,
      "loss": 0.42,
      "step": 15938
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5877705812454224,
      "learning_rate": 1.0447950666976317e-05,
      "loss": 0.4655,
      "step": 15939
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5278061628341675,
      "learning_rate": 1.0446933276456812e-05,
      "loss": 0.4271,
      "step": 15940
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7833847999572754,
      "learning_rate": 1.0445915881301895e-05,
      "loss": 1.0505,
      "step": 15941
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4729057550430298,
      "learning_rate": 1.0444898481522117e-05,
      "loss": 1.0866,
      "step": 15942
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6743978261947632,
      "learning_rate": 1.0443881077128032e-05,
      "loss": 0.5324,
      "step": 15943
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4979677200317383,
      "learning_rate": 1.0442863668130187e-05,
      "loss": 0.3941,
      "step": 15944
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7388808727264404,
      "learning_rate": 1.0441846254539142e-05,
      "loss": 0.505,
      "step": 15945
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5522682666778564,
      "learning_rate": 1.044082883636544e-05,
      "loss": 0.4154,
      "step": 15946
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5191360712051392,
      "learning_rate": 1.043981141361964e-05,
      "loss": 0.5126,
      "step": 15947
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5490130186080933,
      "learning_rate": 1.0438793986312294e-05,
      "loss": 0.4284,
      "step": 15948
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5126237869262695,
      "learning_rate": 1.0437776554453953e-05,
      "loss": 0.4304,
      "step": 15949
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6676125526428223,
      "learning_rate": 1.0436759118055164e-05,
      "loss": 0.4086,
      "step": 15950
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5102949142456055,
      "learning_rate": 1.0435741677126493e-05,
      "loss": 0.4324,
      "step": 15951
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5143545866012573,
      "learning_rate": 1.0434724231678477e-05,
      "loss": 0.4264,
      "step": 15952
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.55595064163208,
      "learning_rate": 1.0433706781721677e-05,
      "loss": 0.4804,
      "step": 15953
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5561567544937134,
      "learning_rate": 1.0432689327266645e-05,
      "loss": 0.4017,
      "step": 15954
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5122661590576172,
      "learning_rate": 1.0431671868323933e-05,
      "loss": 0.4548,
      "step": 15955
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.568289041519165,
      "learning_rate": 1.0430654404904093e-05,
      "loss": 0.4229,
      "step": 15956
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5228177309036255,
      "learning_rate": 1.042963693701768e-05,
      "loss": 0.4489,
      "step": 15957
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.460304617881775,
      "learning_rate": 1.0428619464675242e-05,
      "loss": 0.4003,
      "step": 15958
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7082473039627075,
      "learning_rate": 1.0427601987887336e-05,
      "loss": 0.4392,
      "step": 15959
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4901243448257446,
      "learning_rate": 1.0426584506664511e-05,
      "loss": 0.425,
      "step": 15960
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5487674474716187,
      "learning_rate": 1.0425567021017329e-05,
      "loss": 0.4651,
      "step": 15961
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5088955163955688,
      "learning_rate": 1.0424549530956332e-05,
      "loss": 0.4099,
      "step": 15962
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5700976848602295,
      "learning_rate": 1.0423532036492077e-05,
      "loss": 0.4681,
      "step": 15963
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5519344806671143,
      "learning_rate": 1.042251453763512e-05,
      "loss": 0.4276,
      "step": 15964
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5754576921463013,
      "learning_rate": 1.042149703439601e-05,
      "loss": 0.4761,
      "step": 15965
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5206862688064575,
      "learning_rate": 1.0420479526785301e-05,
      "loss": 0.412,
      "step": 15966
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5940148830413818,
      "learning_rate": 1.0419462014813548e-05,
      "loss": 0.4304,
      "step": 15967
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6500530242919922,
      "learning_rate": 1.0418444498491302e-05,
      "loss": 0.45,
      "step": 15968
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8703322410583496,
      "learning_rate": 1.0417426977829121e-05,
      "loss": 0.4778,
      "step": 15969
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.49501371383667,
      "learning_rate": 1.0416409452837551e-05,
      "loss": 0.41,
      "step": 15970
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5912967920303345,
      "learning_rate": 1.041539192352715e-05,
      "loss": 0.4962,
      "step": 15971
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.600917935371399,
      "learning_rate": 1.041437438990847e-05,
      "loss": 0.4091,
      "step": 15972
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5010170936584473,
      "learning_rate": 1.0413356851992067e-05,
      "loss": 0.4572,
      "step": 15973
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4831057786941528,
      "learning_rate": 1.041233930978849e-05,
      "loss": 0.4003,
      "step": 15974
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5780644416809082,
      "learning_rate": 1.0411321763308298e-05,
      "loss": 0.4535,
      "step": 15975
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4878629446029663,
      "learning_rate": 1.0410304212562039e-05,
      "loss": 0.3934,
      "step": 15976
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4971023797988892,
      "learning_rate": 1.0409286657560271e-05,
      "loss": 0.4479,
      "step": 15977
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6702758073806763,
      "learning_rate": 1.0408269098313544e-05,
      "loss": 0.4508,
      "step": 15978
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4881141185760498,
      "learning_rate": 1.0407251534832416e-05,
      "loss": 0.4462,
      "step": 15979
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.47342050075531,
      "learning_rate": 1.0406233967127434e-05,
      "loss": 0.4119,
      "step": 15980
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.593358039855957,
      "learning_rate": 1.0405216395209161e-05,
      "loss": 0.5037,
      "step": 15981
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.525811791419983,
      "learning_rate": 1.0404198819088143e-05,
      "loss": 0.4078,
      "step": 15982
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5658776760101318,
      "learning_rate": 1.0403181238774939e-05,
      "loss": 0.437,
      "step": 15983
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.462345838546753,
      "learning_rate": 1.04021636542801e-05,
      "loss": 0.418,
      "step": 15984
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5626946687698364,
      "learning_rate": 1.0401146065614181e-05,
      "loss": 0.4801,
      "step": 15985
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4482042789459229,
      "learning_rate": 1.0400128472787734e-05,
      "loss": 0.3863,
      "step": 15986
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5506250858306885,
      "learning_rate": 1.0399110875811318e-05,
      "loss": 0.5029,
      "step": 15987
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5724846124649048,
      "learning_rate": 1.039809327469548e-05,
      "loss": 0.4502,
      "step": 15988
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5573780536651611,
      "learning_rate": 1.039707566945078e-05,
      "loss": 0.5002,
      "step": 15989
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4768176078796387,
      "learning_rate": 1.0396058060087772e-05,
      "loss": 0.3875,
      "step": 15990
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4585773944854736,
      "learning_rate": 1.039504044661701e-05,
      "loss": 0.4062,
      "step": 15991
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.554050326347351,
      "learning_rate": 1.0394022829049041e-05,
      "loss": 0.4068,
      "step": 15992
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5191402435302734,
      "learning_rate": 1.0393005207394429e-05,
      "loss": 0.4871,
      "step": 15993
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.500893235206604,
      "learning_rate": 1.0391987581663723e-05,
      "loss": 0.3927,
      "step": 15994
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.592987298965454,
      "learning_rate": 1.0390969951867482e-05,
      "loss": 0.4993,
      "step": 15995
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5657851696014404,
      "learning_rate": 1.0389952318016254e-05,
      "loss": 0.4447,
      "step": 15996
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6525819301605225,
      "learning_rate": 1.0388934680120602e-05,
      "loss": 0.424,
      "step": 15997
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6544842720031738,
      "learning_rate": 1.0387917038191068e-05,
      "loss": 0.4573,
      "step": 15998
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5794357061386108,
      "learning_rate": 1.038689939223822e-05,
      "loss": 0.4634,
      "step": 15999
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5928714275360107,
      "learning_rate": 1.0385881742272603e-05,
      "loss": 0.4298,
      "step": 16000
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5061253309249878,
      "learning_rate": 1.0384864088304776e-05,
      "loss": 0.4621,
      "step": 16001
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5297719240188599,
      "learning_rate": 1.0383846430345295e-05,
      "loss": 0.4271,
      "step": 16002
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4386711120605469,
      "learning_rate": 1.0382828768404712e-05,
      "loss": 0.4501,
      "step": 16003
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4923913478851318,
      "learning_rate": 1.038181110249358e-05,
      "loss": 0.4203,
      "step": 16004
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6487294435501099,
      "learning_rate": 1.038079343262246e-05,
      "loss": 0.4766,
      "step": 16005
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5597840547561646,
      "learning_rate": 1.0379775758801898e-05,
      "loss": 0.4206,
      "step": 16006
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6970248222351074,
      "learning_rate": 1.0378758081042459e-05,
      "loss": 0.4933,
      "step": 16007
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.509756088256836,
      "learning_rate": 1.037774039935469e-05,
      "loss": 0.4084,
      "step": 16008
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6252065896987915,
      "learning_rate": 1.0376722713749152e-05,
      "loss": 0.4779,
      "step": 16009
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.513543963432312,
      "learning_rate": 1.0375705024236396e-05,
      "loss": 0.3943,
      "step": 16010
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5368417501449585,
      "learning_rate": 1.0374687330826978e-05,
      "loss": 0.4519,
      "step": 16011
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6485322713851929,
      "learning_rate": 1.037366963353145e-05,
      "loss": 0.4532,
      "step": 16012
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5783989429473877,
      "learning_rate": 1.0372651932360376e-05,
      "loss": 0.4546,
      "step": 16013
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6634901762008667,
      "learning_rate": 1.03716342273243e-05,
      "loss": 0.4478,
      "step": 16014
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6016404628753662,
      "learning_rate": 1.0370616518433789e-05,
      "loss": 0.4845,
      "step": 16015
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5875035524368286,
      "learning_rate": 1.0369598805699385e-05,
      "loss": 0.4237,
      "step": 16016
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6488057374954224,
      "learning_rate": 1.0368581089131655e-05,
      "loss": 0.5076,
      "step": 16017
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5406612157821655,
      "learning_rate": 1.036756336874115e-05,
      "loss": 0.3962,
      "step": 16018
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5711560249328613,
      "learning_rate": 1.0366545644538423e-05,
      "loss": 0.4258,
      "step": 16019
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4992494583129883,
      "learning_rate": 1.0365527916534033e-05,
      "loss": 0.4137,
      "step": 16020
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.567627191543579,
      "learning_rate": 1.0364510184738534e-05,
      "loss": 0.4726,
      "step": 16021
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.596149206161499,
      "learning_rate": 1.0363492449162478e-05,
      "loss": 0.4309,
      "step": 16022
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.533592700958252,
      "learning_rate": 1.036247470981643e-05,
      "loss": 0.475,
      "step": 16023
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5245815515518188,
      "learning_rate": 1.0361456966710934e-05,
      "loss": 0.4103,
      "step": 16024
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5946987867355347,
      "learning_rate": 1.0360439219856554e-05,
      "loss": 0.4414,
      "step": 16025
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5589758157730103,
      "learning_rate": 1.0359421469263844e-05,
      "loss": 0.406,
      "step": 16026
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.644511103630066,
      "learning_rate": 1.0358403714943357e-05,
      "loss": 0.4748,
      "step": 16027
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5151540040969849,
      "learning_rate": 1.035738595690565e-05,
      "loss": 0.4008,
      "step": 16028
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8331252336502075,
      "learning_rate": 1.0356368195161282e-05,
      "loss": 1.0898,
      "step": 16029
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5749406814575195,
      "learning_rate": 1.0355350429720801e-05,
      "loss": 0.9947,
      "step": 16030
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5387006998062134,
      "learning_rate": 1.0354332660594773e-05,
      "loss": 0.4666,
      "step": 16031
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5814318656921387,
      "learning_rate": 1.0353314887793744e-05,
      "loss": 0.4449,
      "step": 16032
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6382604837417603,
      "learning_rate": 1.035229711132828e-05,
      "loss": 0.4892,
      "step": 16033
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5210013389587402,
      "learning_rate": 1.0351279331208925e-05,
      "loss": 0.4036,
      "step": 16034
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.571975588798523,
      "learning_rate": 1.0350261547446245e-05,
      "loss": 0.4822,
      "step": 16035
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.474632978439331,
      "learning_rate": 1.0349243760050793e-05,
      "loss": 0.4048,
      "step": 16036
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.088308334350586,
      "learning_rate": 1.0348225969033126e-05,
      "loss": 0.5003,
      "step": 16037
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5055897235870361,
      "learning_rate": 1.0347208174403798e-05,
      "loss": 0.4109,
      "step": 16038
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6035236120224,
      "learning_rate": 1.0346190376173366e-05,
      "loss": 0.4967,
      "step": 16039
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5284584760665894,
      "learning_rate": 1.0345172574352383e-05,
      "loss": 0.4149,
      "step": 16040
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5344070196151733,
      "learning_rate": 1.0344154768951413e-05,
      "loss": 0.5532,
      "step": 16041
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.401911973953247,
      "learning_rate": 1.0343136959981004e-05,
      "loss": 0.3778,
      "step": 16042
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5072022676467896,
      "learning_rate": 1.034211914745172e-05,
      "loss": 0.471,
      "step": 16043
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4891633987426758,
      "learning_rate": 1.034110133137411e-05,
      "loss": 0.3844,
      "step": 16044
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5912699699401855,
      "learning_rate": 1.0340083511758734e-05,
      "loss": 0.4576,
      "step": 16045
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5525462627410889,
      "learning_rate": 1.033906568861615e-05,
      "loss": 0.4091,
      "step": 16046
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5555893182754517,
      "learning_rate": 1.0338047861956912e-05,
      "loss": 0.4717,
      "step": 16047
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5901907682418823,
      "learning_rate": 1.0337030031791576e-05,
      "loss": 0.432,
      "step": 16048
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5278394222259521,
      "learning_rate": 1.03360121981307e-05,
      "loss": 0.4708,
      "step": 16049
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5413389205932617,
      "learning_rate": 1.0334994360984841e-05,
      "loss": 0.4685,
      "step": 16050
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5633584260940552,
      "learning_rate": 1.0333976520364556e-05,
      "loss": 0.4232,
      "step": 16051
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5501338243484497,
      "learning_rate": 1.0332958676280396e-05,
      "loss": 0.4239,
      "step": 16052
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.507191777229309,
      "learning_rate": 1.0331940828742925e-05,
      "loss": 0.4908,
      "step": 16053
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6096197366714478,
      "learning_rate": 1.03309229777627e-05,
      "loss": 0.4433,
      "step": 16054
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5200048685073853,
      "learning_rate": 1.032990512335027e-05,
      "loss": 0.4501,
      "step": 16055
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.458042860031128,
      "learning_rate": 1.0328887265516197e-05,
      "loss": 0.3969,
      "step": 16056
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.512619137763977,
      "learning_rate": 1.0327869404271038e-05,
      "loss": 0.4354,
      "step": 16057
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.513566017150879,
      "learning_rate": 1.0326851539625346e-05,
      "loss": 0.445,
      "step": 16058
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6025410890579224,
      "learning_rate": 1.0325833671589687e-05,
      "loss": 0.446,
      "step": 16059
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4963946342468262,
      "learning_rate": 1.0324815800174606e-05,
      "loss": 0.4267,
      "step": 16060
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4779995679855347,
      "learning_rate": 1.0323797925390669e-05,
      "loss": 0.4258,
      "step": 16061
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.521350622177124,
      "learning_rate": 1.0322780047248429e-05,
      "loss": 0.4086,
      "step": 16062
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.644067645072937,
      "learning_rate": 1.0321762165758445e-05,
      "loss": 0.4903,
      "step": 16063
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5072890520095825,
      "learning_rate": 1.032074428093127e-05,
      "loss": 0.3843,
      "step": 16064
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6191799640655518,
      "learning_rate": 1.0319726392777467e-05,
      "loss": 0.4666,
      "step": 16065
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6118992567062378,
      "learning_rate": 1.0318708501307587e-05,
      "loss": 0.4241,
      "step": 16066
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.537036418914795,
      "learning_rate": 1.0317690606532193e-05,
      "loss": 0.4294,
      "step": 16067
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4508183002471924,
      "learning_rate": 1.0316672708461837e-05,
      "loss": 0.3957,
      "step": 16068
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5378690958023071,
      "learning_rate": 1.0315654807107078e-05,
      "loss": 0.4751,
      "step": 16069
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6914279460906982,
      "learning_rate": 1.0314636902478477e-05,
      "loss": 0.4171,
      "step": 16070
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5349799394607544,
      "learning_rate": 1.0313618994586588e-05,
      "loss": 0.4866,
      "step": 16071
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4982401132583618,
      "learning_rate": 1.0312601083441967e-05,
      "loss": 0.4066,
      "step": 16072
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6070972681045532,
      "learning_rate": 1.0311583169055172e-05,
      "loss": 0.4583,
      "step": 16073
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.611311674118042,
      "learning_rate": 1.0310565251436763e-05,
      "loss": 0.4322,
      "step": 16074
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5286316871643066,
      "learning_rate": 1.0309547330597295e-05,
      "loss": 0.4464,
      "step": 16075
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5277607440948486,
      "learning_rate": 1.0308529406547325e-05,
      "loss": 0.4311,
      "step": 16076
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5165570974349976,
      "learning_rate": 1.0307511479297414e-05,
      "loss": 0.4607,
      "step": 16077
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.549363374710083,
      "learning_rate": 1.0306493548858117e-05,
      "loss": 0.4085,
      "step": 16078
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5788698196411133,
      "learning_rate": 1.030547561523999e-05,
      "loss": 0.4707,
      "step": 16079
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.462226390838623,
      "learning_rate": 1.0304457678453592e-05,
      "loss": 0.3866,
      "step": 16080
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6282638311386108,
      "learning_rate": 1.0303439738509484e-05,
      "loss": 0.4936,
      "step": 16081
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.527817726135254,
      "learning_rate": 1.0302421795418218e-05,
      "loss": 0.4009,
      "step": 16082
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.690529465675354,
      "learning_rate": 1.0301403849190356e-05,
      "loss": 0.4869,
      "step": 16083
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5131902694702148,
      "learning_rate": 1.0300385899836454e-05,
      "loss": 0.403,
      "step": 16084
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.525105357170105,
      "learning_rate": 1.0299367947367071e-05,
      "loss": 0.439,
      "step": 16085
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5551766157150269,
      "learning_rate": 1.029834999179276e-05,
      "loss": 0.4238,
      "step": 16086
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5025287866592407,
      "learning_rate": 1.0297332033124085e-05,
      "loss": 0.4305,
      "step": 16087
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6924772262573242,
      "learning_rate": 1.0296314071371601e-05,
      "loss": 0.415,
      "step": 16088
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.559532880783081,
      "learning_rate": 1.0295296106545867e-05,
      "loss": 0.5017,
      "step": 16089
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4612973928451538,
      "learning_rate": 1.0294278138657439e-05,
      "loss": 0.4214,
      "step": 16090
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6610490083694458,
      "learning_rate": 1.0293260167716876e-05,
      "loss": 0.5068,
      "step": 16091
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5165518522262573,
      "learning_rate": 1.0292242193734739e-05,
      "loss": 0.3731,
      "step": 16092
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6473751068115234,
      "learning_rate": 1.029122421672158e-05,
      "loss": 0.4509,
      "step": 16093
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.622632384300232,
      "learning_rate": 1.0290206236687958e-05,
      "loss": 0.4546,
      "step": 16094
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.515760898590088,
      "learning_rate": 1.028918825364444e-05,
      "loss": 0.4515,
      "step": 16095
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4820095300674438,
      "learning_rate": 1.0288170267601571e-05,
      "loss": 0.4071,
      "step": 16096
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5969949960708618,
      "learning_rate": 1.0287152278569918e-05,
      "loss": 0.4582,
      "step": 16097
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5734487771987915,
      "learning_rate": 1.028613428656004e-05,
      "loss": 0.3904,
      "step": 16098
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.556010127067566,
      "learning_rate": 1.0285116291582486e-05,
      "loss": 0.4827,
      "step": 16099
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5248397588729858,
      "learning_rate": 1.0284098293647823e-05,
      "loss": 0.4075,
      "step": 16100
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5438520908355713,
      "learning_rate": 1.0283080292766608e-05,
      "loss": 0.4528,
      "step": 16101
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5371649265289307,
      "learning_rate": 1.0282062288949395e-05,
      "loss": 0.4008,
      "step": 16102
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5861274003982544,
      "learning_rate": 1.0281044282206747e-05,
      "loss": 0.4676,
      "step": 16103
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5418189764022827,
      "learning_rate": 1.0280026272549216e-05,
      "loss": 0.4464,
      "step": 16104
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.50095796585083,
      "learning_rate": 1.027900825998737e-05,
      "loss": 0.4798,
      "step": 16105
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5861095190048218,
      "learning_rate": 1.027799024453176e-05,
      "loss": 0.4524,
      "step": 16106
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4593430757522583,
      "learning_rate": 1.0276972226192947e-05,
      "loss": 0.4143,
      "step": 16107
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5245741605758667,
      "learning_rate": 1.027595420498149e-05,
      "loss": 0.4156,
      "step": 16108
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5752733945846558,
      "learning_rate": 1.0274936180907946e-05,
      "loss": 0.5355,
      "step": 16109
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4951187372207642,
      "learning_rate": 1.0273918153982874e-05,
      "loss": 0.4445,
      "step": 16110
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.570063829421997,
      "learning_rate": 1.0272900124216834e-05,
      "loss": 0.5031,
      "step": 16111
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4001034498214722,
      "learning_rate": 1.027188209162038e-05,
      "loss": 0.3996,
      "step": 16112
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5094268321990967,
      "learning_rate": 1.0270864056204077e-05,
      "loss": 0.4845,
      "step": 16113
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.505048155784607,
      "learning_rate": 1.0269846017978478e-05,
      "loss": 0.4376,
      "step": 16114
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6008390188217163,
      "learning_rate": 1.0268827976954145e-05,
      "loss": 0.5044,
      "step": 16115
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5306569337844849,
      "learning_rate": 1.0267809933141637e-05,
      "loss": 0.4051,
      "step": 16116
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6125807762145996,
      "learning_rate": 1.0266791886551513e-05,
      "loss": 0.4918,
      "step": 16117
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4645798206329346,
      "learning_rate": 1.0265773837194328e-05,
      "loss": 0.4391,
      "step": 16118
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.475804090499878,
      "learning_rate": 1.0264755785080645e-05,
      "loss": 0.4847,
      "step": 16119
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4417022466659546,
      "learning_rate": 1.026373773022102e-05,
      "loss": 0.4105,
      "step": 16120
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5628067255020142,
      "learning_rate": 1.0262719672626015e-05,
      "loss": 0.4979,
      "step": 16121
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5608502626419067,
      "learning_rate": 1.0261701612306184e-05,
      "loss": 0.3935,
      "step": 16122
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5263803005218506,
      "learning_rate": 1.0260683549272091e-05,
      "loss": 0.4331,
      "step": 16123
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5405505895614624,
      "learning_rate": 1.0259665483534291e-05,
      "loss": 0.4244,
      "step": 16124
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5984071493148804,
      "learning_rate": 1.0258647415103347e-05,
      "loss": 0.5063,
      "step": 16125
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5635676383972168,
      "learning_rate": 1.0257629343989811e-05,
      "loss": 0.4089,
      "step": 16126
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.653881311416626,
      "learning_rate": 1.0256611270204253e-05,
      "loss": 0.537,
      "step": 16127
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.539055585861206,
      "learning_rate": 1.0255593193757221e-05,
      "loss": 0.4094,
      "step": 16128
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7708059549331665,
      "learning_rate": 1.0254575114659282e-05,
      "loss": 0.5102,
      "step": 16129
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7332618236541748,
      "learning_rate": 1.025355703292099e-05,
      "loss": 0.4081,
      "step": 16130
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6170917749404907,
      "learning_rate": 1.0252538948552908e-05,
      "loss": 0.5114,
      "step": 16131
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6515371799468994,
      "learning_rate": 1.0251520861565591e-05,
      "loss": 0.4446,
      "step": 16132
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6303211450576782,
      "learning_rate": 1.0250502771969602e-05,
      "loss": 0.5279,
      "step": 16133
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5453523397445679,
      "learning_rate": 1.02494846797755e-05,
      "loss": 0.4215,
      "step": 16134
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5064244270324707,
      "learning_rate": 1.0248466584993839e-05,
      "loss": 0.4793,
      "step": 16135
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7198344469070435,
      "learning_rate": 1.0247448487635185e-05,
      "loss": 0.4335,
      "step": 16136
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5476759672164917,
      "learning_rate": 1.0246430387710093e-05,
      "loss": 0.5371,
      "step": 16137
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.522934913635254,
      "learning_rate": 1.0245412285229124e-05,
      "loss": 0.403,
      "step": 16138
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6454228162765503,
      "learning_rate": 1.0244394180202837e-05,
      "loss": 0.5247,
      "step": 16139
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.484154224395752,
      "learning_rate": 1.0243376072641793e-05,
      "loss": 0.4071,
      "step": 16140
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9531570672988892,
      "learning_rate": 1.024235796255655e-05,
      "loss": 1.0249,
      "step": 16141
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5069530010223389,
      "learning_rate": 1.0241339849957665e-05,
      "loss": 0.9986,
      "step": 16142
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6284788846969604,
      "learning_rate": 1.02403217348557e-05,
      "loss": 0.4895,
      "step": 16143
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6714560985565186,
      "learning_rate": 1.023930361726122e-05,
      "loss": 0.4548,
      "step": 16144
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6680296659469604,
      "learning_rate": 1.0238285497184773e-05,
      "loss": 0.5349,
      "step": 16145
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5881757736206055,
      "learning_rate": 1.0237267374636925e-05,
      "loss": 0.4153,
      "step": 16146
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5037343502044678,
      "learning_rate": 1.0236249249628236e-05,
      "loss": 0.48,
      "step": 16147
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5709714889526367,
      "learning_rate": 1.0235231122169264e-05,
      "loss": 0.4233,
      "step": 16148
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.546042799949646,
      "learning_rate": 1.0234212992270572e-05,
      "loss": 0.4514,
      "step": 16149
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.525217056274414,
      "learning_rate": 1.0233194859942714e-05,
      "loss": 0.3965,
      "step": 16150
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.660345435142517,
      "learning_rate": 1.0232176725196253e-05,
      "loss": 0.4956,
      "step": 16151
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6196351051330566,
      "learning_rate": 1.0231158588041747e-05,
      "loss": 0.3974,
      "step": 16152
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6163831949234009,
      "learning_rate": 1.0230140448489756e-05,
      "loss": 0.5013,
      "step": 16153
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6805459260940552,
      "learning_rate": 1.0229122306550844e-05,
      "loss": 0.4253,
      "step": 16154
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7168078422546387,
      "learning_rate": 1.0228104162235563e-05,
      "loss": 0.4898,
      "step": 16155
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5454976558685303,
      "learning_rate": 1.0227086015554482e-05,
      "loss": 0.4228,
      "step": 16156
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6216572523117065,
      "learning_rate": 1.0226067866518152e-05,
      "loss": 0.4857,
      "step": 16157
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5872507095336914,
      "learning_rate": 1.022504971513714e-05,
      "loss": 0.4094,
      "step": 16158
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.702856183052063,
      "learning_rate": 1.0224031561421998e-05,
      "loss": 0.4793,
      "step": 16159
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.594425916671753,
      "learning_rate": 1.0223013405383295e-05,
      "loss": 0.4296,
      "step": 16160
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4836419820785522,
      "learning_rate": 1.0221995247031582e-05,
      "loss": 0.4917,
      "step": 16161
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6092256307601929,
      "learning_rate": 1.0220977086377428e-05,
      "loss": 0.4386,
      "step": 16162
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6433101892471313,
      "learning_rate": 1.0219958923431386e-05,
      "loss": 0.4533,
      "step": 16163
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5490349531173706,
      "learning_rate": 1.0218940758204017e-05,
      "loss": 0.417,
      "step": 16164
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5224595069885254,
      "learning_rate": 1.0217922590705884e-05,
      "loss": 0.4544,
      "step": 16165
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.666336178779602,
      "learning_rate": 1.0216904420947544e-05,
      "loss": 0.442,
      "step": 16166
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6939445734024048,
      "learning_rate": 1.021588624893956e-05,
      "loss": 1.1082,
      "step": 16167
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.594275951385498,
      "learning_rate": 1.021486807469249e-05,
      "loss": 1.0154,
      "step": 16168
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.575273871421814,
      "learning_rate": 1.0213849898216895e-05,
      "loss": 0.4905,
      "step": 16169
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5813989639282227,
      "learning_rate": 1.0212831719523333e-05,
      "loss": 0.4468,
      "step": 16170
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5713446140289307,
      "learning_rate": 1.0211813538622364e-05,
      "loss": 0.4752,
      "step": 16171
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5601963996887207,
      "learning_rate": 1.0210795355524555e-05,
      "loss": 0.4162,
      "step": 16172
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5450963973999023,
      "learning_rate": 1.0209777170240457e-05,
      "loss": 0.4251,
      "step": 16173
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6701894998550415,
      "learning_rate": 1.0208758982780636e-05,
      "loss": 0.4447,
      "step": 16174
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5373176336288452,
      "learning_rate": 1.020774079315565e-05,
      "loss": 0.4433,
      "step": 16175
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5614261627197266,
      "learning_rate": 1.020672260137606e-05,
      "loss": 0.4404,
      "step": 16176
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5025594234466553,
      "learning_rate": 1.0205704407452427e-05,
      "loss": 0.441,
      "step": 16177
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6234345436096191,
      "learning_rate": 1.0204686211395308e-05,
      "loss": 0.4353,
      "step": 16178
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5187900066375732,
      "learning_rate": 1.0203668013215265e-05,
      "loss": 0.4644,
      "step": 16179
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.619741678237915,
      "learning_rate": 1.0202649812922865e-05,
      "loss": 0.4355,
      "step": 16180
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.734516978263855,
      "learning_rate": 1.0201631610528655e-05,
      "loss": 0.4666,
      "step": 16181
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6261430978775024,
      "learning_rate": 1.0200613406043208e-05,
      "loss": 0.4574,
      "step": 16182
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.658703327178955,
      "learning_rate": 1.0199595199477077e-05,
      "loss": 0.4792,
      "step": 16183
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4867892265319824,
      "learning_rate": 1.0198576990840826e-05,
      "loss": 0.4316,
      "step": 16184
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.505260944366455,
      "learning_rate": 1.019755878014501e-05,
      "loss": 0.4365,
      "step": 16185
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.556679129600525,
      "learning_rate": 1.0196540567400198e-05,
      "loss": 0.4274,
      "step": 16186
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.536903977394104,
      "learning_rate": 1.0195522352616942e-05,
      "loss": 0.4521,
      "step": 16187
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6529039144515991,
      "learning_rate": 1.0194504135805812e-05,
      "loss": 0.3985,
      "step": 16188
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5503116846084595,
      "learning_rate": 1.019348591697736e-05,
      "loss": 0.4788,
      "step": 16189
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4924099445343018,
      "learning_rate": 1.019246769614215e-05,
      "loss": 0.4103,
      "step": 16190
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.491490364074707,
      "learning_rate": 1.0191449473310741e-05,
      "loss": 0.4396,
      "step": 16191
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.524133563041687,
      "learning_rate": 1.0190431248493695e-05,
      "loss": 0.4132,
      "step": 16192
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5544815063476562,
      "learning_rate": 1.0189413021701575e-05,
      "loss": 0.4333,
      "step": 16193
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.512603759765625,
      "learning_rate": 1.0188394792944938e-05,
      "loss": 0.4316,
      "step": 16194
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.012134313583374,
      "learning_rate": 1.0187376562234344e-05,
      "loss": 1.0145,
      "step": 16195
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6617220640182495,
      "learning_rate": 1.0186358329580357e-05,
      "loss": 1.0246,
      "step": 16196
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7339386940002441,
      "learning_rate": 1.0185340094993534e-05,
      "loss": 0.4535,
      "step": 16197
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5977251529693604,
      "learning_rate": 1.0184321858484442e-05,
      "loss": 0.4222,
      "step": 16198
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5510197877883911,
      "learning_rate": 1.0183303620063633e-05,
      "loss": 0.4912,
      "step": 16199
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5295817852020264,
      "learning_rate": 1.0182285379741675e-05,
      "loss": 0.4109,
      "step": 16200
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5959274768829346,
      "learning_rate": 1.0181267137529124e-05,
      "loss": 0.4689,
      "step": 16201
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5403547286987305,
      "learning_rate": 1.0180248893436547e-05,
      "loss": 0.4265,
      "step": 16202
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7885202169418335,
      "learning_rate": 1.0179230647474498e-05,
      "loss": 0.5053,
      "step": 16203
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5356885194778442,
      "learning_rate": 1.017821239965354e-05,
      "loss": 0.4048,
      "step": 16204
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5896332263946533,
      "learning_rate": 1.0177194149984232e-05,
      "loss": 0.4606,
      "step": 16205
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.574444055557251,
      "learning_rate": 1.0176175898477145e-05,
      "loss": 0.4031,
      "step": 16206
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5470812320709229,
      "learning_rate": 1.0175157645142825e-05,
      "loss": 0.4796,
      "step": 16207
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4749329090118408,
      "learning_rate": 1.0174139389991846e-05,
      "loss": 0.3855,
      "step": 16208
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7119946479797363,
      "learning_rate": 1.0173121133034758e-05,
      "loss": 0.4857,
      "step": 16209
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6218258142471313,
      "learning_rate": 1.0172102874282128e-05,
      "loss": 0.4392,
      "step": 16210
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5554131269454956,
      "learning_rate": 1.0171084613744518e-05,
      "loss": 0.4338,
      "step": 16211
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6580764055252075,
      "learning_rate": 1.0170066351432486e-05,
      "loss": 0.4419,
      "step": 16212
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5219014883041382,
      "learning_rate": 1.0169048087356595e-05,
      "loss": 0.4276,
      "step": 16213
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5487034320831299,
      "learning_rate": 1.0168029821527404e-05,
      "loss": 0.4239,
      "step": 16214
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.710132122039795,
      "learning_rate": 1.0167011553955473e-05,
      "loss": 0.4983,
      "step": 16215
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5170867443084717,
      "learning_rate": 1.016599328465137e-05,
      "loss": 0.4154,
      "step": 16216
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6709972620010376,
      "learning_rate": 1.0164975013625646e-05,
      "loss": 0.4774,
      "step": 16217
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5942809581756592,
      "learning_rate": 1.016395674088887e-05,
      "loss": 0.4502,
      "step": 16218
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6683260202407837,
      "learning_rate": 1.0162938466451599e-05,
      "loss": 0.4739,
      "step": 16219
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4838932752609253,
      "learning_rate": 1.0161920190324397e-05,
      "loss": 0.395,
      "step": 16220
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6252697706222534,
      "learning_rate": 1.0160901912517824e-05,
      "loss": 0.4226,
      "step": 16221
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5224555730819702,
      "learning_rate": 1.015988363304244e-05,
      "loss": 0.4234,
      "step": 16222
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5775023698806763,
      "learning_rate": 1.0158865351908806e-05,
      "loss": 0.4739,
      "step": 16223
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5361884832382202,
      "learning_rate": 1.0157847069127489e-05,
      "loss": 0.426,
      "step": 16224
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6056914329528809,
      "learning_rate": 1.0156828784709038e-05,
      "loss": 0.4774,
      "step": 16225
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4508804082870483,
      "learning_rate": 1.015581049866403e-05,
      "loss": 0.4139,
      "step": 16226
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.563210129737854,
      "learning_rate": 1.015479221100301e-05,
      "loss": 0.466,
      "step": 16227
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5964151620864868,
      "learning_rate": 1.0153773921736551e-05,
      "loss": 0.4292,
      "step": 16228
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.610259771347046,
      "learning_rate": 1.015275563087521e-05,
      "loss": 0.5338,
      "step": 16229
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6005513668060303,
      "learning_rate": 1.015173733842955e-05,
      "loss": 0.4052,
      "step": 16230
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5725796222686768,
      "learning_rate": 1.0150719044410132e-05,
      "loss": 0.4814,
      "step": 16231
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.56147301197052,
      "learning_rate": 1.0149700748827516e-05,
      "loss": 0.4033,
      "step": 16232
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5608443021774292,
      "learning_rate": 1.014868245169226e-05,
      "loss": 0.4774,
      "step": 16233
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4844698905944824,
      "learning_rate": 1.0147664153014934e-05,
      "loss": 0.4136,
      "step": 16234
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5944138765335083,
      "learning_rate": 1.0146645852806092e-05,
      "loss": 0.5099,
      "step": 16235
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5919008255004883,
      "learning_rate": 1.0145627551076299e-05,
      "loss": 0.4127,
      "step": 16236
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5980181694030762,
      "learning_rate": 1.0144609247836115e-05,
      "loss": 0.4942,
      "step": 16237
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4911409616470337,
      "learning_rate": 1.0143590943096102e-05,
      "loss": 0.3944,
      "step": 16238
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6400388479232788,
      "learning_rate": 1.014257263686682e-05,
      "loss": 0.4691,
      "step": 16239
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5129108428955078,
      "learning_rate": 1.0141554329158833e-05,
      "loss": 0.3955,
      "step": 16240
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5722382068634033,
      "learning_rate": 1.01405360199827e-05,
      "loss": 0.4959,
      "step": 16241
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5986793041229248,
      "learning_rate": 1.0139517709348987e-05,
      "loss": 0.4558,
      "step": 16242
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5696555376052856,
      "learning_rate": 1.013849939726825e-05,
      "loss": 0.5234,
      "step": 16243
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4632718563079834,
      "learning_rate": 1.0137481083751056e-05,
      "loss": 0.3885,
      "step": 16244
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5802139043807983,
      "learning_rate": 1.0136462768807958e-05,
      "loss": 0.4916,
      "step": 16245
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5481915473937988,
      "learning_rate": 1.0135444452449523e-05,
      "loss": 0.4165,
      "step": 16246
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6188373565673828,
      "learning_rate": 1.0134426134686314e-05,
      "loss": 0.4945,
      "step": 16247
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.589227318763733,
      "learning_rate": 1.0133407815528893e-05,
      "loss": 0.4478,
      "step": 16248
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6722259521484375,
      "learning_rate": 1.0132389494987819e-05,
      "loss": 0.4944,
      "step": 16249
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5293147563934326,
      "learning_rate": 1.0131371173073654e-05,
      "loss": 0.422,
      "step": 16250
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6672019958496094,
      "learning_rate": 1.0130352849796959e-05,
      "loss": 0.518,
      "step": 16251
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.564199686050415,
      "learning_rate": 1.0129334525168297e-05,
      "loss": 0.431,
      "step": 16252
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.531549334526062,
      "learning_rate": 1.0128316199198227e-05,
      "loss": 0.4449,
      "step": 16253
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.656129002571106,
      "learning_rate": 1.0127297871897317e-05,
      "loss": 0.4409,
      "step": 16254
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.68433678150177,
      "learning_rate": 1.012627954327612e-05,
      "loss": 0.5456,
      "step": 16255
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5742871761322021,
      "learning_rate": 1.0125261213345205e-05,
      "loss": 0.4331,
      "step": 16256
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5303962230682373,
      "learning_rate": 1.0124242882115131e-05,
      "loss": 0.4492,
      "step": 16257
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4917125701904297,
      "learning_rate": 1.012322454959646e-05,
      "loss": 0.4565,
      "step": 16258
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6137100458145142,
      "learning_rate": 1.012220621579975e-05,
      "loss": 0.4553,
      "step": 16259
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5649789571762085,
      "learning_rate": 1.0121187880735572e-05,
      "loss": 0.4568,
      "step": 16260
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6482075452804565,
      "learning_rate": 1.0120169544414478e-05,
      "loss": 0.4813,
      "step": 16261
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.53250253200531,
      "learning_rate": 1.0119151206847034e-05,
      "loss": 0.413,
      "step": 16262
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5578618049621582,
      "learning_rate": 1.0118132868043803e-05,
      "loss": 0.4204,
      "step": 16263
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4993834495544434,
      "learning_rate": 1.0117114528015342e-05,
      "loss": 0.3856,
      "step": 16264
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6189699172973633,
      "learning_rate": 1.011609618677222e-05,
      "loss": 0.4715,
      "step": 16265
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5488708019256592,
      "learning_rate": 1.0115077844324993e-05,
      "loss": 0.4208,
      "step": 16266
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5245335102081299,
      "learning_rate": 1.0114059500684224e-05,
      "loss": 0.4633,
      "step": 16267
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5612300634384155,
      "learning_rate": 1.0113041155860476e-05,
      "loss": 0.4504,
      "step": 16268
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.423588514328003,
      "learning_rate": 1.0112022809864307e-05,
      "loss": 0.4209,
      "step": 16269
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5518507957458496,
      "learning_rate": 1.011100446270629e-05,
      "loss": 0.4433,
      "step": 16270
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6307084560394287,
      "learning_rate": 1.0109986114396973e-05,
      "loss": 0.5044,
      "step": 16271
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5602431297302246,
      "learning_rate": 1.0108967764946926e-05,
      "loss": 0.4172,
      "step": 16272
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6064664125442505,
      "learning_rate": 1.0107949414366711e-05,
      "loss": 0.4536,
      "step": 16273
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5242723226547241,
      "learning_rate": 1.0106931062666885e-05,
      "loss": 0.4359,
      "step": 16274
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5745962858200073,
      "learning_rate": 1.0105912709858015e-05,
      "loss": 0.4603,
      "step": 16275
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4646703004837036,
      "learning_rate": 1.010489435595066e-05,
      "loss": 0.396,
      "step": 16276
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5016261339187622,
      "learning_rate": 1.0103876000955381e-05,
      "loss": 0.4705,
      "step": 16277
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4930031299591064,
      "learning_rate": 1.0102857644882746e-05,
      "loss": 0.3852,
      "step": 16278
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5123568773269653,
      "learning_rate": 1.010183928774331e-05,
      "loss": 0.4872,
      "step": 16279
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5035920143127441,
      "learning_rate": 1.0100820929547637e-05,
      "loss": 0.3936,
      "step": 16280
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.629331350326538,
      "learning_rate": 1.0099802570306292e-05,
      "loss": 0.5115,
      "step": 16281
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.450709342956543,
      "learning_rate": 1.0098784210029832e-05,
      "loss": 0.3891,
      "step": 16282
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.585113763809204,
      "learning_rate": 1.0097765848728825e-05,
      "loss": 0.4717,
      "step": 16283
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4777110815048218,
      "learning_rate": 1.0096747486413826e-05,
      "loss": 0.4124,
      "step": 16284
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7951247692108154,
      "learning_rate": 1.0095729123095405e-05,
      "loss": 0.4563,
      "step": 16285
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6126409769058228,
      "learning_rate": 1.0094710758784119e-05,
      "loss": 0.4496,
      "step": 16286
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6550368070602417,
      "learning_rate": 1.0093692393490526e-05,
      "loss": 0.491,
      "step": 16287
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.502374529838562,
      "learning_rate": 1.00926740272252e-05,
      "loss": 0.3932,
      "step": 16288
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.616734266281128,
      "learning_rate": 1.0091655659998691e-05,
      "loss": 0.4698,
      "step": 16289
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.586991786956787,
      "learning_rate": 1.009063729182157e-05,
      "loss": 0.4101,
      "step": 16290
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6403648853302002,
      "learning_rate": 1.0089618922704394e-05,
      "loss": 0.5299,
      "step": 16291
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6787902116775513,
      "learning_rate": 1.0088600552657726e-05,
      "loss": 0.4161,
      "step": 16292
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8056803941726685,
      "learning_rate": 1.0087582181692129e-05,
      "loss": 0.5085,
      "step": 16293
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5812344551086426,
      "learning_rate": 1.0086563809818163e-05,
      "loss": 0.4183,
      "step": 16294
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5205286741256714,
      "learning_rate": 1.0085545437046393e-05,
      "loss": 0.4702,
      "step": 16295
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5104000568389893,
      "learning_rate": 1.0084527063387382e-05,
      "loss": 0.4052,
      "step": 16296
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5333162546157837,
      "learning_rate": 1.0083508688851687e-05,
      "loss": 0.4803,
      "step": 16297
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5975009202957153,
      "learning_rate": 1.0082490313449875e-05,
      "loss": 0.4468,
      "step": 16298
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4986075162887573,
      "learning_rate": 1.0081471937192507e-05,
      "loss": 0.4633,
      "step": 16299
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6387779712677002,
      "learning_rate": 1.0080453560090144e-05,
      "loss": 0.4387,
      "step": 16300
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6072781085968018,
      "learning_rate": 1.007943518215335e-05,
      "loss": 0.4628,
      "step": 16301
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5163179636001587,
      "learning_rate": 1.0078416803392685e-05,
      "loss": 0.4009,
      "step": 16302
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5159097909927368,
      "learning_rate": 1.0077398423818711e-05,
      "loss": 0.455,
      "step": 16303
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.620309829711914,
      "learning_rate": 1.0076380043441995e-05,
      "loss": 0.4576,
      "step": 16304
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5513648986816406,
      "learning_rate": 1.0075361662273091e-05,
      "loss": 0.481,
      "step": 16305
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5594570636749268,
      "learning_rate": 1.0074343280322571e-05,
      "loss": 0.4264,
      "step": 16306
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6508188247680664,
      "learning_rate": 1.0073324897600988e-05,
      "loss": 0.4288,
      "step": 16307
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5406813621520996,
      "learning_rate": 1.0072306514118912e-05,
      "loss": 0.4013,
      "step": 16308
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5664459466934204,
      "learning_rate": 1.0071288129886902e-05,
      "loss": 0.5033,
      "step": 16309
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4832898378372192,
      "learning_rate": 1.007026974491552e-05,
      "loss": 0.4402,
      "step": 16310
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.624023199081421,
      "learning_rate": 1.0069251359215326e-05,
      "loss": 0.4534,
      "step": 16311
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4950799942016602,
      "learning_rate": 1.0068232972796888e-05,
      "loss": 0.4102,
      "step": 16312
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5592399835586548,
      "learning_rate": 1.006721458567076e-05,
      "loss": 0.4582,
      "step": 16313
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5060170888900757,
      "learning_rate": 1.0066196197847512e-05,
      "loss": 0.4106,
      "step": 16314
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5638890266418457,
      "learning_rate": 1.0065177809337703e-05,
      "loss": 0.4969,
      "step": 16315
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5542889833450317,
      "learning_rate": 1.0064159420151898e-05,
      "loss": 0.4381,
      "step": 16316
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.501991629600525,
      "learning_rate": 1.0063141030300656e-05,
      "loss": 0.4539,
      "step": 16317
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5092524290084839,
      "learning_rate": 1.0062122639794541e-05,
      "loss": 0.4112,
      "step": 16318
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.598810076713562,
      "learning_rate": 1.0061104248644113e-05,
      "loss": 0.4673,
      "step": 16319
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5804178714752197,
      "learning_rate": 1.0060085856859938e-05,
      "loss": 0.4605,
      "step": 16320
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.502766489982605,
      "learning_rate": 1.0059067464452574e-05,
      "loss": 0.4829,
      "step": 16321
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.472764015197754,
      "learning_rate": 1.0058049071432589e-05,
      "loss": 0.4361,
      "step": 16322
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5891835689544678,
      "learning_rate": 1.005703067781054e-05,
      "loss": 0.4885,
      "step": 16323
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6003739833831787,
      "learning_rate": 1.0056012283596996e-05,
      "loss": 0.4177,
      "step": 16324
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.535959243774414,
      "learning_rate": 1.005499388880251e-05,
      "loss": 0.467,
      "step": 16325
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6475399732589722,
      "learning_rate": 1.0053975493437651e-05,
      "loss": 0.4312,
      "step": 16326
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.64244544506073,
      "learning_rate": 1.0052957097512982e-05,
      "loss": 0.4816,
      "step": 16327
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5255130529403687,
      "learning_rate": 1.0051938701039062e-05,
      "loss": 0.4139,
      "step": 16328
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6445099115371704,
      "learning_rate": 1.0050920304026453e-05,
      "loss": 0.5391,
      "step": 16329
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5185614824295044,
      "learning_rate": 1.0049901906485722e-05,
      "loss": 0.4134,
      "step": 16330
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5654900074005127,
      "learning_rate": 1.0048883508427423e-05,
      "loss": 0.4567,
      "step": 16331
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5254135131835938,
      "learning_rate": 1.004786510986213e-05,
      "loss": 0.4103,
      "step": 16332
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4954875707626343,
      "learning_rate": 1.0046846710800396e-05,
      "loss": 0.432,
      "step": 16333
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.733638048171997,
      "learning_rate": 1.0045828311252785e-05,
      "loss": 0.4614,
      "step": 16334
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6053755283355713,
      "learning_rate": 1.0044809911229864e-05,
      "loss": 0.4699,
      "step": 16335
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6052873134613037,
      "learning_rate": 1.0043791510742194e-05,
      "loss": 0.4558,
      "step": 16336
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5723167657852173,
      "learning_rate": 1.0042773109800335e-05,
      "loss": 0.4569,
      "step": 16337
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5861483812332153,
      "learning_rate": 1.0041754708414849e-05,
      "loss": 0.4314,
      "step": 16338
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.48793363571167,
      "learning_rate": 1.0040736306596296e-05,
      "loss": 0.4466,
      "step": 16339
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4997048377990723,
      "learning_rate": 1.0039717904355251e-05,
      "loss": 0.4416,
      "step": 16340
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5754555463790894,
      "learning_rate": 1.0038699501702263e-05,
      "loss": 0.4656,
      "step": 16341
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4834308624267578,
      "learning_rate": 1.00376810986479e-05,
      "loss": 0.4118,
      "step": 16342
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.552193284034729,
      "learning_rate": 1.0036662695202723e-05,
      "loss": 0.4771,
      "step": 16343
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.458652377128601,
      "learning_rate": 1.0035644291377296e-05,
      "loss": 0.454,
      "step": 16344
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5040082931518555,
      "learning_rate": 1.003462588718218e-05,
      "loss": 0.4494,
      "step": 16345
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4873766899108887,
      "learning_rate": 1.003360748262794e-05,
      "loss": 0.4211,
      "step": 16346
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7403253316879272,
      "learning_rate": 1.0032589077725135e-05,
      "loss": 0.4863,
      "step": 16347
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5060569047927856,
      "learning_rate": 1.003157067248433e-05,
      "loss": 0.3849,
      "step": 16348
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5544472932815552,
      "learning_rate": 1.0030552266916084e-05,
      "loss": 0.456,
      "step": 16349
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5575995445251465,
      "learning_rate": 1.0029533861030966e-05,
      "loss": 0.4405,
      "step": 16350
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5072413682937622,
      "learning_rate": 1.0028515454839532e-05,
      "loss": 0.4241,
      "step": 16351
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4899228811264038,
      "learning_rate": 1.0027497048352348e-05,
      "loss": 0.4034,
      "step": 16352
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.489770770072937,
      "learning_rate": 1.0026478641579976e-05,
      "loss": 0.4176,
      "step": 16353
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5041990280151367,
      "learning_rate": 1.0025460234532977e-05,
      "loss": 0.4001,
      "step": 16354
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5153803825378418,
      "learning_rate": 1.0024441827221918e-05,
      "loss": 0.4708,
      "step": 16355
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5864193439483643,
      "learning_rate": 1.0023423419657355e-05,
      "loss": 0.4279,
      "step": 16356
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5343997478485107,
      "learning_rate": 1.0022405011849851e-05,
      "loss": 0.4393,
      "step": 16357
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6100584268569946,
      "learning_rate": 1.002138660380998e-05,
      "loss": 0.412,
      "step": 16358
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4991931915283203,
      "learning_rate": 1.0020368195548287e-05,
      "loss": 0.4757,
      "step": 16359
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5365972518920898,
      "learning_rate": 1.001934978707535e-05,
      "loss": 0.4471,
      "step": 16360
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.593013882637024,
      "learning_rate": 1.0018331378401718e-05,
      "loss": 0.4519,
      "step": 16361
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4960193634033203,
      "learning_rate": 1.0017312969537965e-05,
      "loss": 0.3909,
      "step": 16362
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.565758228302002,
      "learning_rate": 1.0016294560494647e-05,
      "loss": 0.4426,
      "step": 16363
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5650619268417358,
      "learning_rate": 1.0015276151282329e-05,
      "loss": 0.408,
      "step": 16364
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.674999475479126,
      "learning_rate": 1.0014257741911573e-05,
      "loss": 0.952,
      "step": 16365
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4516003131866455,
      "learning_rate": 1.0013239332392941e-05,
      "loss": 1.0282,
      "step": 16366
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5503398180007935,
      "learning_rate": 1.0012220922736995e-05,
      "loss": 0.4864,
      "step": 16367
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6076089143753052,
      "learning_rate": 1.0011202512954302e-05,
      "loss": 0.4551,
      "step": 16368
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6247870922088623,
      "learning_rate": 1.0010184103055416e-05,
      "loss": 0.4674,
      "step": 16369
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5365028381347656,
      "learning_rate": 1.0009165693050906e-05,
      "loss": 0.4064,
      "step": 16370
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.3197097778320312,
      "learning_rate": 1.0008147282951334e-05,
      "loss": 0.4623,
      "step": 16371
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5609198808670044,
      "learning_rate": 1.0007128872767263e-05,
      "loss": 0.4174,
      "step": 16372
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5484906435012817,
      "learning_rate": 1.0006110462509254e-05,
      "loss": 0.4605,
      "step": 16373
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4663029909133911,
      "learning_rate": 1.0005092052187868e-05,
      "loss": 0.4146,
      "step": 16374
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5529658794403076,
      "learning_rate": 1.000407364181367e-05,
      "loss": 0.4925,
      "step": 16375
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5458558797836304,
      "learning_rate": 1.0003055231397224e-05,
      "loss": 0.4293,
      "step": 16376
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5858780145645142,
      "learning_rate": 1.0002036820949086e-05,
      "loss": 0.4427,
      "step": 16377
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5290387868881226,
      "learning_rate": 1.0001018410479826e-05,
      "loss": 0.4249,
      "step": 16378
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6166356801986694,
      "learning_rate": 1e-05,
      "loss": 0.4928,
      "step": 16379
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.562058925628662,
      "learning_rate": 9.998981589520176e-06,
      "loss": 0.4176,
      "step": 16380
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.512481689453125,
      "learning_rate": 9.997963179050915e-06,
      "loss": 0.4498,
      "step": 16381
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5125983953475952,
      "learning_rate": 9.996944768602781e-06,
      "loss": 0.4438,
      "step": 16382
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8186098337173462,
      "learning_rate": 9.995926358186335e-06,
      "loss": 0.5171,
      "step": 16383
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5082274675369263,
      "learning_rate": 9.994907947812135e-06,
      "loss": 0.4099,
      "step": 16384
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5399550199508667,
      "learning_rate": 9.99388953749075e-06,
      "loss": 0.4496,
      "step": 16385
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4822971820831299,
      "learning_rate": 9.992871127232742e-06,
      "loss": 0.4032,
      "step": 16386
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5368971824645996,
      "learning_rate": 9.991852717048667e-06,
      "loss": 0.4756,
      "step": 16387
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5081967115402222,
      "learning_rate": 9.990834306949096e-06,
      "loss": 0.3871,
      "step": 16388
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.55549955368042,
      "learning_rate": 9.989815896944585e-06,
      "loss": 0.4867,
      "step": 16389
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.611196756362915,
      "learning_rate": 9.988797487045703e-06,
      "loss": 0.428,
      "step": 16390
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.574302315711975,
      "learning_rate": 9.987779077263009e-06,
      "loss": 0.4704,
      "step": 16391
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.533124327659607,
      "learning_rate": 9.986760667607062e-06,
      "loss": 0.4116,
      "step": 16392
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.728442907333374,
      "learning_rate": 9.98574225808843e-06,
      "loss": 0.4505,
      "step": 16393
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5083023309707642,
      "learning_rate": 9.984723848717676e-06,
      "loss": 0.421,
      "step": 16394
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5793006420135498,
      "learning_rate": 9.983705439505357e-06,
      "loss": 0.4647,
      "step": 16395
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5545400381088257,
      "learning_rate": 9.982687030462039e-06,
      "loss": 0.4372,
      "step": 16396
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5779345035552979,
      "learning_rate": 9.981668621598283e-06,
      "loss": 0.4787,
      "step": 16397
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4870492219924927,
      "learning_rate": 9.980650212924656e-06,
      "loss": 0.4337,
      "step": 16398
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5542739629745483,
      "learning_rate": 9.979631804451717e-06,
      "loss": 0.4728,
      "step": 16399
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4913736581802368,
      "learning_rate": 9.978613396190024e-06,
      "loss": 0.4181,
      "step": 16400
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6447428464889526,
      "learning_rate": 9.977594988150148e-06,
      "loss": 0.5143,
      "step": 16401
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.579258918762207,
      "learning_rate": 9.97657658034265e-06,
      "loss": 0.4488,
      "step": 16402
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.565915584564209,
      "learning_rate": 9.975558172778085e-06,
      "loss": 0.4555,
      "step": 16403
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5512198209762573,
      "learning_rate": 9.974539765467024e-06,
      "loss": 0.4226,
      "step": 16404
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6677700281143188,
      "learning_rate": 9.973521358420024e-06,
      "loss": 0.4981,
      "step": 16405
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.528320074081421,
      "learning_rate": 9.972502951647654e-06,
      "loss": 0.4249,
      "step": 16406
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4720714092254639,
      "learning_rate": 9.971484545160472e-06,
      "loss": 0.4243,
      "step": 16407
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8029365539550781,
      "learning_rate": 9.970466138969036e-06,
      "loss": 0.4564,
      "step": 16408
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5945191383361816,
      "learning_rate": 9.969447733083917e-06,
      "loss": 0.4258,
      "step": 16409
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5586986541748047,
      "learning_rate": 9.968429327515676e-06,
      "loss": 0.4006,
      "step": 16410
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5971261262893677,
      "learning_rate": 9.967410922274868e-06,
      "loss": 0.4603,
      "step": 16411
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5718128681182861,
      "learning_rate": 9.966392517372062e-06,
      "loss": 0.4808,
      "step": 16412
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.621843934059143,
      "learning_rate": 9.965374112817821e-06,
      "loss": 0.4578,
      "step": 16413
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5426489114761353,
      "learning_rate": 9.964355708622708e-06,
      "loss": 0.4187,
      "step": 16414
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9208884239196777,
      "learning_rate": 9.96333730479728e-06,
      "loss": 1.0065,
      "step": 16415
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5954629182815552,
      "learning_rate": 9.962318901352102e-06,
      "loss": 1.0185,
      "step": 16416
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7067910432815552,
      "learning_rate": 9.961300498297739e-06,
      "loss": 0.4774,
      "step": 16417
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.520688772201538,
      "learning_rate": 9.960282095644754e-06,
      "loss": 0.4148,
      "step": 16418
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5949594974517822,
      "learning_rate": 9.959263693403705e-06,
      "loss": 0.468,
      "step": 16419
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4663736820220947,
      "learning_rate": 9.958245291585155e-06,
      "loss": 0.4127,
      "step": 16420
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6469558477401733,
      "learning_rate": 9.957226890199669e-06,
      "loss": 0.4893,
      "step": 16421
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5212448835372925,
      "learning_rate": 9.95620848925781e-06,
      "loss": 0.4181,
      "step": 16422
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6606940031051636,
      "learning_rate": 9.955190088770137e-06,
      "loss": 0.4902,
      "step": 16423
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.535143494606018,
      "learning_rate": 9.954171688747217e-06,
      "loss": 0.4184,
      "step": 16424
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.546884536743164,
      "learning_rate": 9.953153289199606e-06,
      "loss": 0.4321,
      "step": 16425
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5008405447006226,
      "learning_rate": 9.952134890137875e-06,
      "loss": 0.418,
      "step": 16426
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7123969793319702,
      "learning_rate": 9.95111649157258e-06,
      "loss": 0.4578,
      "step": 16427
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5945965051651,
      "learning_rate": 9.950098093514282e-06,
      "loss": 0.4205,
      "step": 16428
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5997494459152222,
      "learning_rate": 9.949079695973549e-06,
      "loss": 0.481,
      "step": 16429
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5318201780319214,
      "learning_rate": 9.948061298960943e-06,
      "loss": 0.431,
      "step": 16430
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6454863548278809,
      "learning_rate": 9.947042902487022e-06,
      "loss": 0.5248,
      "step": 16431
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6797754764556885,
      "learning_rate": 9.94602450656235e-06,
      "loss": 0.4268,
      "step": 16432
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.675183892250061,
      "learning_rate": 9.945006111197491e-06,
      "loss": 0.4694,
      "step": 16433
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.614698886871338,
      "learning_rate": 9.943987716403009e-06,
      "loss": 0.4123,
      "step": 16434
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5929518938064575,
      "learning_rate": 9.942969322189462e-06,
      "loss": 0.5207,
      "step": 16435
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7016464471817017,
      "learning_rate": 9.941950928567413e-06,
      "loss": 0.4669,
      "step": 16436
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5062698125839233,
      "learning_rate": 9.940932535547425e-06,
      "loss": 0.4518,
      "step": 16437
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4554848670959473,
      "learning_rate": 9.939914143140067e-06,
      "loss": 0.4068,
      "step": 16438
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5347890853881836,
      "learning_rate": 9.938895751355889e-06,
      "loss": 0.4603,
      "step": 16439
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.51960289478302,
      "learning_rate": 9.937877360205464e-06,
      "loss": 0.4646,
      "step": 16440
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.582022786140442,
      "learning_rate": 9.936858969699345e-06,
      "loss": 0.4889,
      "step": 16441
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8466750383377075,
      "learning_rate": 9.935840579848105e-06,
      "loss": 0.4411,
      "step": 16442
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.509382724761963,
      "learning_rate": 9.934822190662299e-06,
      "loss": 0.4724,
      "step": 16443
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4675194025039673,
      "learning_rate": 9.93380380215249e-06,
      "loss": 0.3971,
      "step": 16444
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6429576873779297,
      "learning_rate": 9.93278541432924e-06,
      "loss": 0.4914,
      "step": 16445
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5272499322891235,
      "learning_rate": 9.931767027203119e-06,
      "loss": 0.4098,
      "step": 16446
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6256613731384277,
      "learning_rate": 9.930748640784676e-06,
      "loss": 0.4934,
      "step": 16447
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.579882264137268,
      "learning_rate": 9.929730255084485e-06,
      "loss": 0.4453,
      "step": 16448
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.492925763130188,
      "learning_rate": 9.9287118701131e-06,
      "loss": 0.4823,
      "step": 16449
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.524300456047058,
      "learning_rate": 9.927693485881091e-06,
      "loss": 0.4374,
      "step": 16450
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4702353477478027,
      "learning_rate": 9.926675102399013e-06,
      "loss": 0.4002,
      "step": 16451
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5135703086853027,
      "learning_rate": 9.92565671967743e-06,
      "loss": 0.457,
      "step": 16452
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5795085430145264,
      "learning_rate": 9.924638337726909e-06,
      "loss": 0.4636,
      "step": 16453
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5767146348953247,
      "learning_rate": 9.92361995655801e-06,
      "loss": 0.4283,
      "step": 16454
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7000590562820435,
      "learning_rate": 9.92260157618129e-06,
      "loss": 0.4107,
      "step": 16455
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4849390983581543,
      "learning_rate": 9.921583196607318e-06,
      "loss": 0.393,
      "step": 16456
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5671510696411133,
      "learning_rate": 9.920564817846651e-06,
      "loss": 0.491,
      "step": 16457
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.617486834526062,
      "learning_rate": 9.91954643990986e-06,
      "loss": 0.4439,
      "step": 16458
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5682857036590576,
      "learning_rate": 9.918528062807496e-06,
      "loss": 0.4647,
      "step": 16459
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5173701047897339,
      "learning_rate": 9.917509686550129e-06,
      "loss": 0.4338,
      "step": 16460
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5874630212783813,
      "learning_rate": 9.916491311148315e-06,
      "loss": 0.4954,
      "step": 16461
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4856916666030884,
      "learning_rate": 9.915472936612623e-06,
      "loss": 0.4218,
      "step": 16462
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5935231447219849,
      "learning_rate": 9.91445456295361e-06,
      "loss": 0.4816,
      "step": 16463
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5856456756591797,
      "learning_rate": 9.913436190181839e-06,
      "loss": 0.4304,
      "step": 16464
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.593761920928955,
      "learning_rate": 9.912417818307875e-06,
      "loss": 0.4623,
      "step": 16465
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4822072982788086,
      "learning_rate": 9.91139944734228e-06,
      "loss": 0.3936,
      "step": 16466
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5668693780899048,
      "learning_rate": 9.91038107729561e-06,
      "loss": 0.514,
      "step": 16467
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5351227521896362,
      "learning_rate": 9.909362708178434e-06,
      "loss": 0.4071,
      "step": 16468
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7105932235717773,
      "learning_rate": 9.908344340001309e-06,
      "loss": 1.0104,
      "step": 16469
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4777268171310425,
      "learning_rate": 9.907325972774805e-06,
      "loss": 0.934,
      "step": 16470
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8021440505981445,
      "learning_rate": 9.906307606509476e-06,
      "loss": 0.5753,
      "step": 16471
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6665066480636597,
      "learning_rate": 9.905289241215886e-06,
      "loss": 0.3879,
      "step": 16472
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.654491662979126,
      "learning_rate": 9.904270876904598e-06,
      "loss": 0.4376,
      "step": 16473
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6639773845672607,
      "learning_rate": 9.903252513586179e-06,
      "loss": 0.4073,
      "step": 16474
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5458769798278809,
      "learning_rate": 9.902234151271178e-06,
      "loss": 0.4371,
      "step": 16475
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5242490768432617,
      "learning_rate": 9.90121578997017e-06,
      "loss": 0.3832,
      "step": 16476
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5631593465805054,
      "learning_rate": 9.90019742969371e-06,
      "loss": 0.4843,
      "step": 16477
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5422818660736084,
      "learning_rate": 9.899179070452366e-06,
      "loss": 0.4246,
      "step": 16478
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.824407696723938,
      "learning_rate": 9.898160712256695e-06,
      "loss": 0.4554,
      "step": 16479
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5131055116653442,
      "learning_rate": 9.897142355117256e-06,
      "loss": 0.4191,
      "step": 16480
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5461345911026,
      "learning_rate": 9.896123999044619e-06,
      "loss": 0.4525,
      "step": 16481
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.756837248802185,
      "learning_rate": 9.895105644049345e-06,
      "loss": 0.4347,
      "step": 16482
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5284755229949951,
      "learning_rate": 9.894087290141987e-06,
      "loss": 0.4829,
      "step": 16483
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6089156866073608,
      "learning_rate": 9.893068937333117e-06,
      "loss": 0.4632,
      "step": 16484
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9533888101577759,
      "learning_rate": 9.89205058563329e-06,
      "loss": 0.9483,
      "step": 16485
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.675496220588684,
      "learning_rate": 9.891032235053076e-06,
      "loss": 1.06,
      "step": 16486
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6202012300491333,
      "learning_rate": 9.890013885603029e-06,
      "loss": 0.468,
      "step": 16487
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5296475887298584,
      "learning_rate": 9.888995537293713e-06,
      "loss": 0.4142,
      "step": 16488
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5990835428237915,
      "learning_rate": 9.887977190135693e-06,
      "loss": 0.4636,
      "step": 16489
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4942303895950317,
      "learning_rate": 9.886958844139529e-06,
      "loss": 0.4353,
      "step": 16490
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7784745693206787,
      "learning_rate": 9.885940499315779e-06,
      "loss": 0.4757,
      "step": 16491
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.449609637260437,
      "learning_rate": 9.88492215567501e-06,
      "loss": 0.3826,
      "step": 16492
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4904049634933472,
      "learning_rate": 9.883903813227781e-06,
      "loss": 0.4254,
      "step": 16493
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4769001007080078,
      "learning_rate": 9.882885471984662e-06,
      "loss": 0.4083,
      "step": 16494
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.490079402923584,
      "learning_rate": 9.881867131956202e-06,
      "loss": 0.4356,
      "step": 16495
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4758644104003906,
      "learning_rate": 9.88084879315297e-06,
      "loss": 0.4393,
      "step": 16496
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5887912511825562,
      "learning_rate": 9.879830455585524e-06,
      "loss": 0.4566,
      "step": 16497
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4599764347076416,
      "learning_rate": 9.878812119264433e-06,
      "loss": 0.4245,
      "step": 16498
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6569377183914185,
      "learning_rate": 9.877793784200251e-06,
      "loss": 0.4732,
      "step": 16499
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5516327619552612,
      "learning_rate": 9.876775450403542e-06,
      "loss": 0.412,
      "step": 16500
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7259080410003662,
      "learning_rate": 9.87575711788487e-06,
      "loss": 0.4731,
      "step": 16501
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5465576648712158,
      "learning_rate": 9.874738786654798e-06,
      "loss": 0.402,
      "step": 16502
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.580885887145996,
      "learning_rate": 9.873720456723882e-06,
      "loss": 0.5028,
      "step": 16503
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5084609985351562,
      "learning_rate": 9.872702128102686e-06,
      "loss": 0.4079,
      "step": 16504
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.572841763496399,
      "learning_rate": 9.871683800801773e-06,
      "loss": 0.5036,
      "step": 16505
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5461825132369995,
      "learning_rate": 9.870665474831708e-06,
      "loss": 0.4228,
      "step": 16506
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7562226057052612,
      "learning_rate": 9.869647150203046e-06,
      "loss": 0.9475,
      "step": 16507
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6316102743148804,
      "learning_rate": 9.868628826926349e-06,
      "loss": 1.0924,
      "step": 16508
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5899548530578613,
      "learning_rate": 9.867610505012184e-06,
      "loss": 0.486,
      "step": 16509
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5691413879394531,
      "learning_rate": 9.86659218447111e-06,
      "loss": 0.4034,
      "step": 16510
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5336558818817139,
      "learning_rate": 9.865573865313687e-06,
      "loss": 0.4427,
      "step": 16511
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4622634649276733,
      "learning_rate": 9.864555547550478e-06,
      "loss": 0.4108,
      "step": 16512
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5617817640304565,
      "learning_rate": 9.863537231192043e-06,
      "loss": 0.4801,
      "step": 16513
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.533262848854065,
      "learning_rate": 9.862518916248951e-06,
      "loss": 0.3935,
      "step": 16514
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8579235076904297,
      "learning_rate": 9.861500602731754e-06,
      "loss": 1.0077,
      "step": 16515
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.63570237159729,
      "learning_rate": 9.860482290651015e-06,
      "loss": 0.9926,
      "step": 16516
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5190883874893188,
      "learning_rate": 9.859463980017301e-06,
      "loss": 0.4571,
      "step": 16517
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.566422700881958,
      "learning_rate": 9.858445670841172e-06,
      "loss": 0.4441,
      "step": 16518
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6231235265731812,
      "learning_rate": 9.857427363133182e-06,
      "loss": 0.4698,
      "step": 16519
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4798427820205688,
      "learning_rate": 9.856409056903901e-06,
      "loss": 0.3929,
      "step": 16520
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9955320358276367,
      "learning_rate": 9.855390752163887e-06,
      "loss": 0.4527,
      "step": 16521
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6239573955535889,
      "learning_rate": 9.854372448923705e-06,
      "loss": 0.4166,
      "step": 16522
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5680276155471802,
      "learning_rate": 9.853354147193911e-06,
      "loss": 0.4905,
      "step": 16523
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5080268383026123,
      "learning_rate": 9.852335846985067e-06,
      "loss": 0.4071,
      "step": 16524
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6278719902038574,
      "learning_rate": 9.85131754830774e-06,
      "loss": 0.467,
      "step": 16525
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.571927785873413,
      "learning_rate": 9.850299251172491e-06,
      "loss": 0.4159,
      "step": 16526
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6527565717697144,
      "learning_rate": 9.84928095558987e-06,
      "loss": 0.548,
      "step": 16527
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.483896017074585,
      "learning_rate": 9.848262661570451e-06,
      "loss": 0.4059,
      "step": 16528
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5028996467590332,
      "learning_rate": 9.847244369124788e-06,
      "loss": 0.4347,
      "step": 16529
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5585664510726929,
      "learning_rate": 9.84622607826345e-06,
      "loss": 0.4101,
      "step": 16530
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5381773710250854,
      "learning_rate": 9.845207788996992e-06,
      "loss": 0.4643,
      "step": 16531
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.479662299156189,
      "learning_rate": 9.844189501335974e-06,
      "loss": 0.424,
      "step": 16532
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6237566471099854,
      "learning_rate": 9.843171215290962e-06,
      "loss": 0.4633,
      "step": 16533
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5259791612625122,
      "learning_rate": 9.842152930872516e-06,
      "loss": 0.3984,
      "step": 16534
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5619359016418457,
      "learning_rate": 9.841134648091198e-06,
      "loss": 0.4836,
      "step": 16535
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6050093173980713,
      "learning_rate": 9.840116366957561e-06,
      "loss": 0.4249,
      "step": 16536
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6025300025939941,
      "learning_rate": 9.839098087482178e-06,
      "loss": 0.5372,
      "step": 16537
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7542681694030762,
      "learning_rate": 9.838079809675607e-06,
      "loss": 0.4101,
      "step": 16538
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5907701253890991,
      "learning_rate": 9.837061533548403e-06,
      "loss": 0.4951,
      "step": 16539
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.598168969154358,
      "learning_rate": 9.836043259111134e-06,
      "loss": 0.429,
      "step": 16540
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5924915075302124,
      "learning_rate": 9.835024986374355e-06,
      "loss": 0.4934,
      "step": 16541
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.492440938949585,
      "learning_rate": 9.834006715348635e-06,
      "loss": 0.4223,
      "step": 16542
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4940017461776733,
      "learning_rate": 9.83298844604453e-06,
      "loss": 0.4366,
      "step": 16543
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6679697036743164,
      "learning_rate": 9.831970178472598e-06,
      "loss": 0.4087,
      "step": 16544
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4896738529205322,
      "learning_rate": 9.830951912643408e-06,
      "loss": 0.4321,
      "step": 16545
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.488492727279663,
      "learning_rate": 9.82993364856752e-06,
      "loss": 0.396,
      "step": 16546
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5879402160644531,
      "learning_rate": 9.828915386255484e-06,
      "loss": 0.4806,
      "step": 16547
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.485445261001587,
      "learning_rate": 9.827897125717873e-06,
      "loss": 0.4099,
      "step": 16548
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5931907892227173,
      "learning_rate": 9.826878866965242e-06,
      "loss": 0.472,
      "step": 16549
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4461451768875122,
      "learning_rate": 9.825860610008159e-06,
      "loss": 0.4062,
      "step": 16550
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4570627212524414,
      "learning_rate": 9.824842354857177e-06,
      "loss": 0.4595,
      "step": 16551
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6124202013015747,
      "learning_rate": 9.823824101522858e-06,
      "loss": 0.4792,
      "step": 16552
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.743698239326477,
      "learning_rate": 9.822805850015766e-06,
      "loss": 0.505,
      "step": 16553
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5836988687515259,
      "learning_rate": 9.821787600346464e-06,
      "loss": 0.4399,
      "step": 16554
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6255221366882324,
      "learning_rate": 9.820769352525505e-06,
      "loss": 0.4924,
      "step": 16555
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.459378719329834,
      "learning_rate": 9.819751106563456e-06,
      "loss": 0.405,
      "step": 16556
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5583112239837646,
      "learning_rate": 9.818732862470875e-06,
      "loss": 0.4329,
      "step": 16557
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5113826990127563,
      "learning_rate": 9.817714620258328e-06,
      "loss": 0.4104,
      "step": 16558
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6093382835388184,
      "learning_rate": 9.81669637993637e-06,
      "loss": 0.4491,
      "step": 16559
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7287594079971313,
      "learning_rate": 9.815678141515561e-06,
      "loss": 0.4874,
      "step": 16560
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.499529480934143,
      "learning_rate": 9.814659905006468e-06,
      "loss": 0.4138,
      "step": 16561
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6224604845046997,
      "learning_rate": 9.813641670419648e-06,
      "loss": 0.4496,
      "step": 16562
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6781929731369019,
      "learning_rate": 9.812623437765658e-06,
      "loss": 0.5732,
      "step": 16563
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5263733863830566,
      "learning_rate": 9.811605207055066e-06,
      "loss": 0.3774,
      "step": 16564
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5861046314239502,
      "learning_rate": 9.810586978298426e-06,
      "loss": 0.457,
      "step": 16565
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5492819547653198,
      "learning_rate": 9.809568751506307e-06,
      "loss": 0.4747,
      "step": 16566
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6265515089035034,
      "learning_rate": 9.808550526689262e-06,
      "loss": 0.5023,
      "step": 16567
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5039225816726685,
      "learning_rate": 9.807532303857852e-06,
      "loss": 0.4034,
      "step": 16568
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.611950159072876,
      "learning_rate": 9.806514083022642e-06,
      "loss": 0.4368,
      "step": 16569
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5033429861068726,
      "learning_rate": 9.805495864194193e-06,
      "loss": 0.4166,
      "step": 16570
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4978529214859009,
      "learning_rate": 9.804477647383061e-06,
      "loss": 0.4496,
      "step": 16571
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5561343431472778,
      "learning_rate": 9.803459432599804e-06,
      "loss": 0.4192,
      "step": 16572
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6571481227874756,
      "learning_rate": 9.80244121985499e-06,
      "loss": 0.4869,
      "step": 16573
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5523232221603394,
      "learning_rate": 9.801423009159181e-06,
      "loss": 0.4004,
      "step": 16574
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5232642889022827,
      "learning_rate": 9.800404800522927e-06,
      "loss": 0.4733,
      "step": 16575
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5981926918029785,
      "learning_rate": 9.799386593956797e-06,
      "loss": 0.439,
      "step": 16576
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5519523620605469,
      "learning_rate": 9.798368389471345e-06,
      "loss": 0.4938,
      "step": 16577
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4752660989761353,
      "learning_rate": 9.797350187077142e-06,
      "loss": 0.382,
      "step": 16578
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5258454084396362,
      "learning_rate": 9.796331986784737e-06,
      "loss": 0.4919,
      "step": 16579
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4782284498214722,
      "learning_rate": 9.795313788604694e-06,
      "loss": 0.3944,
      "step": 16580
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6098310947418213,
      "learning_rate": 9.794295592547577e-06,
      "loss": 0.4392,
      "step": 16581
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5143167972564697,
      "learning_rate": 9.793277398623946e-06,
      "loss": 0.398,
      "step": 16582
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6835397481918335,
      "learning_rate": 9.792259206844354e-06,
      "loss": 0.4418,
      "step": 16583
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9058624505996704,
      "learning_rate": 9.791241017219368e-06,
      "loss": 0.3885,
      "step": 16584
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5408562421798706,
      "learning_rate": 9.790222829759545e-06,
      "loss": 0.481,
      "step": 16585
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4600732326507568,
      "learning_rate": 9.78920464447545e-06,
      "loss": 0.3984,
      "step": 16586
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5292588472366333,
      "learning_rate": 9.788186461377639e-06,
      "loss": 0.452,
      "step": 16587
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4973728656768799,
      "learning_rate": 9.78716828047667e-06,
      "loss": 0.4229,
      "step": 16588
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.653003215789795,
      "learning_rate": 9.786150101783109e-06,
      "loss": 0.5011,
      "step": 16589
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5706384181976318,
      "learning_rate": 9.785131925307515e-06,
      "loss": 0.422,
      "step": 16590
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5345635414123535,
      "learning_rate": 9.784113751060442e-06,
      "loss": 0.5082,
      "step": 16591
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0245606899261475,
      "learning_rate": 9.783095579052457e-06,
      "loss": 0.4454,
      "step": 16592
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.247897148132324,
      "learning_rate": 9.782077409294116e-06,
      "loss": 0.5039,
      "step": 16593
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.568771481513977,
      "learning_rate": 9.781059241795987e-06,
      "loss": 0.429,
      "step": 16594
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5653446912765503,
      "learning_rate": 9.780041076568619e-06,
      "loss": 0.4566,
      "step": 16595
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5025439262390137,
      "learning_rate": 9.779022913622574e-06,
      "loss": 0.4147,
      "step": 16596
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5599168539047241,
      "learning_rate": 9.778004752968418e-06,
      "loss": 0.4974,
      "step": 16597
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4772037267684937,
      "learning_rate": 9.77698659461671e-06,
      "loss": 0.4089,
      "step": 16598
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4637571573257446,
      "learning_rate": 9.775968438578004e-06,
      "loss": 0.4495,
      "step": 16599
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5137863159179688,
      "learning_rate": 9.774950284862864e-06,
      "loss": 0.3911,
      "step": 16600
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5330959558486938,
      "learning_rate": 9.773932133481848e-06,
      "loss": 0.4438,
      "step": 16601
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5594353675842285,
      "learning_rate": 9.772913984445522e-06,
      "loss": 0.4161,
      "step": 16602
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.588800072669983,
      "learning_rate": 9.771895837764438e-06,
      "loss": 0.4729,
      "step": 16603
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5696076154708862,
      "learning_rate": 9.770877693449157e-06,
      "loss": 0.4188,
      "step": 16604
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6529186964035034,
      "learning_rate": 9.769859551510244e-06,
      "loss": 0.4967,
      "step": 16605
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4920674562454224,
      "learning_rate": 9.768841411958258e-06,
      "loss": 0.3958,
      "step": 16606
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5947035551071167,
      "learning_rate": 9.76782327480375e-06,
      "loss": 0.4866,
      "step": 16607
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5257593393325806,
      "learning_rate": 9.766805140057289e-06,
      "loss": 0.4374,
      "step": 16608
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6759588718414307,
      "learning_rate": 9.76578700772943e-06,
      "loss": 0.5276,
      "step": 16609
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4987280368804932,
      "learning_rate": 9.764768877830741e-06,
      "loss": 0.4124,
      "step": 16610
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5925202369689941,
      "learning_rate": 9.763750750371766e-06,
      "loss": 0.4425,
      "step": 16611
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6522893905639648,
      "learning_rate": 9.762732625363078e-06,
      "loss": 0.4053,
      "step": 16612
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4672996997833252,
      "learning_rate": 9.761714502815229e-06,
      "loss": 0.4499,
      "step": 16613
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5663708448410034,
      "learning_rate": 9.760696382738785e-06,
      "loss": 0.4764,
      "step": 16614
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5810133218765259,
      "learning_rate": 9.759678265144302e-06,
      "loss": 0.4862,
      "step": 16615
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.552992343902588,
      "learning_rate": 9.758660150042336e-06,
      "loss": 0.4351,
      "step": 16616
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6005380153656006,
      "learning_rate": 9.757642037443453e-06,
      "loss": 0.5226,
      "step": 16617
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5069432258605957,
      "learning_rate": 9.756623927358212e-06,
      "loss": 0.37,
      "step": 16618
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6789690256118774,
      "learning_rate": 9.755605819797164e-06,
      "loss": 0.4998,
      "step": 16619
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3824807405471802,
      "learning_rate": 9.75458771477088e-06,
      "loss": 0.3834,
      "step": 16620
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5550237894058228,
      "learning_rate": 9.753569612289908e-06,
      "loss": 0.4719,
      "step": 16621
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.561877965927124,
      "learning_rate": 9.752551512364819e-06,
      "loss": 0.4131,
      "step": 16622
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5334933996200562,
      "learning_rate": 9.751533415006166e-06,
      "loss": 0.4711,
      "step": 16623
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5237243175506592,
      "learning_rate": 9.750515320224506e-06,
      "loss": 0.4048,
      "step": 16624
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5449546575546265,
      "learning_rate": 9.749497228030401e-06,
      "loss": 0.5477,
      "step": 16625
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5946439504623413,
      "learning_rate": 9.74847913843441e-06,
      "loss": 0.437,
      "step": 16626
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5681817531585693,
      "learning_rate": 9.747461051447097e-06,
      "loss": 0.485,
      "step": 16627
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.522612452507019,
      "learning_rate": 9.746442967079014e-06,
      "loss": 0.4572,
      "step": 16628
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6022312641143799,
      "learning_rate": 9.745424885340721e-06,
      "loss": 0.4807,
      "step": 16629
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.621688961982727,
      "learning_rate": 9.74440680624278e-06,
      "loss": 0.4203,
      "step": 16630
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5833600759506226,
      "learning_rate": 9.743388729795753e-06,
      "loss": 0.5087,
      "step": 16631
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.442409634590149,
      "learning_rate": 9.74237065601019e-06,
      "loss": 0.4296,
      "step": 16632
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4979314804077148,
      "learning_rate": 9.741352584896657e-06,
      "loss": 0.4621,
      "step": 16633
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5797617435455322,
      "learning_rate": 9.740334516465709e-06,
      "loss": 0.444,
      "step": 16634
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6186237335205078,
      "learning_rate": 9.739316450727914e-06,
      "loss": 0.4739,
      "step": 16635
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5181058645248413,
      "learning_rate": 9.73829838769382e-06,
      "loss": 0.4147,
      "step": 16636
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.523457407951355,
      "learning_rate": 9.737280327373988e-06,
      "loss": 0.4736,
      "step": 16637
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5995522737503052,
      "learning_rate": 9.736262269778982e-06,
      "loss": 0.4434,
      "step": 16638
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6629705429077148,
      "learning_rate": 9.735244214919359e-06,
      "loss": 0.4952,
      "step": 16639
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5004810094833374,
      "learning_rate": 9.734226162805674e-06,
      "loss": 0.3994,
      "step": 16640
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5829954147338867,
      "learning_rate": 9.73320811344849e-06,
      "loss": 0.4326,
      "step": 16641
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.477677822113037,
      "learning_rate": 9.732190066858363e-06,
      "loss": 0.4048,
      "step": 16642
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.53182852268219,
      "learning_rate": 9.731172023045857e-06,
      "loss": 0.4521,
      "step": 16643
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6064784526824951,
      "learning_rate": 9.730153982021525e-06,
      "loss": 0.4307,
      "step": 16644
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.600944995880127,
      "learning_rate": 9.729135943795925e-06,
      "loss": 0.5007,
      "step": 16645
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5043153762817383,
      "learning_rate": 9.728117908379621e-06,
      "loss": 0.4124,
      "step": 16646
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.599453091621399,
      "learning_rate": 9.727099875783173e-06,
      "loss": 0.5184,
      "step": 16647
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4881619215011597,
      "learning_rate": 9.726081846017127e-06,
      "loss": 0.4051,
      "step": 16648
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6158201694488525,
      "learning_rate": 9.725063819092056e-06,
      "loss": 0.4425,
      "step": 16649
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5481319427490234,
      "learning_rate": 9.72404579501851e-06,
      "loss": 0.4292,
      "step": 16650
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6354825496673584,
      "learning_rate": 9.723027773807056e-06,
      "loss": 0.4623,
      "step": 16651
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5722713470458984,
      "learning_rate": 9.722009755468243e-06,
      "loss": 0.4221,
      "step": 16652
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5327801704406738,
      "learning_rate": 9.720991740012633e-06,
      "loss": 0.529,
      "step": 16653
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5273411273956299,
      "learning_rate": 9.719973727450784e-06,
      "loss": 0.4366,
      "step": 16654
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4868091344833374,
      "learning_rate": 9.718955717793258e-06,
      "loss": 0.448,
      "step": 16655
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4855071306228638,
      "learning_rate": 9.717937711050608e-06,
      "loss": 0.415,
      "step": 16656
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.508331298828125,
      "learning_rate": 9.716919707233395e-06,
      "loss": 0.4669,
      "step": 16657
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5402604341506958,
      "learning_rate": 9.715901706352178e-06,
      "loss": 0.4416,
      "step": 16658
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4674524068832397,
      "learning_rate": 9.71488370841752e-06,
      "loss": 0.425,
      "step": 16659
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5080995559692383,
      "learning_rate": 9.713865713439966e-06,
      "loss": 0.4105,
      "step": 16660
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.601823091506958,
      "learning_rate": 9.712847721430084e-06,
      "loss": 0.495,
      "step": 16661
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5039525032043457,
      "learning_rate": 9.711829732398429e-06,
      "loss": 0.4089,
      "step": 16662
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5471292734146118,
      "learning_rate": 9.710811746355565e-06,
      "loss": 0.5102,
      "step": 16663
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6024495363235474,
      "learning_rate": 9.709793763312044e-06,
      "loss": 0.464,
      "step": 16664
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5199933052062988,
      "learning_rate": 9.708775783278423e-06,
      "loss": 0.5043,
      "step": 16665
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.503429651260376,
      "learning_rate": 9.707757806265265e-06,
      "loss": 0.408,
      "step": 16666
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7072147130966187,
      "learning_rate": 9.706739832283127e-06,
      "loss": 0.4798,
      "step": 16667
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5111466646194458,
      "learning_rate": 9.705721861342563e-06,
      "loss": 0.4229,
      "step": 16668
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6877546310424805,
      "learning_rate": 9.704703893454136e-06,
      "loss": 0.4392,
      "step": 16669
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8246121406555176,
      "learning_rate": 9.703685928628399e-06,
      "loss": 0.4087,
      "step": 16670
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6085789203643799,
      "learning_rate": 9.702667966875918e-06,
      "loss": 0.4927,
      "step": 16671
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4835745096206665,
      "learning_rate": 9.701650008207243e-06,
      "loss": 0.3913,
      "step": 16672
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7269340753555298,
      "learning_rate": 9.700632052632932e-06,
      "loss": 0.5179,
      "step": 16673
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4828495979309082,
      "learning_rate": 9.699614100163548e-06,
      "loss": 0.4268,
      "step": 16674
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4941251277923584,
      "learning_rate": 9.698596150809649e-06,
      "loss": 0.4332,
      "step": 16675
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5476880073547363,
      "learning_rate": 9.697578204581784e-06,
      "loss": 0.433,
      "step": 16676
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5280200242996216,
      "learning_rate": 9.696560261490518e-06,
      "loss": 0.4512,
      "step": 16677
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5339736938476562,
      "learning_rate": 9.695542321546408e-06,
      "loss": 0.4273,
      "step": 16678
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5194413661956787,
      "learning_rate": 9.694524384760013e-06,
      "loss": 0.4261,
      "step": 16679
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.540808081626892,
      "learning_rate": 9.693506451141887e-06,
      "loss": 0.3873,
      "step": 16680
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5652611255645752,
      "learning_rate": 9.692488520702587e-06,
      "loss": 0.4445,
      "step": 16681
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.437142252922058,
      "learning_rate": 9.691470593452677e-06,
      "loss": 0.3987,
      "step": 16682
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5578744411468506,
      "learning_rate": 9.69045266940271e-06,
      "loss": 0.4482,
      "step": 16683
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5681357383728027,
      "learning_rate": 9.68943474856324e-06,
      "loss": 0.4318,
      "step": 16684
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.61354398727417,
      "learning_rate": 9.688416830944831e-06,
      "loss": 0.4983,
      "step": 16685
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4920216798782349,
      "learning_rate": 9.687398916558034e-06,
      "loss": 0.4145,
      "step": 16686
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5314794778823853,
      "learning_rate": 9.686381005413419e-06,
      "loss": 0.461,
      "step": 16687
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5654903650283813,
      "learning_rate": 9.685363097521527e-06,
      "loss": 0.4067,
      "step": 16688
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5141522884368896,
      "learning_rate": 9.684345192892924e-06,
      "loss": 0.4928,
      "step": 16689
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4525372982025146,
      "learning_rate": 9.683327291538164e-06,
      "loss": 0.4,
      "step": 16690
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6099116802215576,
      "learning_rate": 9.682309393467812e-06,
      "loss": 0.4793,
      "step": 16691
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5203590393066406,
      "learning_rate": 9.681291498692418e-06,
      "loss": 0.4204,
      "step": 16692
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5666559934616089,
      "learning_rate": 9.680273607222537e-06,
      "loss": 0.4458,
      "step": 16693
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.508671760559082,
      "learning_rate": 9.679255719068731e-06,
      "loss": 0.4085,
      "step": 16694
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5390443801879883,
      "learning_rate": 9.678237834241562e-06,
      "loss": 0.4958,
      "step": 16695
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5541191101074219,
      "learning_rate": 9.677219952751575e-06,
      "loss": 0.4136,
      "step": 16696
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3974946737289429,
      "learning_rate": 9.676202074609335e-06,
      "loss": 0.4669,
      "step": 16697
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5687085390090942,
      "learning_rate": 9.675184199825394e-06,
      "loss": 0.4391,
      "step": 16698
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6139322519302368,
      "learning_rate": 9.674166328410318e-06,
      "loss": 0.4979,
      "step": 16699
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.575239658355713,
      "learning_rate": 9.673148460374657e-06,
      "loss": 0.4241,
      "step": 16700
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.560180902481079,
      "learning_rate": 9.672130595728965e-06,
      "loss": 0.4817,
      "step": 16701
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5040929317474365,
      "learning_rate": 9.671112734483804e-06,
      "loss": 0.4301,
      "step": 16702
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5724639892578125,
      "learning_rate": 9.670094876649737e-06,
      "loss": 0.451,
      "step": 16703
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5967568159103394,
      "learning_rate": 9.669077022237306e-06,
      "loss": 0.4204,
      "step": 16704
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5074480772018433,
      "learning_rate": 9.668059171257076e-06,
      "loss": 0.4319,
      "step": 16705
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4489891529083252,
      "learning_rate": 9.667041323719604e-06,
      "loss": 0.3836,
      "step": 16706
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.609156608581543,
      "learning_rate": 9.666023479635449e-06,
      "loss": 0.4822,
      "step": 16707
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5915279388427734,
      "learning_rate": 9.665005639015162e-06,
      "loss": 0.4222,
      "step": 16708
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5102747678756714,
      "learning_rate": 9.663987801869302e-06,
      "loss": 0.4676,
      "step": 16709
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.527829885482788,
      "learning_rate": 9.662969968208426e-06,
      "loss": 0.4065,
      "step": 16710
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5736325979232788,
      "learning_rate": 9.661952138043094e-06,
      "loss": 0.4503,
      "step": 16711
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5135247707366943,
      "learning_rate": 9.660934311383852e-06,
      "loss": 0.4333,
      "step": 16712
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4854038953781128,
      "learning_rate": 9.659916488241267e-06,
      "loss": 0.4347,
      "step": 16713
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5594289302825928,
      "learning_rate": 9.658898668625892e-06,
      "loss": 0.4888,
      "step": 16714
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.523966908454895,
      "learning_rate": 9.657880852548285e-06,
      "loss": 0.4219,
      "step": 16715
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5817694664001465,
      "learning_rate": 9.656863040019e-06,
      "loss": 0.4264,
      "step": 16716
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9989134073257446,
      "learning_rate": 9.65584523104859e-06,
      "loss": 0.5018,
      "step": 16717
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4860607385635376,
      "learning_rate": 9.654827425647619e-06,
      "loss": 0.4237,
      "step": 16718
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5665866136550903,
      "learning_rate": 9.65380962382664e-06,
      "loss": 0.4535,
      "step": 16719
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5338332653045654,
      "learning_rate": 9.652791825596206e-06,
      "loss": 0.4006,
      "step": 16720
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5537166595458984,
      "learning_rate": 9.651774030966878e-06,
      "loss": 0.4306,
      "step": 16721
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5457985401153564,
      "learning_rate": 9.650756239949206e-06,
      "loss": 0.4216,
      "step": 16722
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.584949016571045,
      "learning_rate": 9.649738452553757e-06,
      "loss": 0.4406,
      "step": 16723
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4948906898498535,
      "learning_rate": 9.648720668791077e-06,
      "loss": 0.3991,
      "step": 16724
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.586316704750061,
      "learning_rate": 9.647702888671725e-06,
      "loss": 0.4549,
      "step": 16725
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7329350709915161,
      "learning_rate": 9.646685112206257e-06,
      "loss": 0.4388,
      "step": 16726
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5353078842163086,
      "learning_rate": 9.645667339405232e-06,
      "loss": 0.4845,
      "step": 16727
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.576425313949585,
      "learning_rate": 9.644649570279202e-06,
      "loss": 0.434,
      "step": 16728
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5255221128463745,
      "learning_rate": 9.643631804838723e-06,
      "loss": 0.4527,
      "step": 16729
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.532139778137207,
      "learning_rate": 9.642614043094353e-06,
      "loss": 0.421,
      "step": 16730
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5207040309906006,
      "learning_rate": 9.641596285056648e-06,
      "loss": 0.4427,
      "step": 16731
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5168912410736084,
      "learning_rate": 9.64057853073616e-06,
      "loss": 0.4182,
      "step": 16732
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5009238719940186,
      "learning_rate": 9.639560780143449e-06,
      "loss": 0.4225,
      "step": 16733
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5029072761535645,
      "learning_rate": 9.638543033289066e-06,
      "loss": 0.4,
      "step": 16734
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9291255474090576,
      "learning_rate": 9.637525290183576e-06,
      "loss": 0.4874,
      "step": 16735
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5775216817855835,
      "learning_rate": 9.636507550837524e-06,
      "loss": 0.4422,
      "step": 16736
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.592725157737732,
      "learning_rate": 9.63548981526147e-06,
      "loss": 0.4774,
      "step": 16737
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5580146312713623,
      "learning_rate": 9.63447208346597e-06,
      "loss": 0.4335,
      "step": 16738
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.495336890220642,
      "learning_rate": 9.63345435546158e-06,
      "loss": 0.4338,
      "step": 16739
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5249879360198975,
      "learning_rate": 9.632436631258854e-06,
      "loss": 0.4304,
      "step": 16740
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.502915382385254,
      "learning_rate": 9.631418910868347e-06,
      "loss": 0.4668,
      "step": 16741
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4731864929199219,
      "learning_rate": 9.630401194300615e-06,
      "loss": 0.4295,
      "step": 16742
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8551371097564697,
      "learning_rate": 9.629383481566216e-06,
      "loss": 1.0635,
      "step": 16743
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.456580400466919,
      "learning_rate": 9.628365772675701e-06,
      "loss": 1.0715,
      "step": 16744
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.558475375175476,
      "learning_rate": 9.627348067639628e-06,
      "loss": 0.437,
      "step": 16745
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.685502290725708,
      "learning_rate": 9.62633036646855e-06,
      "loss": 0.4303,
      "step": 16746
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6756291389465332,
      "learning_rate": 9.625312669173029e-06,
      "loss": 0.4568,
      "step": 16747
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5750490427017212,
      "learning_rate": 9.624294975763607e-06,
      "loss": 0.4245,
      "step": 16748
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5120959281921387,
      "learning_rate": 9.623277286250851e-06,
      "loss": 0.4415,
      "step": 16749
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5423612594604492,
      "learning_rate": 9.622259600645309e-06,
      "loss": 0.4158,
      "step": 16750
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.671444296836853,
      "learning_rate": 9.621241918957545e-06,
      "loss": 0.4566,
      "step": 16751
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5514570474624634,
      "learning_rate": 9.620224241198104e-06,
      "loss": 0.3971,
      "step": 16752
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5641510486602783,
      "learning_rate": 9.619206567377543e-06,
      "loss": 0.4655,
      "step": 16753
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5659829378128052,
      "learning_rate": 9.618188897506422e-06,
      "loss": 0.3985,
      "step": 16754
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8313353061676025,
      "learning_rate": 9.617171231595293e-06,
      "loss": 0.9706,
      "step": 16755
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5144598484039307,
      "learning_rate": 9.616153569654708e-06,
      "loss": 1.068,
      "step": 16756
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5999383926391602,
      "learning_rate": 9.615135911695225e-06,
      "loss": 0.4701,
      "step": 16757
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5547971725463867,
      "learning_rate": 9.614118257727398e-06,
      "loss": 0.4073,
      "step": 16758
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7315216064453125,
      "learning_rate": 9.613100607761786e-06,
      "loss": 0.5176,
      "step": 16759
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.530771017074585,
      "learning_rate": 9.612082961808935e-06,
      "loss": 0.414,
      "step": 16760
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5894182920455933,
      "learning_rate": 9.611065319879402e-06,
      "loss": 0.4819,
      "step": 16761
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5516468286514282,
      "learning_rate": 9.610047681983747e-06,
      "loss": 0.4256,
      "step": 16762
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6592782735824585,
      "learning_rate": 9.609030048132523e-06,
      "loss": 0.4721,
      "step": 16763
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5884215831756592,
      "learning_rate": 9.60801241833628e-06,
      "loss": 0.4468,
      "step": 16764
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7451387643814087,
      "learning_rate": 9.606994792605573e-06,
      "loss": 0.46,
      "step": 16765
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5189847946166992,
      "learning_rate": 9.605977170950959e-06,
      "loss": 0.4145,
      "step": 16766
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5151842832565308,
      "learning_rate": 9.604959553382997e-06,
      "loss": 0.4735,
      "step": 16767
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5154836177825928,
      "learning_rate": 9.603941939912231e-06,
      "loss": 0.4181,
      "step": 16768
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7084381580352783,
      "learning_rate": 9.602924330549222e-06,
      "loss": 1.012,
      "step": 16769
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4678664207458496,
      "learning_rate": 9.601906725304521e-06,
      "loss": 1.0268,
      "step": 16770
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.625142216682434,
      "learning_rate": 9.600889124188687e-06,
      "loss": 0.4654,
      "step": 16771
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.682686448097229,
      "learning_rate": 9.59987152721227e-06,
      "loss": 0.4111,
      "step": 16772
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.53640615940094,
      "learning_rate": 9.598853934385822e-06,
      "loss": 0.4794,
      "step": 16773
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4911850690841675,
      "learning_rate": 9.597836345719902e-06,
      "loss": 0.4234,
      "step": 16774
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6223317384719849,
      "learning_rate": 9.596818761225066e-06,
      "loss": 0.5053,
      "step": 16775
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5055636167526245,
      "learning_rate": 9.595801180911858e-06,
      "loss": 0.4162,
      "step": 16776
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.481903076171875,
      "learning_rate": 9.594783604790842e-06,
      "loss": 0.4221,
      "step": 16777
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5601131916046143,
      "learning_rate": 9.593766032872566e-06,
      "loss": 0.4245,
      "step": 16778
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.571226954460144,
      "learning_rate": 9.592748465167588e-06,
      "loss": 0.4611,
      "step": 16779
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5070106983184814,
      "learning_rate": 9.59173090168646e-06,
      "loss": 0.4161,
      "step": 16780
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.66275954246521,
      "learning_rate": 9.590713342439732e-06,
      "loss": 0.5117,
      "step": 16781
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4976458549499512,
      "learning_rate": 9.589695787437963e-06,
      "loss": 0.4143,
      "step": 16782
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.623136281967163,
      "learning_rate": 9.588678236691707e-06,
      "loss": 0.4452,
      "step": 16783
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5591844320297241,
      "learning_rate": 9.587660690211511e-06,
      "loss": 0.4055,
      "step": 16784
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.433518648147583,
      "learning_rate": 9.586643148007935e-06,
      "loss": 0.4295,
      "step": 16785
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6400959491729736,
      "learning_rate": 9.58562561009153e-06,
      "loss": 0.4548,
      "step": 16786
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4438362121582031,
      "learning_rate": 9.584608076472853e-06,
      "loss": 0.4124,
      "step": 16787
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5121523141860962,
      "learning_rate": 9.583590547162452e-06,
      "loss": 0.414,
      "step": 16788
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5865671634674072,
      "learning_rate": 9.582573022170882e-06,
      "loss": 0.465,
      "step": 16789
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4934051036834717,
      "learning_rate": 9.581555501508698e-06,
      "loss": 0.4077,
      "step": 16790
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7788640260696411,
      "learning_rate": 9.580537985186456e-06,
      "loss": 0.4373,
      "step": 16791
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4961298704147339,
      "learning_rate": 9.5795204732147e-06,
      "loss": 0.402,
      "step": 16792
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.983400583267212,
      "learning_rate": 9.578502965603992e-06,
      "loss": 0.4106,
      "step": 16793
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.446624517440796,
      "learning_rate": 9.57748546236488e-06,
      "loss": 0.3905,
      "step": 16794
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5743892192840576,
      "learning_rate": 9.576467963507926e-06,
      "loss": 0.4652,
      "step": 16795
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.565688133239746,
      "learning_rate": 9.575450469043673e-06,
      "loss": 0.3887,
      "step": 16796
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5667023658752441,
      "learning_rate": 9.574432978982674e-06,
      "loss": 0.4312,
      "step": 16797
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5200098752975464,
      "learning_rate": 9.573415493335488e-06,
      "loss": 0.4165,
      "step": 16798
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5609230995178223,
      "learning_rate": 9.572398012112667e-06,
      "loss": 0.439,
      "step": 16799
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7117385864257812,
      "learning_rate": 9.57138053532476e-06,
      "loss": 0.4806,
      "step": 16800
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5598595142364502,
      "learning_rate": 9.570363062982324e-06,
      "loss": 0.4875,
      "step": 16801
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.568209171295166,
      "learning_rate": 9.569345595095907e-06,
      "loss": 0.4293,
      "step": 16802
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.597685694694519,
      "learning_rate": 9.568328131676072e-06,
      "loss": 0.4653,
      "step": 16803
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6194772720336914,
      "learning_rate": 9.567310672733358e-06,
      "loss": 0.4489,
      "step": 16804
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6470236778259277,
      "learning_rate": 9.566293218278327e-06,
      "loss": 0.4685,
      "step": 16805
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4953076839447021,
      "learning_rate": 9.565275768321524e-06,
      "loss": 0.3865,
      "step": 16806
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.528415322303772,
      "learning_rate": 9.564258322873512e-06,
      "loss": 0.4391,
      "step": 16807
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5121785402297974,
      "learning_rate": 9.563240881944838e-06,
      "loss": 0.4231,
      "step": 16808
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5764131546020508,
      "learning_rate": 9.56222344554605e-06,
      "loss": 0.4863,
      "step": 16809
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.473248839378357,
      "learning_rate": 9.561206013687707e-06,
      "loss": 0.3952,
      "step": 16810
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5839574337005615,
      "learning_rate": 9.560188586380363e-06,
      "loss": 0.4736,
      "step": 16811
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5041942596435547,
      "learning_rate": 9.559171163634562e-06,
      "loss": 0.4028,
      "step": 16812
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5401781797409058,
      "learning_rate": 9.558153745460863e-06,
      "loss": 0.4413,
      "step": 16813
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5010666847229004,
      "learning_rate": 9.557136331869815e-06,
      "loss": 0.4086,
      "step": 16814
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.605601191520691,
      "learning_rate": 9.556118922871974e-06,
      "loss": 0.4694,
      "step": 16815
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5526106357574463,
      "learning_rate": 9.555101518477887e-06,
      "loss": 0.4227,
      "step": 16816
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.546865701675415,
      "learning_rate": 9.554084118698108e-06,
      "loss": 0.4553,
      "step": 16817
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.623372197151184,
      "learning_rate": 9.553066723543191e-06,
      "loss": 0.4564,
      "step": 16818
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.636314868927002,
      "learning_rate": 9.55204933302369e-06,
      "loss": 0.4943,
      "step": 16819
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4909634590148926,
      "learning_rate": 9.551031947150148e-06,
      "loss": 0.4389,
      "step": 16820
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8276190757751465,
      "learning_rate": 9.550014565933126e-06,
      "loss": 0.9966,
      "step": 16821
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5076205730438232,
      "learning_rate": 9.548997189383168e-06,
      "loss": 1.0079,
      "step": 16822
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5708584785461426,
      "learning_rate": 9.547979817510839e-06,
      "loss": 0.4623,
      "step": 16823
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.505685806274414,
      "learning_rate": 9.546962450326677e-06,
      "loss": 0.4009,
      "step": 16824
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.683936595916748,
      "learning_rate": 9.545945087841236e-06,
      "loss": 1.0153,
      "step": 16825
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4855694770812988,
      "learning_rate": 9.544927730065074e-06,
      "loss": 1.053,
      "step": 16826
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5952366590499878,
      "learning_rate": 9.543910377008741e-06,
      "loss": 0.5012,
      "step": 16827
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.542333722114563,
      "learning_rate": 9.542893028682783e-06,
      "loss": 0.4224,
      "step": 16828
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5064635276794434,
      "learning_rate": 9.541875685097757e-06,
      "loss": 0.4772,
      "step": 16829
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4815597534179688,
      "learning_rate": 9.54085834626421e-06,
      "loss": 0.3983,
      "step": 16830
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.581286907196045,
      "learning_rate": 9.5398410121927e-06,
      "loss": 0.4443,
      "step": 16831
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.501064658164978,
      "learning_rate": 9.538823682893775e-06,
      "loss": 0.4115,
      "step": 16832
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6220968961715698,
      "learning_rate": 9.537806358377982e-06,
      "loss": 0.4776,
      "step": 16833
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5608960390090942,
      "learning_rate": 9.536789038655878e-06,
      "loss": 0.3792,
      "step": 16834
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5331249237060547,
      "learning_rate": 9.535771723738015e-06,
      "loss": 0.4385,
      "step": 16835
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5469379425048828,
      "learning_rate": 9.534754413634936e-06,
      "loss": 0.4252,
      "step": 16836
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6580334901809692,
      "learning_rate": 9.533737108357202e-06,
      "loss": 0.4816,
      "step": 16837
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5094071626663208,
      "learning_rate": 9.532719807915356e-06,
      "loss": 0.4396,
      "step": 16838
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5102895498275757,
      "learning_rate": 9.531702512319958e-06,
      "loss": 0.456,
      "step": 16839
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.585133671760559,
      "learning_rate": 9.530685221581551e-06,
      "loss": 0.4093,
      "step": 16840
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6567941904067993,
      "learning_rate": 9.529667935710688e-06,
      "loss": 0.4695,
      "step": 16841
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.466615080833435,
      "learning_rate": 9.528650654717922e-06,
      "loss": 0.4169,
      "step": 16842
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.602303385734558,
      "learning_rate": 9.527633378613804e-06,
      "loss": 0.4857,
      "step": 16843
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4942442178726196,
      "learning_rate": 9.52661610740888e-06,
      "loss": 0.4122,
      "step": 16844
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5333787202835083,
      "learning_rate": 9.525598841113703e-06,
      "loss": 0.4649,
      "step": 16845
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5918607711791992,
      "learning_rate": 9.524581579738827e-06,
      "loss": 0.4132,
      "step": 16846
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6960080862045288,
      "learning_rate": 9.523564323294803e-06,
      "loss": 0.5182,
      "step": 16847
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6138606071472168,
      "learning_rate": 9.522547071792174e-06,
      "loss": 0.4052,
      "step": 16848
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6902000904083252,
      "learning_rate": 9.521529825241496e-06,
      "loss": 1.0215,
      "step": 16849
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5638105869293213,
      "learning_rate": 9.520512583653318e-06,
      "loss": 1.0239,
      "step": 16850
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6548726558685303,
      "learning_rate": 9.519495347038196e-06,
      "loss": 0.5077,
      "step": 16851
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5013432502746582,
      "learning_rate": 9.518478115406674e-06,
      "loss": 0.4099,
      "step": 16852
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.627419114112854,
      "learning_rate": 9.5174608887693e-06,
      "loss": 0.4851,
      "step": 16853
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6007487773895264,
      "learning_rate": 9.51644366713663e-06,
      "loss": 0.4181,
      "step": 16854
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5137485265731812,
      "learning_rate": 9.515426450519218e-06,
      "loss": 0.4731,
      "step": 16855
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.522782802581787,
      "learning_rate": 9.5144092389276e-06,
      "loss": 0.4103,
      "step": 16856
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5510896444320679,
      "learning_rate": 9.51339203237234e-06,
      "loss": 0.5113,
      "step": 16857
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6281534433364868,
      "learning_rate": 9.512374830863979e-06,
      "loss": 0.4725,
      "step": 16858
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.607545018196106,
      "learning_rate": 9.511357634413077e-06,
      "loss": 0.5686,
      "step": 16859
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5676876306533813,
      "learning_rate": 9.510340443030174e-06,
      "loss": 0.4341,
      "step": 16860
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5824569463729858,
      "learning_rate": 9.50932325672582e-06,
      "loss": 0.4717,
      "step": 16861
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5871963500976562,
      "learning_rate": 9.508306075510573e-06,
      "loss": 0.43,
      "step": 16862
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5580112934112549,
      "learning_rate": 9.507288899394976e-06,
      "loss": 0.4697,
      "step": 16863
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5378108024597168,
      "learning_rate": 9.506271728389583e-06,
      "loss": 0.4223,
      "step": 16864
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6216050386428833,
      "learning_rate": 9.505254562504942e-06,
      "loss": 0.48,
      "step": 16865
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6592410802841187,
      "learning_rate": 9.504237401751599e-06,
      "loss": 0.4277,
      "step": 16866
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5616018772125244,
      "learning_rate": 9.50322024614011e-06,
      "loss": 0.4409,
      "step": 16867
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6147384643554688,
      "learning_rate": 9.502203095681021e-06,
      "loss": 0.4714,
      "step": 16868
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6227952241897583,
      "learning_rate": 9.50118595038488e-06,
      "loss": 0.5009,
      "step": 16869
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5558820962905884,
      "learning_rate": 9.50016881026224e-06,
      "loss": 0.4521,
      "step": 16870
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5213669538497925,
      "learning_rate": 9.499151675323646e-06,
      "loss": 0.4307,
      "step": 16871
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4973663091659546,
      "learning_rate": 9.498134545579655e-06,
      "loss": 0.4054,
      "step": 16872
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5876251459121704,
      "learning_rate": 9.49711742104081e-06,
      "loss": 0.4704,
      "step": 16873
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4657418727874756,
      "learning_rate": 9.49610030171766e-06,
      "loss": 0.3928,
      "step": 16874
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5070476531982422,
      "learning_rate": 9.495083187620756e-06,
      "loss": 0.428,
      "step": 16875
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5985199213027954,
      "learning_rate": 9.494066078760647e-06,
      "loss": 0.4212,
      "step": 16876
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7962723970413208,
      "learning_rate": 9.49304897514788e-06,
      "loss": 0.4565,
      "step": 16877
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.85061776638031,
      "learning_rate": 9.492031876793007e-06,
      "loss": 0.4082,
      "step": 16878
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6103624105453491,
      "learning_rate": 9.491014783706575e-06,
      "loss": 0.4242,
      "step": 16879
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5445153713226318,
      "learning_rate": 9.48999769589914e-06,
      "loss": 0.3812,
      "step": 16880
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6249566078186035,
      "learning_rate": 9.488980613381235e-06,
      "loss": 0.4436,
      "step": 16881
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4702489376068115,
      "learning_rate": 9.487963536163424e-06,
      "loss": 0.4022,
      "step": 16882
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5616999864578247,
      "learning_rate": 9.486946464256246e-06,
      "loss": 0.4435,
      "step": 16883
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.574537992477417,
      "learning_rate": 9.485929397670257e-06,
      "loss": 0.417,
      "step": 16884
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5664360523223877,
      "learning_rate": 9.484912336416002e-06,
      "loss": 0.4833,
      "step": 16885
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5203503370285034,
      "learning_rate": 9.483895280504026e-06,
      "loss": 0.4125,
      "step": 16886
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5479117631912231,
      "learning_rate": 9.482878229944885e-06,
      "loss": 0.4598,
      "step": 16887
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5129587650299072,
      "learning_rate": 9.481861184749126e-06,
      "loss": 0.4224,
      "step": 16888
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4952532052993774,
      "learning_rate": 9.480844144927288e-06,
      "loss": 0.4456,
      "step": 16889
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4761772155761719,
      "learning_rate": 9.47982711048993e-06,
      "loss": 0.4055,
      "step": 16890
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5007829666137695,
      "learning_rate": 9.478810081447595e-06,
      "loss": 0.4763,
      "step": 16891
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5561425685882568,
      "learning_rate": 9.477793057810837e-06,
      "loss": 0.4333,
      "step": 16892
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5465736389160156,
      "learning_rate": 9.476776039590198e-06,
      "loss": 0.4491,
      "step": 16893
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4813456535339355,
      "learning_rate": 9.475759026796226e-06,
      "loss": 0.4017,
      "step": 16894
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6422325372695923,
      "learning_rate": 9.474742019439473e-06,
      "loss": 0.5391,
      "step": 16895
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5264954566955566,
      "learning_rate": 9.473725017530487e-06,
      "loss": 0.4071,
      "step": 16896
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6274389028549194,
      "learning_rate": 9.472708021079811e-06,
      "loss": 0.4771,
      "step": 16897
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.570119023323059,
      "learning_rate": 9.471691030097998e-06,
      "loss": 0.4253,
      "step": 16898
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.469658374786377,
      "learning_rate": 9.47067404459559e-06,
      "loss": 0.4584,
      "step": 16899
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4455573558807373,
      "learning_rate": 9.469657064583144e-06,
      "loss": 0.3826,
      "step": 16900
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.787156581878662,
      "learning_rate": 9.4686400900712e-06,
      "loss": 0.453,
      "step": 16901
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4770958423614502,
      "learning_rate": 9.467623121070306e-06,
      "loss": 0.3821,
      "step": 16902
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5574123859405518,
      "learning_rate": 9.466606157591013e-06,
      "loss": 0.4601,
      "step": 16903
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5984673500061035,
      "learning_rate": 9.465589199643871e-06,
      "loss": 0.419,
      "step": 16904
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5253719091415405,
      "learning_rate": 9.464572247239418e-06,
      "loss": 0.4505,
      "step": 16905
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4915549755096436,
      "learning_rate": 9.463555300388209e-06,
      "loss": 0.4038,
      "step": 16906
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5641167163848877,
      "learning_rate": 9.462538359100789e-06,
      "loss": 0.4618,
      "step": 16907
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5500822067260742,
      "learning_rate": 9.461521423387708e-06,
      "loss": 0.4179,
      "step": 16908
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6804832220077515,
      "learning_rate": 9.460504493259508e-06,
      "loss": 0.518,
      "step": 16909
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6953827142715454,
      "learning_rate": 9.459487568726738e-06,
      "loss": 0.4594,
      "step": 16910
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.710121750831604,
      "learning_rate": 9.458470649799949e-06,
      "loss": 0.5871,
      "step": 16911
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.530713677406311,
      "learning_rate": 9.457453736489688e-06,
      "loss": 0.4148,
      "step": 16912
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.701988697052002,
      "learning_rate": 9.456436828806494e-06,
      "loss": 0.4795,
      "step": 16913
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4613838195800781,
      "learning_rate": 9.45541992676092e-06,
      "loss": 0.3919,
      "step": 16914
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5550227165222168,
      "learning_rate": 9.454403030363512e-06,
      "loss": 0.4513,
      "step": 16915
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5354926586151123,
      "learning_rate": 9.453386139624821e-06,
      "loss": 0.3743,
      "step": 16916
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.60630464553833,
      "learning_rate": 9.452369254555388e-06,
      "loss": 0.4553,
      "step": 16917
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.565114974975586,
      "learning_rate": 9.451352375165759e-06,
      "loss": 0.4299,
      "step": 16918
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6291756629943848,
      "learning_rate": 9.450335501466485e-06,
      "loss": 0.4594,
      "step": 16919
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.7034547328948975,
      "learning_rate": 9.449318633468113e-06,
      "loss": 0.4217,
      "step": 16920
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5458589792251587,
      "learning_rate": 9.448301771181186e-06,
      "loss": 0.4588,
      "step": 16921
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5377813577651978,
      "learning_rate": 9.447284914616249e-06,
      "loss": 0.4002,
      "step": 16922
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6184881925582886,
      "learning_rate": 9.446268063783853e-06,
      "loss": 0.4945,
      "step": 16923
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6310313940048218,
      "learning_rate": 9.445251218694546e-06,
      "loss": 0.4089,
      "step": 16924
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5572599172592163,
      "learning_rate": 9.444234379358865e-06,
      "loss": 0.429,
      "step": 16925
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5791354179382324,
      "learning_rate": 9.443217545787366e-06,
      "loss": 0.4177,
      "step": 16926
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6495157480239868,
      "learning_rate": 9.442200717990589e-06,
      "loss": 0.9807,
      "step": 16927
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.545323371887207,
      "learning_rate": 9.441183895979085e-06,
      "loss": 1.0061,
      "step": 16928
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6538077592849731,
      "learning_rate": 9.440167079763398e-06,
      "loss": 0.4716,
      "step": 16929
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5644301176071167,
      "learning_rate": 9.439150269354071e-06,
      "loss": 0.4566,
      "step": 16930
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6806186437606812,
      "learning_rate": 9.438133464761653e-06,
      "loss": 0.5143,
      "step": 16931
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5515716075897217,
      "learning_rate": 9.437116665996692e-06,
      "loss": 0.4021,
      "step": 16932
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.525781512260437,
      "learning_rate": 9.436099873069729e-06,
      "loss": 0.4406,
      "step": 16933
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.515753149986267,
      "learning_rate": 9.435083085991312e-06,
      "loss": 0.4026,
      "step": 16934
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5970847606658936,
      "learning_rate": 9.434066304771986e-06,
      "loss": 0.483,
      "step": 16935
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5456938743591309,
      "learning_rate": 9.4330495294223e-06,
      "loss": 0.4183,
      "step": 16936
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5731358528137207,
      "learning_rate": 9.432032759952796e-06,
      "loss": 0.4558,
      "step": 16937
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5376733541488647,
      "learning_rate": 9.431015996374018e-06,
      "loss": 0.3923,
      "step": 16938
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6119906902313232,
      "learning_rate": 9.429999238696517e-06,
      "loss": 0.5449,
      "step": 16939
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5545423030853271,
      "learning_rate": 9.428982486930837e-06,
      "loss": 0.4194,
      "step": 16940
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5117812156677246,
      "learning_rate": 9.427965741087517e-06,
      "loss": 0.4473,
      "step": 16941
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.457926869392395,
      "learning_rate": 9.42694900117711e-06,
      "loss": 0.3946,
      "step": 16942
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5002485513687134,
      "learning_rate": 9.425932267210155e-06,
      "loss": 0.4255,
      "step": 16943
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5934593677520752,
      "learning_rate": 9.424915539197204e-06,
      "loss": 0.4091,
      "step": 16944
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6452635526657104,
      "learning_rate": 9.423898817148799e-06,
      "loss": 0.4635,
      "step": 16945
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.526025652885437,
      "learning_rate": 9.42288210107548e-06,
      "loss": 0.4227,
      "step": 16946
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4910531044006348,
      "learning_rate": 9.4218653909878e-06,
      "loss": 0.4424,
      "step": 16947
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6047070026397705,
      "learning_rate": 9.420848686896303e-06,
      "loss": 0.4406,
      "step": 16948
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5968927145004272,
      "learning_rate": 9.419831988811525e-06,
      "loss": 0.4462,
      "step": 16949
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5039023160934448,
      "learning_rate": 9.41881529674402e-06,
      "loss": 0.4125,
      "step": 16950
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5751733779907227,
      "learning_rate": 9.417798610704327e-06,
      "loss": 0.438,
      "step": 16951
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5010370016098022,
      "learning_rate": 9.416781930702998e-06,
      "loss": 0.4312,
      "step": 16952
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6085057258605957,
      "learning_rate": 9.415765256750572e-06,
      "loss": 0.5223,
      "step": 16953
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4939967393875122,
      "learning_rate": 9.414748588857592e-06,
      "loss": 0.3857,
      "step": 16954
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.565995693206787,
      "learning_rate": 9.413731927034607e-06,
      "loss": 0.4925,
      "step": 16955
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.550049901008606,
      "learning_rate": 9.41271527129216e-06,
      "loss": 0.4205,
      "step": 16956
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4792792797088623,
      "learning_rate": 9.411698621640795e-06,
      "loss": 0.4269,
      "step": 16957
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.570736289024353,
      "learning_rate": 9.410681978091053e-06,
      "loss": 0.437,
      "step": 16958
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5197265148162842,
      "learning_rate": 9.40966534065348e-06,
      "loss": 0.4609,
      "step": 16959
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.505760669708252,
      "learning_rate": 9.408648709338629e-06,
      "loss": 0.4072,
      "step": 16960
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5239346027374268,
      "learning_rate": 9.40763208415703e-06,
      "loss": 0.4684,
      "step": 16961
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5108225345611572,
      "learning_rate": 9.406615465119236e-06,
      "loss": 0.3918,
      "step": 16962
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6009374856948853,
      "learning_rate": 9.405598852235786e-06,
      "loss": 0.5285,
      "step": 16963
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4888548851013184,
      "learning_rate": 9.40458224551723e-06,
      "loss": 0.4236,
      "step": 16964
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4804073572158813,
      "learning_rate": 9.403565644974107e-06,
      "loss": 0.4639,
      "step": 16965
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5146762132644653,
      "learning_rate": 9.402549050616958e-06,
      "loss": 0.4164,
      "step": 16966
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4955989122390747,
      "learning_rate": 9.401532462456336e-06,
      "loss": 0.4174,
      "step": 16967
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6503894329071045,
      "learning_rate": 9.40051588050278e-06,
      "loss": 0.4572,
      "step": 16968
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4782311916351318,
      "learning_rate": 9.399499304766828e-06,
      "loss": 0.437,
      "step": 16969
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5105351209640503,
      "learning_rate": 9.39848273525903e-06,
      "loss": 0.4093,
      "step": 16970
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6224656105041504,
      "learning_rate": 9.397466171989928e-06,
      "loss": 0.5072,
      "step": 16971
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4911445379257202,
      "learning_rate": 9.39644961497007e-06,
      "loss": 0.4151,
      "step": 16972
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6302651166915894,
      "learning_rate": 9.39543306420999e-06,
      "loss": 0.4737,
      "step": 16973
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4690477848052979,
      "learning_rate": 9.394416519720236e-06,
      "loss": 0.4055,
      "step": 16974
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5797580480575562,
      "learning_rate": 9.393399981511352e-06,
      "loss": 0.4658,
      "step": 16975
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5278701782226562,
      "learning_rate": 9.39238344959388e-06,
      "loss": 0.4171,
      "step": 16976
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5244114398956299,
      "learning_rate": 9.391366923978362e-06,
      "loss": 0.4359,
      "step": 16977
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5988857746124268,
      "learning_rate": 9.390350404675343e-06,
      "loss": 0.434,
      "step": 16978
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.516263484954834,
      "learning_rate": 9.389333891695365e-06,
      "loss": 0.4313,
      "step": 16979
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5083919763565063,
      "learning_rate": 9.388317385048973e-06,
      "loss": 0.4174,
      "step": 16980
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5078121423721313,
      "learning_rate": 9.387300884746707e-06,
      "loss": 0.4943,
      "step": 16981
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4996838569641113,
      "learning_rate": 9.38628439079911e-06,
      "loss": 0.3946,
      "step": 16982
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4608012437820435,
      "learning_rate": 9.385267903216723e-06,
      "loss": 0.497,
      "step": 16983
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5275802612304688,
      "learning_rate": 9.384251422010097e-06,
      "loss": 0.3976,
      "step": 16984
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5941476821899414,
      "learning_rate": 9.383234947189762e-06,
      "loss": 0.4439,
      "step": 16985
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4584120512008667,
      "learning_rate": 9.38221847876627e-06,
      "loss": 0.3968,
      "step": 16986
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4989362955093384,
      "learning_rate": 9.381202016750158e-06,
      "loss": 0.4483,
      "step": 16987
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.568158507347107,
      "learning_rate": 9.380185561151975e-06,
      "loss": 0.4524,
      "step": 16988
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.620574951171875,
      "learning_rate": 9.379169111982255e-06,
      "loss": 0.4763,
      "step": 16989
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.701685905456543,
      "learning_rate": 9.378152669251545e-06,
      "loss": 0.4673,
      "step": 16990
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5544370412826538,
      "learning_rate": 9.377136232970387e-06,
      "loss": 0.5109,
      "step": 16991
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4825270175933838,
      "learning_rate": 9.376119803149322e-06,
      "loss": 0.4118,
      "step": 16992
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4512451887130737,
      "learning_rate": 9.375103379798892e-06,
      "loss": 0.4283,
      "step": 16993
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5383445024490356,
      "learning_rate": 9.374086962929639e-06,
      "loss": 0.434,
      "step": 16994
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.558784008026123,
      "learning_rate": 9.373070552552102e-06,
      "loss": 0.448,
      "step": 16995
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5595016479492188,
      "learning_rate": 9.372054148676834e-06,
      "loss": 0.4283,
      "step": 16996
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8037155866622925,
      "learning_rate": 9.371037751314362e-06,
      "loss": 0.4685,
      "step": 16997
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5281281471252441,
      "learning_rate": 9.370021360475236e-06,
      "loss": 0.4135,
      "step": 16998
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5847216844558716,
      "learning_rate": 9.369004976169995e-06,
      "loss": 0.4457,
      "step": 16999
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5284208059310913,
      "learning_rate": 9.367988598409185e-06,
      "loss": 0.3951,
      "step": 17000
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5172241926193237,
      "learning_rate": 9.366972227203341e-06,
      "loss": 0.5113,
      "step": 17001
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4543793201446533,
      "learning_rate": 9.365955862563005e-06,
      "loss": 0.3827,
      "step": 17002
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.590262532234192,
      "learning_rate": 9.364939504498723e-06,
      "loss": 0.4464,
      "step": 17003
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5114518404006958,
      "learning_rate": 9.363923153021037e-06,
      "loss": 0.4071,
      "step": 17004
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4718289375305176,
      "learning_rate": 9.36290680814048e-06,
      "loss": 0.4372,
      "step": 17005
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6142432689666748,
      "learning_rate": 9.3618904698676e-06,
      "loss": 0.4533,
      "step": 17006
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.616923451423645,
      "learning_rate": 9.360874138212935e-06,
      "loss": 0.48,
      "step": 17007
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6596530675888062,
      "learning_rate": 9.359857813187031e-06,
      "loss": 0.4099,
      "step": 17008
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.2267777919769287,
      "learning_rate": 9.358841494800423e-06,
      "loss": 0.4225,
      "step": 17009
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6088716983795166,
      "learning_rate": 9.357825183063651e-06,
      "loss": 0.4337,
      "step": 17010
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6097973585128784,
      "learning_rate": 9.356808877987262e-06,
      "loss": 0.5663,
      "step": 17011
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4657756090164185,
      "learning_rate": 9.355792579581795e-06,
      "loss": 0.4094,
      "step": 17012
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5892454385757446,
      "learning_rate": 9.354776287857786e-06,
      "loss": 0.4674,
      "step": 17013
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4934731721878052,
      "learning_rate": 9.353760002825779e-06,
      "loss": 0.423,
      "step": 17014
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6360641717910767,
      "learning_rate": 9.352743724496313e-06,
      "loss": 0.4593,
      "step": 17015
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5695585012435913,
      "learning_rate": 9.351727452879934e-06,
      "loss": 0.4228,
      "step": 17016
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6227922439575195,
      "learning_rate": 9.350711187987176e-06,
      "loss": 0.4999,
      "step": 17017
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5656095743179321,
      "learning_rate": 9.349694929828578e-06,
      "loss": 0.4229,
      "step": 17018
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5988738536834717,
      "learning_rate": 9.348678678414687e-06,
      "loss": 0.5063,
      "step": 17019
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6197807788848877,
      "learning_rate": 9.34766243375604e-06,
      "loss": 0.4515,
      "step": 17020
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5559769868850708,
      "learning_rate": 9.346646195863174e-06,
      "loss": 0.4717,
      "step": 17021
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5758087635040283,
      "learning_rate": 9.345629964746634e-06,
      "loss": 0.4407,
      "step": 17022
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6263519525527954,
      "learning_rate": 9.344613740416953e-06,
      "loss": 0.5014,
      "step": 17023
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4573638439178467,
      "learning_rate": 9.343597522884682e-06,
      "loss": 0.396,
      "step": 17024
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6369004249572754,
      "learning_rate": 9.342581312160351e-06,
      "loss": 0.5024,
      "step": 17025
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4216159582138062,
      "learning_rate": 9.341565108254502e-06,
      "loss": 0.3923,
      "step": 17026
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.523424506187439,
      "learning_rate": 9.340548911177678e-06,
      "loss": 0.4295,
      "step": 17027
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.598314642906189,
      "learning_rate": 9.339532720940418e-06,
      "loss": 0.4009,
      "step": 17028
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5994096994400024,
      "learning_rate": 9.338516537553255e-06,
      "loss": 0.4643,
      "step": 17029
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5366206169128418,
      "learning_rate": 9.337500361026736e-06,
      "loss": 0.3994,
      "step": 17030
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6108039617538452,
      "learning_rate": 9.336484191371396e-06,
      "loss": 0.544,
      "step": 17031
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5028451681137085,
      "learning_rate": 9.33546802859778e-06,
      "loss": 0.3893,
      "step": 17032
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.546344518661499,
      "learning_rate": 9.33445187271642e-06,
      "loss": 0.4684,
      "step": 17033
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6585460901260376,
      "learning_rate": 9.333435723737858e-06,
      "loss": 0.4207,
      "step": 17034
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6031322479248047,
      "learning_rate": 9.332419581672634e-06,
      "loss": 0.453,
      "step": 17035
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5454299449920654,
      "learning_rate": 9.331403446531289e-06,
      "loss": 0.387,
      "step": 17036
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5913724899291992,
      "learning_rate": 9.330387318324357e-06,
      "loss": 0.4911,
      "step": 17037
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4677072763442993,
      "learning_rate": 9.329371197062378e-06,
      "loss": 0.3894,
      "step": 17038
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5614129304885864,
      "learning_rate": 9.328355082755894e-06,
      "loss": 0.4525,
      "step": 17039
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5563039779663086,
      "learning_rate": 9.327338975415443e-06,
      "loss": 0.4139,
      "step": 17040
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6552391052246094,
      "learning_rate": 9.326322875051558e-06,
      "loss": 0.5104,
      "step": 17041
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.504339575767517,
      "learning_rate": 9.325306781674786e-06,
      "loss": 0.4312,
      "step": 17042
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.605921745300293,
      "learning_rate": 9.324290695295657e-06,
      "loss": 0.5189,
      "step": 17043
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5056171417236328,
      "learning_rate": 9.32327461592472e-06,
      "loss": 0.4051,
      "step": 17044
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.558750867843628,
      "learning_rate": 9.322258543572507e-06,
      "loss": 0.4712,
      "step": 17045
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5281782150268555,
      "learning_rate": 9.321242478249552e-06,
      "loss": 0.404,
      "step": 17046
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5149527788162231,
      "learning_rate": 9.3202264199664e-06,
      "loss": 0.4535,
      "step": 17047
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5699316263198853,
      "learning_rate": 9.31921036873359e-06,
      "loss": 0.3939,
      "step": 17048
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5782831907272339,
      "learning_rate": 9.318194324561652e-06,
      "loss": 0.4871,
      "step": 17049
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5986528396606445,
      "learning_rate": 9.317178287461133e-06,
      "loss": 0.4196,
      "step": 17050
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5754749774932861,
      "learning_rate": 9.316162257442562e-06,
      "loss": 0.4643,
      "step": 17051
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4305695295333862,
      "learning_rate": 9.315146234516488e-06,
      "loss": 0.3886,
      "step": 17052
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5173962116241455,
      "learning_rate": 9.314130218693441e-06,
      "loss": 0.4382,
      "step": 17053
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5488914251327515,
      "learning_rate": 9.313114209983958e-06,
      "loss": 0.4423,
      "step": 17054
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5403379201889038,
      "learning_rate": 9.312098208398582e-06,
      "loss": 0.4574,
      "step": 17055
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5493310689926147,
      "learning_rate": 9.311082213947848e-06,
      "loss": 0.4148,
      "step": 17056
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5143619775772095,
      "learning_rate": 9.310066226642289e-06,
      "loss": 0.4533,
      "step": 17057
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5795077085494995,
      "learning_rate": 9.309050246492448e-06,
      "loss": 0.4116,
      "step": 17058
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.57291579246521,
      "learning_rate": 9.30803427350886e-06,
      "loss": 0.4752,
      "step": 17059
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6141430139541626,
      "learning_rate": 9.307018307702067e-06,
      "loss": 0.448,
      "step": 17060
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5486416816711426,
      "learning_rate": 9.3060023490826e-06,
      "loss": 0.4917,
      "step": 17061
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4391099214553833,
      "learning_rate": 9.304986397660998e-06,
      "loss": 0.4073,
      "step": 17062
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5133440494537354,
      "learning_rate": 9.303970453447798e-06,
      "loss": 0.4693,
      "step": 17063
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5712538957595825,
      "learning_rate": 9.30295451645354e-06,
      "loss": 0.4487,
      "step": 17064
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6054471731185913,
      "learning_rate": 9.301938586688755e-06,
      "loss": 0.4933,
      "step": 17065
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4588350057601929,
      "learning_rate": 9.300922664163985e-06,
      "loss": 0.389,
      "step": 17066
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7959660291671753,
      "learning_rate": 9.299906748889762e-06,
      "loss": 0.9986,
      "step": 17067
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5242093801498413,
      "learning_rate": 9.298890840876631e-06,
      "loss": 1.003,
      "step": 17068
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5253254175186157,
      "learning_rate": 9.297874940135121e-06,
      "loss": 0.4915,
      "step": 17069
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5703470706939697,
      "learning_rate": 9.296859046675768e-06,
      "loss": 0.4115,
      "step": 17070
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5996350049972534,
      "learning_rate": 9.295843160509112e-06,
      "loss": 0.4639,
      "step": 17071
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5014028549194336,
      "learning_rate": 9.294827281645692e-06,
      "loss": 0.4303,
      "step": 17072
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6346701383590698,
      "learning_rate": 9.293811410096039e-06,
      "loss": 0.4907,
      "step": 17073
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7096694707870483,
      "learning_rate": 9.292795545870688e-06,
      "loss": 0.4023,
      "step": 17074
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5831592082977295,
      "learning_rate": 9.29177968898018e-06,
      "loss": 0.44,
      "step": 17075
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5140599012374878,
      "learning_rate": 9.290763839435051e-06,
      "loss": 0.3817,
      "step": 17076
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6003787517547607,
      "learning_rate": 9.289747997245832e-06,
      "loss": 0.4877,
      "step": 17077
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.593454122543335,
      "learning_rate": 9.288732162423063e-06,
      "loss": 0.4213,
      "step": 17078
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5927314758300781,
      "learning_rate": 9.287716334977276e-06,
      "loss": 0.4464,
      "step": 17079
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5738922357559204,
      "learning_rate": 9.286700514919017e-06,
      "loss": 0.3926,
      "step": 17080
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4949547052383423,
      "learning_rate": 9.285684702258809e-06,
      "loss": 0.4312,
      "step": 17081
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.648141622543335,
      "learning_rate": 9.284668897007192e-06,
      "loss": 0.4696,
      "step": 17082
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6059279441833496,
      "learning_rate": 9.283653099174704e-06,
      "loss": 0.4154,
      "step": 17083
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5695664882659912,
      "learning_rate": 9.282637308771883e-06,
      "loss": 0.4066,
      "step": 17084
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.538065791130066,
      "learning_rate": 9.281621525809253e-06,
      "loss": 0.457,
      "step": 17085
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6094365119934082,
      "learning_rate": 9.28060575029736e-06,
      "loss": 0.4072,
      "step": 17086
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4611551761627197,
      "learning_rate": 9.279589982246734e-06,
      "loss": 0.4284,
      "step": 17087
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5650639533996582,
      "learning_rate": 9.278574221667916e-06,
      "loss": 0.4198,
      "step": 17088
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5326441526412964,
      "learning_rate": 9.277558468571434e-06,
      "loss": 0.4529,
      "step": 17089
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.522252082824707,
      "learning_rate": 9.276542722967825e-06,
      "loss": 0.4256,
      "step": 17090
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4934955835342407,
      "learning_rate": 9.275526984867626e-06,
      "loss": 0.4467,
      "step": 17091
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.541419267654419,
      "learning_rate": 9.274511254281374e-06,
      "loss": 0.4442,
      "step": 17092
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5998185873031616,
      "learning_rate": 9.273495531219594e-06,
      "loss": 0.4904,
      "step": 17093
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6077690124511719,
      "learning_rate": 9.272479815692832e-06,
      "loss": 0.4143,
      "step": 17094
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.523171067237854,
      "learning_rate": 9.271464107711614e-06,
      "loss": 0.4138,
      "step": 17095
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4508647918701172,
      "learning_rate": 9.270448407286482e-06,
      "loss": 0.4143,
      "step": 17096
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.718859076499939,
      "learning_rate": 9.269432714427965e-06,
      "loss": 0.4667,
      "step": 17097
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4820423126220703,
      "learning_rate": 9.268417029146598e-06,
      "loss": 0.4003,
      "step": 17098
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6109579801559448,
      "learning_rate": 9.267401351452916e-06,
      "loss": 0.4512,
      "step": 17099
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4941496849060059,
      "learning_rate": 9.266385681357457e-06,
      "loss": 0.3937,
      "step": 17100
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.510449767112732,
      "learning_rate": 9.265370018870748e-06,
      "loss": 0.4529,
      "step": 17101
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5073083639144897,
      "learning_rate": 9.264354364003327e-06,
      "loss": 0.4089,
      "step": 17102
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.52485990524292,
      "learning_rate": 9.263338716765726e-06,
      "loss": 0.474,
      "step": 17103
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5997660160064697,
      "learning_rate": 9.262323077168483e-06,
      "loss": 0.4548,
      "step": 17104
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5880595445632935,
      "learning_rate": 9.261307445222133e-06,
      "loss": 0.5012,
      "step": 17105
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6136338710784912,
      "learning_rate": 9.2602918209372e-06,
      "loss": 0.4583,
      "step": 17106
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.563149333000183,
      "learning_rate": 9.259276204324226e-06,
      "loss": 0.4484,
      "step": 17107
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4758837223052979,
      "learning_rate": 9.25826059539374e-06,
      "loss": 0.4105,
      "step": 17108
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5144546031951904,
      "learning_rate": 9.257244994156282e-06,
      "loss": 0.444,
      "step": 17109
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5032241344451904,
      "learning_rate": 9.25622940062238e-06,
      "loss": 0.4294,
      "step": 17110
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5221903324127197,
      "learning_rate": 9.255213814802566e-06,
      "loss": 0.4763,
      "step": 17111
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7117422819137573,
      "learning_rate": 9.254198236707375e-06,
      "loss": 0.4588,
      "step": 17112
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.441696047782898,
      "learning_rate": 9.253182666347347e-06,
      "loss": 0.4452,
      "step": 17113
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.514769434928894,
      "learning_rate": 9.252167103733005e-06,
      "loss": 0.4137,
      "step": 17114
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.596035361289978,
      "learning_rate": 9.251151548874883e-06,
      "loss": 0.466,
      "step": 17115
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5565990209579468,
      "learning_rate": 9.250136001783521e-06,
      "loss": 0.4341,
      "step": 17116
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7151602506637573,
      "learning_rate": 9.24912046246945e-06,
      "loss": 0.4322,
      "step": 17117
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5431017875671387,
      "learning_rate": 9.248104930943194e-06,
      "loss": 0.4396,
      "step": 17118
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5982028245925903,
      "learning_rate": 9.247089407215296e-06,
      "loss": 0.4465,
      "step": 17119
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.522943377494812,
      "learning_rate": 9.246073891296283e-06,
      "loss": 0.4233,
      "step": 17120
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5257799625396729,
      "learning_rate": 9.245058383196693e-06,
      "loss": 0.4737,
      "step": 17121
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.50452721118927,
      "learning_rate": 9.24404288292705e-06,
      "loss": 0.4029,
      "step": 17122
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5120625495910645,
      "learning_rate": 9.243027390497893e-06,
      "loss": 0.4385,
      "step": 17123
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4879320859909058,
      "learning_rate": 9.242011905919752e-06,
      "loss": 0.433,
      "step": 17124
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5550366640090942,
      "learning_rate": 9.240996429203164e-06,
      "loss": 0.4663,
      "step": 17125
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5564359426498413,
      "learning_rate": 9.23998096035865e-06,
      "loss": 0.4036,
      "step": 17126
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.980781555175781,
      "learning_rate": 9.238965499396753e-06,
      "loss": 0.471,
      "step": 17127
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4271904230117798,
      "learning_rate": 9.237950046327998e-06,
      "loss": 0.3862,
      "step": 17128
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.572407603263855,
      "learning_rate": 9.236934601162924e-06,
      "loss": 0.4159,
      "step": 17129
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5017013549804688,
      "learning_rate": 9.235919163912056e-06,
      "loss": 0.4154,
      "step": 17130
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5219166278839111,
      "learning_rate": 9.234903734585925e-06,
      "loss": 0.4879,
      "step": 17131
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5069687366485596,
      "learning_rate": 9.23388831319507e-06,
      "loss": 0.4172,
      "step": 17132
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5958316326141357,
      "learning_rate": 9.232872899750019e-06,
      "loss": 0.461,
      "step": 17133
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5805586576461792,
      "learning_rate": 9.231857494261298e-06,
      "loss": 0.4321,
      "step": 17134
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6490540504455566,
      "learning_rate": 9.230842096739445e-06,
      "loss": 0.4872,
      "step": 17135
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.520842432975769,
      "learning_rate": 9.22982670719499e-06,
      "loss": 0.3974,
      "step": 17136
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4993804693222046,
      "learning_rate": 9.228811325638465e-06,
      "loss": 0.4357,
      "step": 17137
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5699681043624878,
      "learning_rate": 9.227795952080398e-06,
      "loss": 0.4354,
      "step": 17138
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.516473412513733,
      "learning_rate": 9.226780586531322e-06,
      "loss": 0.4525,
      "step": 17139
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.501227855682373,
      "learning_rate": 9.225765229001766e-06,
      "loss": 0.3996,
      "step": 17140
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.52940034866333,
      "learning_rate": 9.224749879502269e-06,
      "loss": 0.4692,
      "step": 17141
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.79112708568573,
      "learning_rate": 9.223734538043347e-06,
      "loss": 0.3971,
      "step": 17142
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5378512144088745,
      "learning_rate": 9.222719204635545e-06,
      "loss": 0.4696,
      "step": 17143
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.46817946434021,
      "learning_rate": 9.221703879289383e-06,
      "loss": 0.4121,
      "step": 17144
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5355263948440552,
      "learning_rate": 9.220688562015402e-06,
      "loss": 0.4686,
      "step": 17145
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5725404024124146,
      "learning_rate": 9.219673252824124e-06,
      "loss": 0.4353,
      "step": 17146
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.469367504119873,
      "learning_rate": 9.21865795172608e-06,
      "loss": 0.4457,
      "step": 17147
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.458968162536621,
      "learning_rate": 9.217642658731805e-06,
      "loss": 0.3986,
      "step": 17148
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6994670629501343,
      "learning_rate": 9.21662737385183e-06,
      "loss": 0.4904,
      "step": 17149
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.471138834953308,
      "learning_rate": 9.215612097096677e-06,
      "loss": 0.4178,
      "step": 17150
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5799050331115723,
      "learning_rate": 9.21459682847688e-06,
      "loss": 0.449,
      "step": 17151
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5493097305297852,
      "learning_rate": 9.21358156800297e-06,
      "loss": 0.4644,
      "step": 17152
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6228243112564087,
      "learning_rate": 9.212566315685484e-06,
      "loss": 0.4481,
      "step": 17153
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5109094381332397,
      "learning_rate": 9.211551071534937e-06,
      "loss": 0.4014,
      "step": 17154
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5986236333847046,
      "learning_rate": 9.210535835561868e-06,
      "loss": 0.5128,
      "step": 17155
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5215895175933838,
      "learning_rate": 9.209520607776802e-06,
      "loss": 0.408,
      "step": 17156
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5805416107177734,
      "learning_rate": 9.208505388190279e-06,
      "loss": 0.5002,
      "step": 17157
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5531686544418335,
      "learning_rate": 9.207490176812816e-06,
      "loss": 0.4096,
      "step": 17158
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4981693029403687,
      "learning_rate": 9.206474973654945e-06,
      "loss": 0.4374,
      "step": 17159
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5944093465805054,
      "learning_rate": 9.205459778727199e-06,
      "loss": 0.4173,
      "step": 17160
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5880705118179321,
      "learning_rate": 9.204444592040109e-06,
      "loss": 0.4661,
      "step": 17161
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.449224591255188,
      "learning_rate": 9.203429413604198e-06,
      "loss": 0.4077,
      "step": 17162
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5828804969787598,
      "learning_rate": 9.202414243429997e-06,
      "loss": 1.0058,
      "step": 17163
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4230390787124634,
      "learning_rate": 9.201399081528034e-06,
      "loss": 1.0412,
      "step": 17164
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.662265658378601,
      "learning_rate": 9.200383927908845e-06,
      "loss": 0.5043,
      "step": 17165
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.53154456615448,
      "learning_rate": 9.19936878258295e-06,
      "loss": 0.4437,
      "step": 17166
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5048422813415527,
      "learning_rate": 9.19835364556088e-06,
      "loss": 0.4541,
      "step": 17167
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.565582275390625,
      "learning_rate": 9.197338516853167e-06,
      "loss": 0.4242,
      "step": 17168
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9019341468811035,
      "learning_rate": 9.196323396470337e-06,
      "loss": 0.4999,
      "step": 17169
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.481253743171692,
      "learning_rate": 9.195308284422917e-06,
      "loss": 0.3869,
      "step": 17170
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5364049673080444,
      "learning_rate": 9.194293180721437e-06,
      "loss": 0.4632,
      "step": 17171
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.472670555114746,
      "learning_rate": 9.193278085376424e-06,
      "loss": 0.4092,
      "step": 17172
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5464234352111816,
      "learning_rate": 9.192262998398412e-06,
      "loss": 0.4351,
      "step": 17173
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4889545440673828,
      "learning_rate": 9.19124791979792e-06,
      "loss": 0.3838,
      "step": 17174
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6477800607681274,
      "learning_rate": 9.190232849585479e-06,
      "loss": 0.4838,
      "step": 17175
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5107368230819702,
      "learning_rate": 9.189217787771622e-06,
      "loss": 0.4288,
      "step": 17176
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5450149774551392,
      "learning_rate": 9.188202734366874e-06,
      "loss": 0.4681,
      "step": 17177
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5308021306991577,
      "learning_rate": 9.187187689381757e-06,
      "loss": 0.4081,
      "step": 17178
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5757967233657837,
      "learning_rate": 9.186172652826807e-06,
      "loss": 0.453,
      "step": 17179
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5053768157958984,
      "learning_rate": 9.185157624712545e-06,
      "loss": 0.3864,
      "step": 17180
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.635074257850647,
      "learning_rate": 9.184142605049506e-06,
      "loss": 0.4961,
      "step": 17181
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5144131183624268,
      "learning_rate": 9.183127593848212e-06,
      "loss": 0.3999,
      "step": 17182
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.51847505569458,
      "learning_rate": 9.182112591119185e-06,
      "loss": 0.4555,
      "step": 17183
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.490569829940796,
      "learning_rate": 9.181097596872965e-06,
      "loss": 0.4024,
      "step": 17184
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.616112232208252,
      "learning_rate": 9.180082611120071e-06,
      "loss": 0.484,
      "step": 17185
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6277122497558594,
      "learning_rate": 9.179067633871029e-06,
      "loss": 0.4177,
      "step": 17186
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5394551753997803,
      "learning_rate": 9.17805266513637e-06,
      "loss": 0.4743,
      "step": 17187
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5685460567474365,
      "learning_rate": 9.177037704926618e-06,
      "loss": 0.399,
      "step": 17188
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5471659898757935,
      "learning_rate": 9.176022753252305e-06,
      "loss": 0.4198,
      "step": 17189
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6088124513626099,
      "learning_rate": 9.17500781012395e-06,
      "loss": 0.4984,
      "step": 17190
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5461615324020386,
      "learning_rate": 9.173992875552083e-06,
      "loss": 0.486,
      "step": 17191
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4618130922317505,
      "learning_rate": 9.172977949547231e-06,
      "loss": 0.4148,
      "step": 17192
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5607014894485474,
      "learning_rate": 9.171963032119923e-06,
      "loss": 0.4585,
      "step": 17193
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.616646647453308,
      "learning_rate": 9.17094812328068e-06,
      "loss": 0.4612,
      "step": 17194
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6564587354660034,
      "learning_rate": 9.16993322304003e-06,
      "loss": 0.5054,
      "step": 17195
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.1481475830078125,
      "learning_rate": 9.1689183314085e-06,
      "loss": 0.4264,
      "step": 17196
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5277928113937378,
      "learning_rate": 9.167903448396619e-06,
      "loss": 0.4482,
      "step": 17197
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4831289052963257,
      "learning_rate": 9.166888574014907e-06,
      "loss": 0.431,
      "step": 17198
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6049233675003052,
      "learning_rate": 9.165873708273893e-06,
      "loss": 0.45,
      "step": 17199
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4899221658706665,
      "learning_rate": 9.1648588511841e-06,
      "loss": 0.4368,
      "step": 17200
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5757709741592407,
      "learning_rate": 9.163844002756061e-06,
      "loss": 0.4748,
      "step": 17201
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.486459732055664,
      "learning_rate": 9.162829163000295e-06,
      "loss": 0.4088,
      "step": 17202
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5654346942901611,
      "learning_rate": 9.161814331927327e-06,
      "loss": 0.413,
      "step": 17203
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.485874891281128,
      "learning_rate": 9.160799509547687e-06,
      "loss": 0.3949,
      "step": 17204
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6046802997589111,
      "learning_rate": 9.1597846958719e-06,
      "loss": 0.4639,
      "step": 17205
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.492224931716919,
      "learning_rate": 9.158769890910485e-06,
      "loss": 0.4156,
      "step": 17206
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6225595474243164,
      "learning_rate": 9.157755094673973e-06,
      "loss": 0.4323,
      "step": 17207
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5811654329299927,
      "learning_rate": 9.156740307172886e-06,
      "loss": 0.4149,
      "step": 17208
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6237181425094604,
      "learning_rate": 9.155725528417755e-06,
      "loss": 0.4685,
      "step": 17209
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5010383129119873,
      "learning_rate": 9.154710758419098e-06,
      "loss": 0.4128,
      "step": 17210
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5792458057403564,
      "learning_rate": 9.15369599718744e-06,
      "loss": 0.432,
      "step": 17211
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5377565622329712,
      "learning_rate": 9.152681244733308e-06,
      "loss": 0.4225,
      "step": 17212
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4575796127319336,
      "learning_rate": 9.151666501067232e-06,
      "loss": 0.4052,
      "step": 17213
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5367141962051392,
      "learning_rate": 9.150651766199724e-06,
      "loss": 0.4282,
      "step": 17214
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6485340595245361,
      "learning_rate": 9.149637040141319e-06,
      "loss": 0.4566,
      "step": 17215
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5548787117004395,
      "learning_rate": 9.148622322902535e-06,
      "loss": 0.3991,
      "step": 17216
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5862399339675903,
      "learning_rate": 9.147607614493903e-06,
      "loss": 0.4589,
      "step": 17217
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4752179384231567,
      "learning_rate": 9.146592914925941e-06,
      "loss": 0.3863,
      "step": 17218
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5138330459594727,
      "learning_rate": 9.145578224209173e-06,
      "loss": 0.4178,
      "step": 17219
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5371867418289185,
      "learning_rate": 9.144563542354128e-06,
      "loss": 0.3929,
      "step": 17220
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5211714506149292,
      "learning_rate": 9.143548869371328e-06,
      "loss": 0.4522,
      "step": 17221
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5572508573532104,
      "learning_rate": 9.142534205271293e-06,
      "loss": 0.4234,
      "step": 17222
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5181576013565063,
      "learning_rate": 9.14151955006455e-06,
      "loss": 0.458,
      "step": 17223
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5311732292175293,
      "learning_rate": 9.14050490376162e-06,
      "loss": 0.4216,
      "step": 17224
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.593501329421997,
      "learning_rate": 9.139490266373035e-06,
      "loss": 0.4541,
      "step": 17225
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.523896336555481,
      "learning_rate": 9.138475637909307e-06,
      "loss": 0.4094,
      "step": 17226
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7079030275344849,
      "learning_rate": 9.137461018380962e-06,
      "loss": 0.9263,
      "step": 17227
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.585931420326233,
      "learning_rate": 9.13644640779853e-06,
      "loss": 1.0537,
      "step": 17228
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6697750091552734,
      "learning_rate": 9.13543180617253e-06,
      "loss": 0.4939,
      "step": 17229
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6486166715621948,
      "learning_rate": 9.134417213513484e-06,
      "loss": 0.4572,
      "step": 17230
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5814975500106812,
      "learning_rate": 9.133402629831914e-06,
      "loss": 0.4988,
      "step": 17231
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6500108242034912,
      "learning_rate": 9.132388055138345e-06,
      "loss": 0.4032,
      "step": 17232
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.544752836227417,
      "learning_rate": 9.1313734894433e-06,
      "loss": 0.4626,
      "step": 17233
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5337920188903809,
      "learning_rate": 9.1303589327573e-06,
      "loss": 0.4261,
      "step": 17234
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4577561616897583,
      "learning_rate": 9.12934438509087e-06,
      "loss": 0.4179,
      "step": 17235
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5500959157943726,
      "learning_rate": 9.128329846454527e-06,
      "loss": 0.4002,
      "step": 17236
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5115231275558472,
      "learning_rate": 9.127315316858802e-06,
      "loss": 0.4602,
      "step": 17237
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5340021848678589,
      "learning_rate": 9.126300796314211e-06,
      "loss": 0.3891,
      "step": 17238
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6551868915557861,
      "learning_rate": 9.125286284831277e-06,
      "loss": 0.5473,
      "step": 17239
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.534911870956421,
      "learning_rate": 9.124271782420523e-06,
      "loss": 0.4112,
      "step": 17240
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5370750427246094,
      "learning_rate": 9.123257289092474e-06,
      "loss": 0.4522,
      "step": 17241
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5518730878829956,
      "learning_rate": 9.122242804857646e-06,
      "loss": 0.4171,
      "step": 17242
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7490612268447876,
      "learning_rate": 9.121228329726563e-06,
      "loss": 0.9774,
      "step": 17243
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.535304307937622,
      "learning_rate": 9.120213863709747e-06,
      "loss": 1.0127,
      "step": 17244
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6242327690124512,
      "learning_rate": 9.119199406817725e-06,
      "loss": 0.4825,
      "step": 17245
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5728399753570557,
      "learning_rate": 9.118184959061011e-06,
      "loss": 0.414,
      "step": 17246
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5905003547668457,
      "learning_rate": 9.117170520450126e-06,
      "loss": 0.5012,
      "step": 17247
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4468845129013062,
      "learning_rate": 9.116156090995596e-06,
      "loss": 0.3867,
      "step": 17248
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6440123319625854,
      "learning_rate": 9.115141670707944e-06,
      "loss": 0.4721,
      "step": 17249
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7077487707138062,
      "learning_rate": 9.114127259597684e-06,
      "loss": 0.4396,
      "step": 17250
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5234692096710205,
      "learning_rate": 9.113112857675342e-06,
      "loss": 0.4779,
      "step": 17251
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.530135989189148,
      "learning_rate": 9.112098464951436e-06,
      "loss": 0.4258,
      "step": 17252
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4926860332489014,
      "learning_rate": 9.111084081436495e-06,
      "loss": 0.4427,
      "step": 17253
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.437663197517395,
      "learning_rate": 9.110069707141029e-06,
      "loss": 0.3906,
      "step": 17254
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5791627168655396,
      "learning_rate": 9.109055342075561e-06,
      "loss": 0.4398,
      "step": 17255
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5376989841461182,
      "learning_rate": 9.108040986250618e-06,
      "loss": 0.4054,
      "step": 17256
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.458939552307129,
      "learning_rate": 9.107026639676717e-06,
      "loss": 0.4404,
      "step": 17257
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.632377028465271,
      "learning_rate": 9.106012302364374e-06,
      "loss": 0.4413,
      "step": 17258
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7230863571166992,
      "learning_rate": 9.104997974324113e-06,
      "loss": 1.0099,
      "step": 17259
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5073896646499634,
      "learning_rate": 9.103983655566453e-06,
      "loss": 1.0108,
      "step": 17260
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.641189694404602,
      "learning_rate": 9.102969346101922e-06,
      "loss": 0.467,
      "step": 17261
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5863063335418701,
      "learning_rate": 9.10195504594103e-06,
      "loss": 0.4067,
      "step": 17262
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.589756727218628,
      "learning_rate": 9.100940755094296e-06,
      "loss": 0.5162,
      "step": 17263
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5314867496490479,
      "learning_rate": 9.099926473572248e-06,
      "loss": 0.4175,
      "step": 17264
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5936325788497925,
      "learning_rate": 9.098912201385403e-06,
      "loss": 0.4421,
      "step": 17265
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4766966104507446,
      "learning_rate": 9.09789793854428e-06,
      "loss": 0.4008,
      "step": 17266
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5210391283035278,
      "learning_rate": 9.096883685059392e-06,
      "loss": 0.4618,
      "step": 17267
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4879279136657715,
      "learning_rate": 9.095869440941268e-06,
      "loss": 0.4028,
      "step": 17268
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.622304916381836,
      "learning_rate": 9.094855206200425e-06,
      "loss": 0.4354,
      "step": 17269
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5463590621948242,
      "learning_rate": 9.093840980847379e-06,
      "loss": 0.4036,
      "step": 17270
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.541885256767273,
      "learning_rate": 9.092826764892652e-06,
      "loss": 0.4818,
      "step": 17271
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5577954053878784,
      "learning_rate": 9.091812558346759e-06,
      "loss": 0.413,
      "step": 17272
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6496237516403198,
      "learning_rate": 9.090798361220228e-06,
      "loss": 0.5389,
      "step": 17273
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5125097036361694,
      "learning_rate": 9.089784173523568e-06,
      "loss": 0.3901,
      "step": 17274
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5110222101211548,
      "learning_rate": 9.0887699952673e-06,
      "loss": 0.4471,
      "step": 17275
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5030527114868164,
      "learning_rate": 9.087755826461947e-06,
      "loss": 0.4155,
      "step": 17276
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.631375789642334,
      "learning_rate": 9.086741667118026e-06,
      "loss": 0.5007,
      "step": 17277
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5924057960510254,
      "learning_rate": 9.085727517246052e-06,
      "loss": 0.4352,
      "step": 17278
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7671501636505127,
      "learning_rate": 9.084713376856546e-06,
      "loss": 0.5459,
      "step": 17279
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5084797143936157,
      "learning_rate": 9.083699245960024e-06,
      "loss": 0.3784,
      "step": 17280
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7071669101715088,
      "learning_rate": 9.08268512456701e-06,
      "loss": 0.4525,
      "step": 17281
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5403045415878296,
      "learning_rate": 9.081671012688015e-06,
      "loss": 0.4227,
      "step": 17282
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.591530442237854,
      "learning_rate": 9.080656910333561e-06,
      "loss": 0.441,
      "step": 17283
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.520071029663086,
      "learning_rate": 9.079642817514164e-06,
      "loss": 0.4148,
      "step": 17284
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5479915142059326,
      "learning_rate": 9.078628734240347e-06,
      "loss": 0.4816,
      "step": 17285
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6046652793884277,
      "learning_rate": 9.077614660522619e-06,
      "loss": 0.4305,
      "step": 17286
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.658716082572937,
      "learning_rate": 9.076600596371502e-06,
      "loss": 0.48,
      "step": 17287
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4648284912109375,
      "learning_rate": 9.075586541797515e-06,
      "loss": 0.4119,
      "step": 17288
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5868079662322998,
      "learning_rate": 9.074572496811174e-06,
      "loss": 0.4556,
      "step": 17289
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5264898538589478,
      "learning_rate": 9.073558461422996e-06,
      "loss": 0.3897,
      "step": 17290
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.527397871017456,
      "learning_rate": 9.072544435643497e-06,
      "loss": 0.4601,
      "step": 17291
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5188390016555786,
      "learning_rate": 9.071530419483197e-06,
      "loss": 0.4197,
      "step": 17292
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.561226487159729,
      "learning_rate": 9.070516412952612e-06,
      "loss": 0.4906,
      "step": 17293
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4734045267105103,
      "learning_rate": 9.069502416062254e-06,
      "loss": 0.3941,
      "step": 17294
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5958964824676514,
      "learning_rate": 9.068488428822646e-06,
      "loss": 0.5053,
      "step": 17295
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5364854335784912,
      "learning_rate": 9.067474451244302e-06,
      "loss": 0.4295,
      "step": 17296
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5821834802627563,
      "learning_rate": 9.066460483337743e-06,
      "loss": 0.4919,
      "step": 17297
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5789107084274292,
      "learning_rate": 9.065446525113477e-06,
      "loss": 0.4105,
      "step": 17298
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4609464406967163,
      "learning_rate": 9.064432576582026e-06,
      "loss": 0.4161,
      "step": 17299
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.558607578277588,
      "learning_rate": 9.063418637753904e-06,
      "loss": 0.436,
      "step": 17300
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5473321676254272,
      "learning_rate": 9.062404708639633e-06,
      "loss": 0.4937,
      "step": 17301
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5329443216323853,
      "learning_rate": 9.06139078924972e-06,
      "loss": 0.4019,
      "step": 17302
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5632684230804443,
      "learning_rate": 9.060376879594688e-06,
      "loss": 0.4458,
      "step": 17303
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4918229579925537,
      "learning_rate": 9.059362979685047e-06,
      "loss": 0.4245,
      "step": 17304
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.621390700340271,
      "learning_rate": 9.058349089531323e-06,
      "loss": 0.4565,
      "step": 17305
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7857681512832642,
      "learning_rate": 9.057335209144018e-06,
      "loss": 0.4245,
      "step": 17306
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4655431509017944,
      "learning_rate": 9.056321338533657e-06,
      "loss": 0.4244,
      "step": 17307
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5992075204849243,
      "learning_rate": 9.05530747771075e-06,
      "loss": 0.43,
      "step": 17308
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6654384136199951,
      "learning_rate": 9.05429362668582e-06,
      "loss": 0.5106,
      "step": 17309
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.599305272102356,
      "learning_rate": 9.053279785469374e-06,
      "loss": 0.4387,
      "step": 17310
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6322383880615234,
      "learning_rate": 9.052265954071929e-06,
      "loss": 0.4744,
      "step": 17311
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.564680814743042,
      "learning_rate": 9.051252132504003e-06,
      "loss": 0.4213,
      "step": 17312
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4639487266540527,
      "learning_rate": 9.050238320776113e-06,
      "loss": 0.4586,
      "step": 17313
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5691044330596924,
      "learning_rate": 9.049224518898765e-06,
      "loss": 0.4028,
      "step": 17314
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6184449195861816,
      "learning_rate": 9.048210726882482e-06,
      "loss": 0.4708,
      "step": 17315
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4171346426010132,
      "learning_rate": 9.047196944737774e-06,
      "loss": 0.3865,
      "step": 17316
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.453048586845398,
      "learning_rate": 9.046183172475161e-06,
      "loss": 0.4348,
      "step": 17317
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5277056694030762,
      "learning_rate": 9.045169410105152e-06,
      "loss": 0.4173,
      "step": 17318
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.55172860622406,
      "learning_rate": 9.04415565763826e-06,
      "loss": 0.4663,
      "step": 17319
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5881725549697876,
      "learning_rate": 9.043141915085006e-06,
      "loss": 0.4132,
      "step": 17320
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.464626669883728,
      "learning_rate": 9.0421281824559e-06,
      "loss": 0.4502,
      "step": 17321
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.53237783908844,
      "learning_rate": 9.041114459761454e-06,
      "loss": 0.3988,
      "step": 17322
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5519309043884277,
      "learning_rate": 9.040100747012187e-06,
      "loss": 0.5005,
      "step": 17323
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4932405948638916,
      "learning_rate": 9.039087044218607e-06,
      "loss": 0.411,
      "step": 17324
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.559531331062317,
      "learning_rate": 9.038073351391236e-06,
      "loss": 0.5051,
      "step": 17325
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5270758867263794,
      "learning_rate": 9.037059668540579e-06,
      "loss": 0.4341,
      "step": 17326
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.683193564414978,
      "learning_rate": 9.036045995677152e-06,
      "loss": 0.4797,
      "step": 17327
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6256966590881348,
      "learning_rate": 9.035032332811473e-06,
      "loss": 0.4164,
      "step": 17328
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4472500085830688,
      "learning_rate": 9.034018679954051e-06,
      "loss": 0.4317,
      "step": 17329
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4927557706832886,
      "learning_rate": 9.033005037115397e-06,
      "loss": 0.4225,
      "step": 17330
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7120966911315918,
      "learning_rate": 9.031991404306029e-06,
      "loss": 0.489,
      "step": 17331
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.527663230895996,
      "learning_rate": 9.030977781536456e-06,
      "loss": 0.3893,
      "step": 17332
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.554490327835083,
      "learning_rate": 9.029964168817198e-06,
      "loss": 0.4414,
      "step": 17333
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5049245357513428,
      "learning_rate": 9.02895056615876e-06,
      "loss": 0.3865,
      "step": 17334
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.512269139289856,
      "learning_rate": 9.027936973571656e-06,
      "loss": 0.4287,
      "step": 17335
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4720619916915894,
      "learning_rate": 9.0269233910664e-06,
      "loss": 0.3903,
      "step": 17336
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5013248920440674,
      "learning_rate": 9.025909818653509e-06,
      "loss": 0.444,
      "step": 17337
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6067811250686646,
      "learning_rate": 9.024896256343484e-06,
      "loss": 0.408,
      "step": 17338
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6377747058868408,
      "learning_rate": 9.023882704146848e-06,
      "loss": 0.4862,
      "step": 17339
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5496788024902344,
      "learning_rate": 9.022869162074108e-06,
      "loss": 0.42,
      "step": 17340
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5644274950027466,
      "learning_rate": 9.021855630135783e-06,
      "loss": 0.46,
      "step": 17341
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.562017560005188,
      "learning_rate": 9.020842108342372e-06,
      "loss": 0.4096,
      "step": 17342
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5489548444747925,
      "learning_rate": 9.019828596704394e-06,
      "loss": 0.4148,
      "step": 17343
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5973726511001587,
      "learning_rate": 9.018815095232363e-06,
      "loss": 0.407,
      "step": 17344
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4647879600524902,
      "learning_rate": 9.017801603936788e-06,
      "loss": 0.4637,
      "step": 17345
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6297061443328857,
      "learning_rate": 9.016788122828187e-06,
      "loss": 0.419,
      "step": 17346
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5317589044570923,
      "learning_rate": 9.015774651917058e-06,
      "loss": 0.4387,
      "step": 17347
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5021392107009888,
      "learning_rate": 9.01476119121392e-06,
      "loss": 0.4224,
      "step": 17348
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5006672143936157,
      "learning_rate": 9.013747740729286e-06,
      "loss": 0.4355,
      "step": 17349
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.578314185142517,
      "learning_rate": 9.012734300473666e-06,
      "loss": 0.3882,
      "step": 17350
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5894770622253418,
      "learning_rate": 9.011720870457569e-06,
      "loss": 0.4497,
      "step": 17351
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.531024694442749,
      "learning_rate": 9.010707450691506e-06,
      "loss": 0.4307,
      "step": 17352
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5734961032867432,
      "learning_rate": 9.009694041185987e-06,
      "loss": 0.4674,
      "step": 17353
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.459031343460083,
      "learning_rate": 9.00868064195153e-06,
      "loss": 0.3852,
      "step": 17354
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.530516266822815,
      "learning_rate": 9.007667252998636e-06,
      "loss": 0.4678,
      "step": 17355
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5253939628601074,
      "learning_rate": 9.00665387433782e-06,
      "loss": 0.4134,
      "step": 17356
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7918810844421387,
      "learning_rate": 9.00564050597959e-06,
      "loss": 0.5186,
      "step": 17357
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4949301481246948,
      "learning_rate": 9.00462714793446e-06,
      "loss": 0.4028,
      "step": 17358
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.459112524986267,
      "learning_rate": 9.00361380021294e-06,
      "loss": 0.4224,
      "step": 17359
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4470041990280151,
      "learning_rate": 9.002600462825535e-06,
      "loss": 0.4021,
      "step": 17360
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5240126848220825,
      "learning_rate": 9.00158713578276e-06,
      "loss": 0.5067,
      "step": 17361
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5118290185928345,
      "learning_rate": 9.000573819095123e-06,
      "loss": 0.4113,
      "step": 17362
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6673252582550049,
      "learning_rate": 8.999560512773133e-06,
      "loss": 0.5018,
      "step": 17363
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5803955793380737,
      "learning_rate": 8.9985472168273e-06,
      "loss": 0.4665,
      "step": 17364
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6832714080810547,
      "learning_rate": 8.997533931268133e-06,
      "loss": 0.5007,
      "step": 17365
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5382663011550903,
      "learning_rate": 8.996520656106145e-06,
      "loss": 0.4086,
      "step": 17366
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5852065086364746,
      "learning_rate": 8.99550739135184e-06,
      "loss": 0.4495,
      "step": 17367
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5211864709854126,
      "learning_rate": 8.994494137015728e-06,
      "loss": 0.4029,
      "step": 17368
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5301121473312378,
      "learning_rate": 8.993480893108322e-06,
      "loss": 0.4389,
      "step": 17369
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.545922875404358,
      "learning_rate": 8.992467659640132e-06,
      "loss": 0.3912,
      "step": 17370
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.649802327156067,
      "learning_rate": 8.991454436621657e-06,
      "loss": 0.4585,
      "step": 17371
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.569895625114441,
      "learning_rate": 8.990441224063416e-06,
      "loss": 0.4117,
      "step": 17372
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5329699516296387,
      "learning_rate": 8.989428021975911e-06,
      "loss": 0.4185,
      "step": 17373
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5303269624710083,
      "learning_rate": 8.988414830369658e-06,
      "loss": 0.4166,
      "step": 17374
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.649942398071289,
      "learning_rate": 8.987401649255158e-06,
      "loss": 0.4747,
      "step": 17375
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4643597602844238,
      "learning_rate": 8.986388478642921e-06,
      "loss": 0.3919,
      "step": 17376
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4896348714828491,
      "learning_rate": 8.985375318543459e-06,
      "loss": 0.4129,
      "step": 17377
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6241543292999268,
      "learning_rate": 8.984362168967278e-06,
      "loss": 0.4324,
      "step": 17378
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.647700309753418,
      "learning_rate": 8.983349029924881e-06,
      "loss": 0.4525,
      "step": 17379
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.668263554573059,
      "learning_rate": 8.982335901426784e-06,
      "loss": 0.4223,
      "step": 17380
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6464699506759644,
      "learning_rate": 8.981322783483489e-06,
      "loss": 0.4822,
      "step": 17381
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5807256698608398,
      "learning_rate": 8.980309676105509e-06,
      "loss": 0.413,
      "step": 17382
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.673555612564087,
      "learning_rate": 8.979296579303346e-06,
      "loss": 0.4594,
      "step": 17383
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5926146507263184,
      "learning_rate": 8.97828349308751e-06,
      "loss": 0.4106,
      "step": 17384
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.57613205909729,
      "learning_rate": 8.977270417468505e-06,
      "loss": 0.4628,
      "step": 17385
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5239739418029785,
      "learning_rate": 8.976257352456845e-06,
      "loss": 0.4016,
      "step": 17386
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6889369487762451,
      "learning_rate": 8.975244298063032e-06,
      "loss": 0.5175,
      "step": 17387
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5234159231185913,
      "learning_rate": 8.974231254297572e-06,
      "loss": 0.3838,
      "step": 17388
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6046416759490967,
      "learning_rate": 8.973218221170979e-06,
      "loss": 0.4686,
      "step": 17389
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5259073972702026,
      "learning_rate": 8.972205198693754e-06,
      "loss": 0.4271,
      "step": 17390
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.502455234527588,
      "learning_rate": 8.971192186876401e-06,
      "loss": 0.4224,
      "step": 17391
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4960216283798218,
      "learning_rate": 8.970179185729431e-06,
      "loss": 0.3987,
      "step": 17392
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5771596431732178,
      "learning_rate": 8.96916619526335e-06,
      "loss": 0.4705,
      "step": 17393
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6572657823562622,
      "learning_rate": 8.968153215488668e-06,
      "loss": 0.4464,
      "step": 17394
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6224030256271362,
      "learning_rate": 8.967140246415883e-06,
      "loss": 0.4545,
      "step": 17395
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.658006191253662,
      "learning_rate": 8.966127288055505e-06,
      "loss": 0.4374,
      "step": 17396
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6019901037216187,
      "learning_rate": 8.965114340418042e-06,
      "loss": 0.4415,
      "step": 17397
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5366969108581543,
      "learning_rate": 8.964101403514e-06,
      "loss": 0.3742,
      "step": 17398
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.525994896888733,
      "learning_rate": 8.963088477353879e-06,
      "loss": 0.4456,
      "step": 17399
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.497501254081726,
      "learning_rate": 8.96207556194819e-06,
      "loss": 0.4166,
      "step": 17400
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6118727922439575,
      "learning_rate": 8.961062657307434e-06,
      "loss": 0.4663,
      "step": 17401
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5603054761886597,
      "learning_rate": 8.960049763442125e-06,
      "loss": 0.3922,
      "step": 17402
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.604585886001587,
      "learning_rate": 8.959036880362762e-06,
      "loss": 0.5332,
      "step": 17403
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4894801378250122,
      "learning_rate": 8.95802400807985e-06,
      "loss": 0.4199,
      "step": 17404
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6706511974334717,
      "learning_rate": 8.957011146603894e-06,
      "loss": 0.4733,
      "step": 17405
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9160100221633911,
      "learning_rate": 8.955998295945402e-06,
      "loss": 0.4426,
      "step": 17406
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5550596714019775,
      "learning_rate": 8.954985456114876e-06,
      "loss": 0.4368,
      "step": 17407
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.658690094947815,
      "learning_rate": 8.953972627122822e-06,
      "loss": 0.4196,
      "step": 17408
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5859860181808472,
      "learning_rate": 8.952959808979743e-06,
      "loss": 0.4308,
      "step": 17409
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5934412479400635,
      "learning_rate": 8.95194700169615e-06,
      "loss": 0.4199,
      "step": 17410
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.718297004699707,
      "learning_rate": 8.95093420528254e-06,
      "loss": 0.4808,
      "step": 17411
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5959423780441284,
      "learning_rate": 8.949921419749416e-06,
      "loss": 0.442,
      "step": 17412
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5774229764938354,
      "learning_rate": 8.948908645107291e-06,
      "loss": 0.5131,
      "step": 17413
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5946630239486694,
      "learning_rate": 8.947895881366665e-06,
      "loss": 0.4623,
      "step": 17414
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5384172201156616,
      "learning_rate": 8.946883128538037e-06,
      "loss": 0.4554,
      "step": 17415
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7138501405715942,
      "learning_rate": 8.945870386631916e-06,
      "loss": 0.4346,
      "step": 17416
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.577301263809204,
      "learning_rate": 8.944857655658804e-06,
      "loss": 0.5378,
      "step": 17417
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5311223268508911,
      "learning_rate": 8.943844935629208e-06,
      "loss": 0.3923,
      "step": 17418
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5915929079055786,
      "learning_rate": 8.942832226553628e-06,
      "loss": 0.444,
      "step": 17419
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.541795015335083,
      "learning_rate": 8.941819528442566e-06,
      "loss": 0.3983,
      "step": 17420
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9332078695297241,
      "learning_rate": 8.94080684130653e-06,
      "loss": 0.4945,
      "step": 17421
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.52403724193573,
      "learning_rate": 8.939794165156023e-06,
      "loss": 0.4051,
      "step": 17422
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7432193756103516,
      "learning_rate": 8.938781500001543e-06,
      "loss": 0.4746,
      "step": 17423
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.530452847480774,
      "learning_rate": 8.937768845853595e-06,
      "loss": 0.4279,
      "step": 17424
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4723389148712158,
      "learning_rate": 8.936756202722684e-06,
      "loss": 0.445,
      "step": 17425
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.490976095199585,
      "learning_rate": 8.935743570619313e-06,
      "loss": 0.4203,
      "step": 17426
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.718540072441101,
      "learning_rate": 8.93473094955398e-06,
      "loss": 0.4856,
      "step": 17427
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5746945142745972,
      "learning_rate": 8.93371833953719e-06,
      "loss": 0.4256,
      "step": 17428
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5403540134429932,
      "learning_rate": 8.932705740579446e-06,
      "loss": 0.4268,
      "step": 17429
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7883273363113403,
      "learning_rate": 8.931693152691254e-06,
      "loss": 0.4295,
      "step": 17430
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5889408588409424,
      "learning_rate": 8.930680575883111e-06,
      "loss": 0.4839,
      "step": 17431
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5072559118270874,
      "learning_rate": 8.929668010165518e-06,
      "loss": 0.4132,
      "step": 17432
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.827239990234375,
      "learning_rate": 8.92865545554898e-06,
      "loss": 1.0557,
      "step": 17433
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5833399295806885,
      "learning_rate": 8.927642912044e-06,
      "loss": 1.0386,
      "step": 17434
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.668864130973816,
      "learning_rate": 8.926630379661076e-06,
      "loss": 0.4359,
      "step": 17435
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5667022466659546,
      "learning_rate": 8.925617858410712e-06,
      "loss": 0.3806,
      "step": 17436
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6161634922027588,
      "learning_rate": 8.924605348303405e-06,
      "loss": 0.487,
      "step": 17437
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6621469259262085,
      "learning_rate": 8.923592849349667e-06,
      "loss": 0.4205,
      "step": 17438
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4466434717178345,
      "learning_rate": 8.92258036155999e-06,
      "loss": 0.4036,
      "step": 17439
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4524677991867065,
      "learning_rate": 8.921567884944876e-06,
      "loss": 0.4037,
      "step": 17440
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.535878300666809,
      "learning_rate": 8.920555419514827e-06,
      "loss": 0.4465,
      "step": 17441
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5397627353668213,
      "learning_rate": 8.919542965280349e-06,
      "loss": 0.4131,
      "step": 17442
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5012916326522827,
      "learning_rate": 8.918530522251933e-06,
      "loss": 0.4484,
      "step": 17443
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5746222734451294,
      "learning_rate": 8.917518090440086e-06,
      "loss": 0.4424,
      "step": 17444
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6813715696334839,
      "learning_rate": 8.916505669855307e-06,
      "loss": 0.4822,
      "step": 17445
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5873780250549316,
      "learning_rate": 8.9154932605081e-06,
      "loss": 0.4173,
      "step": 17446
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.570165991783142,
      "learning_rate": 8.914480862408961e-06,
      "loss": 0.4847,
      "step": 17447
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.568626880645752,
      "learning_rate": 8.913468475568387e-06,
      "loss": 0.4269,
      "step": 17448
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6952887773513794,
      "learning_rate": 8.912456099996887e-06,
      "loss": 1.0451,
      "step": 17449
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5845710039138794,
      "learning_rate": 8.911443735704956e-06,
      "loss": 1.0149,
      "step": 17450
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6219562292099,
      "learning_rate": 8.910431382703092e-06,
      "loss": 0.4251,
      "step": 17451
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6304665803909302,
      "learning_rate": 8.909419041001799e-06,
      "loss": 0.4245,
      "step": 17452
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5183502435684204,
      "learning_rate": 8.908406710611571e-06,
      "loss": 0.433,
      "step": 17453
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5123841762542725,
      "learning_rate": 8.907394391542917e-06,
      "loss": 0.3938,
      "step": 17454
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6538360118865967,
      "learning_rate": 8.906382083806327e-06,
      "loss": 0.4742,
      "step": 17455
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6390498876571655,
      "learning_rate": 8.9053697874123e-06,
      "loss": 0.4041,
      "step": 17456
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6246602535247803,
      "learning_rate": 8.904357502371343e-06,
      "loss": 0.4364,
      "step": 17457
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5513107776641846,
      "learning_rate": 8.903345228693952e-06,
      "loss": 0.438,
      "step": 17458
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.620998740196228,
      "learning_rate": 8.902332966390622e-06,
      "loss": 0.4626,
      "step": 17459
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.635511875152588,
      "learning_rate": 8.901320715471853e-06,
      "loss": 0.4646,
      "step": 17460
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.196074962615967,
      "learning_rate": 8.900308475948147e-06,
      "loss": 0.5247,
      "step": 17461
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.503641963005066,
      "learning_rate": 8.899296247830004e-06,
      "loss": 0.3881,
      "step": 17462
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7447354793548584,
      "learning_rate": 8.898284031127912e-06,
      "loss": 0.4725,
      "step": 17463
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4941284656524658,
      "learning_rate": 8.897271825852381e-06,
      "loss": 0.4022,
      "step": 17464
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5883909463882446,
      "learning_rate": 8.896259632013902e-06,
      "loss": 0.48,
      "step": 17465
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5518128871917725,
      "learning_rate": 8.895247449622979e-06,
      "loss": 0.4191,
      "step": 17466
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5085493326187134,
      "learning_rate": 8.894235278690104e-06,
      "loss": 0.4659,
      "step": 17467
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5352777242660522,
      "learning_rate": 8.893223119225775e-06,
      "loss": 0.409,
      "step": 17468
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5192149877548218,
      "learning_rate": 8.892210971240497e-06,
      "loss": 0.4712,
      "step": 17469
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5769519805908203,
      "learning_rate": 8.891198834744763e-06,
      "loss": 0.4391,
      "step": 17470
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.562883734703064,
      "learning_rate": 8.890186709749064e-06,
      "loss": 0.4492,
      "step": 17471
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5058598518371582,
      "learning_rate": 8.889174596263908e-06,
      "loss": 0.3935,
      "step": 17472
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5381613969802856,
      "learning_rate": 8.888162494299787e-06,
      "loss": 0.4612,
      "step": 17473
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5162055492401123,
      "learning_rate": 8.8871504038672e-06,
      "loss": 0.3764,
      "step": 17474
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6417597532272339,
      "learning_rate": 8.886138324976642e-06,
      "loss": 0.5366,
      "step": 17475
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5250846147537231,
      "learning_rate": 8.885126257638608e-06,
      "loss": 0.3904,
      "step": 17476
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.551207184791565,
      "learning_rate": 8.884114201863601e-06,
      "loss": 0.4516,
      "step": 17477
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5186145305633545,
      "learning_rate": 8.883102157662115e-06,
      "loss": 0.3957,
      "step": 17478
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6045053005218506,
      "learning_rate": 8.882090125044642e-06,
      "loss": 0.4637,
      "step": 17479
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5406509637832642,
      "learning_rate": 8.881078104021684e-06,
      "loss": 0.4049,
      "step": 17480
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8311101198196411,
      "learning_rate": 8.880066094603733e-06,
      "loss": 0.5061,
      "step": 17481
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5050331354141235,
      "learning_rate": 8.87905409680129e-06,
      "loss": 0.3877,
      "step": 17482
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5860885381698608,
      "learning_rate": 8.87804211062485e-06,
      "loss": 0.4532,
      "step": 17483
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5395201444625854,
      "learning_rate": 8.877030136084903e-06,
      "loss": 0.4169,
      "step": 17484
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7256280183792114,
      "learning_rate": 8.876018173191951e-06,
      "loss": 0.4815,
      "step": 17485
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5780614614486694,
      "learning_rate": 8.875006221956492e-06,
      "loss": 0.4283,
      "step": 17486
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6417975425720215,
      "learning_rate": 8.873994282389011e-06,
      "loss": 0.4757,
      "step": 17487
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7311598062515259,
      "learning_rate": 8.872982354500013e-06,
      "loss": 0.4223,
      "step": 17488
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.5864009857177734,
      "learning_rate": 8.871970438299989e-06,
      "loss": 0.4811,
      "step": 17489
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6880269050598145,
      "learning_rate": 8.870958533799437e-06,
      "loss": 0.4892,
      "step": 17490
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.623159646987915,
      "learning_rate": 8.869946641008851e-06,
      "loss": 0.4424,
      "step": 17491
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.552003264427185,
      "learning_rate": 8.868934759938723e-06,
      "loss": 0.4177,
      "step": 17492
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5351991653442383,
      "learning_rate": 8.86792289059955e-06,
      "loss": 0.4334,
      "step": 17493
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8057312965393066,
      "learning_rate": 8.86691103300183e-06,
      "loss": 0.3983,
      "step": 17494
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5398046970367432,
      "learning_rate": 8.865899187156051e-06,
      "loss": 0.4679,
      "step": 17495
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.513398289680481,
      "learning_rate": 8.864887353072712e-06,
      "loss": 0.4158,
      "step": 17496
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6625986099243164,
      "learning_rate": 8.863875530762307e-06,
      "loss": 0.4438,
      "step": 17497
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.542946696281433,
      "learning_rate": 8.862863720235334e-06,
      "loss": 0.3869,
      "step": 17498
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.715295433998108,
      "learning_rate": 8.861851921502276e-06,
      "loss": 0.4838,
      "step": 17499
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5909759998321533,
      "learning_rate": 8.860840134573634e-06,
      "loss": 0.4206,
      "step": 17500
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5446217060089111,
      "learning_rate": 8.859828359459903e-06,
      "loss": 0.4808,
      "step": 17501
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.598486065864563,
      "learning_rate": 8.858816596171576e-06,
      "loss": 0.4427,
      "step": 17502
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5587719678878784,
      "learning_rate": 8.857804844719147e-06,
      "loss": 0.4888,
      "step": 17503
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.719979166984558,
      "learning_rate": 8.856793105113103e-06,
      "loss": 0.4061,
      "step": 17504
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.573392391204834,
      "learning_rate": 8.855781377363946e-06,
      "loss": 0.4545,
      "step": 17505
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5536584854125977,
      "learning_rate": 8.854769661482168e-06,
      "loss": 0.4274,
      "step": 17506
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5475826263427734,
      "learning_rate": 8.853757957478255e-06,
      "loss": 0.4238,
      "step": 17507
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.519704818725586,
      "learning_rate": 8.852746265362708e-06,
      "loss": 0.3713,
      "step": 17508
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5866042375564575,
      "learning_rate": 8.851734585146015e-06,
      "loss": 0.5111,
      "step": 17509
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5343003273010254,
      "learning_rate": 8.850722916838674e-06,
      "loss": 0.4127,
      "step": 17510
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6740647554397583,
      "learning_rate": 8.849711260451172e-06,
      "loss": 0.4395,
      "step": 17511
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5243194103240967,
      "learning_rate": 8.848699615994e-06,
      "loss": 0.3842,
      "step": 17512
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5993074178695679,
      "learning_rate": 8.847687983477659e-06,
      "loss": 0.4686,
      "step": 17513
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.49919855594635,
      "learning_rate": 8.846676362912636e-06,
      "loss": 0.4112,
      "step": 17514
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.550132155418396,
      "learning_rate": 8.845664754309422e-06,
      "loss": 0.4716,
      "step": 17515
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5324161052703857,
      "learning_rate": 8.844653157678509e-06,
      "loss": 0.4297,
      "step": 17516
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8395811319351196,
      "learning_rate": 8.843641573030391e-06,
      "loss": 0.984,
      "step": 17517
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5300542116165161,
      "learning_rate": 8.84263000037556e-06,
      "loss": 0.9822,
      "step": 17518
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6335135698318481,
      "learning_rate": 8.841618439724508e-06,
      "loss": 0.4815,
      "step": 17519
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6287508010864258,
      "learning_rate": 8.840606891087721e-06,
      "loss": 0.4217,
      "step": 17520
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6567631959915161,
      "learning_rate": 8.839595354475698e-06,
      "loss": 0.413,
      "step": 17521
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5231168270111084,
      "learning_rate": 8.838583829898928e-06,
      "loss": 0.3993,
      "step": 17522
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5888144969940186,
      "learning_rate": 8.837572317367897e-06,
      "loss": 0.4138,
      "step": 17523
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.514571189880371,
      "learning_rate": 8.836560816893101e-06,
      "loss": 0.4357,
      "step": 17524
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7687830924987793,
      "learning_rate": 8.835549328485028e-06,
      "loss": 0.4968,
      "step": 17525
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5261328220367432,
      "learning_rate": 8.834537852154174e-06,
      "loss": 0.4018,
      "step": 17526
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.569565773010254,
      "learning_rate": 8.833526387911027e-06,
      "loss": 0.429,
      "step": 17527
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5299896001815796,
      "learning_rate": 8.832514935766073e-06,
      "loss": 0.3878,
      "step": 17528
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6022132635116577,
      "learning_rate": 8.831503495729807e-06,
      "loss": 0.5227,
      "step": 17529
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5640852451324463,
      "learning_rate": 8.83049206781272e-06,
      "loss": 0.3945,
      "step": 17530
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.551612138748169,
      "learning_rate": 8.829480652025298e-06,
      "loss": 0.4318,
      "step": 17531
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6107397079467773,
      "learning_rate": 8.828469248378035e-06,
      "loss": 0.3911,
      "step": 17532
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5972870588302612,
      "learning_rate": 8.827457856881417e-06,
      "loss": 0.4656,
      "step": 17533
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5506733655929565,
      "learning_rate": 8.826446477545942e-06,
      "loss": 0.3895,
      "step": 17534
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6949865818023682,
      "learning_rate": 8.825435110382087e-06,
      "loss": 0.4214,
      "step": 17535
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.633957862854004,
      "learning_rate": 8.824423755400348e-06,
      "loss": 0.4336,
      "step": 17536
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.570954442024231,
      "learning_rate": 8.823412412611218e-06,
      "loss": 0.4566,
      "step": 17537
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5912514925003052,
      "learning_rate": 8.822401082025185e-06,
      "loss": 0.4311,
      "step": 17538
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.639224648475647,
      "learning_rate": 8.821389763652733e-06,
      "loss": 0.5058,
      "step": 17539
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.489187479019165,
      "learning_rate": 8.820378457504351e-06,
      "loss": 0.382,
      "step": 17540
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4362746477127075,
      "learning_rate": 8.819367163590535e-06,
      "loss": 0.4355,
      "step": 17541
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5332640409469604,
      "learning_rate": 8.818355881921769e-06,
      "loss": 0.4028,
      "step": 17542
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6283156871795654,
      "learning_rate": 8.817344612508538e-06,
      "loss": 0.4811,
      "step": 17543
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.540305733680725,
      "learning_rate": 8.816333355361339e-06,
      "loss": 0.4533,
      "step": 17544
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.620017409324646,
      "learning_rate": 8.815322110490652e-06,
      "loss": 0.4554,
      "step": 17545
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.519265055656433,
      "learning_rate": 8.814310877906973e-06,
      "loss": 0.441,
      "step": 17546
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5545367002487183,
      "learning_rate": 8.813299657620784e-06,
      "loss": 0.4635,
      "step": 17547
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4532285928726196,
      "learning_rate": 8.812288449642574e-06,
      "loss": 0.3716,
      "step": 17548
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4562411308288574,
      "learning_rate": 8.811277253982835e-06,
      "loss": 0.4512,
      "step": 17549
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5319024324417114,
      "learning_rate": 8.810266070652053e-06,
      "loss": 0.3975,
      "step": 17550
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6112642288208008,
      "learning_rate": 8.80925489966071e-06,
      "loss": 0.4101,
      "step": 17551
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5466350317001343,
      "learning_rate": 8.808243741019299e-06,
      "loss": 0.4286,
      "step": 17552
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6333913803100586,
      "learning_rate": 8.807232594738304e-06,
      "loss": 0.4882,
      "step": 17553
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5092450380325317,
      "learning_rate": 8.80622146082822e-06,
      "loss": 0.3917,
      "step": 17554
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5415276288986206,
      "learning_rate": 8.805210339299526e-06,
      "loss": 0.4492,
      "step": 17555
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6976985931396484,
      "learning_rate": 8.80419923016271e-06,
      "loss": 0.3925,
      "step": 17556
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7011852264404297,
      "learning_rate": 8.80318813342826e-06,
      "loss": 0.5013,
      "step": 17557
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.506179690361023,
      "learning_rate": 8.802177049106666e-06,
      "loss": 0.3965,
      "step": 17558
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6140716075897217,
      "learning_rate": 8.801165977208407e-06,
      "loss": 0.4502,
      "step": 17559
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.541587471961975,
      "learning_rate": 8.800154917743976e-06,
      "loss": 0.4103,
      "step": 17560
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.45269775390625,
      "learning_rate": 8.799143870723855e-06,
      "loss": 0.4175,
      "step": 17561
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5267958641052246,
      "learning_rate": 8.798132836158537e-06,
      "loss": 0.423,
      "step": 17562
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.52967369556427,
      "learning_rate": 8.797121814058502e-06,
      "loss": 0.4673,
      "step": 17563
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5795055627822876,
      "learning_rate": 8.796110804434233e-06,
      "loss": 0.4321,
      "step": 17564
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6781749725341797,
      "learning_rate": 8.795099807296225e-06,
      "loss": 0.4451,
      "step": 17565
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5200812816619873,
      "learning_rate": 8.79408882265496e-06,
      "loss": 0.413,
      "step": 17566
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4965050220489502,
      "learning_rate": 8.793077850520916e-06,
      "loss": 0.4315,
      "step": 17567
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.549797534942627,
      "learning_rate": 8.792066890904589e-06,
      "loss": 0.4158,
      "step": 17568
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.593965768814087,
      "learning_rate": 8.791055943816456e-06,
      "loss": 0.4681,
      "step": 17569
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4387414455413818,
      "learning_rate": 8.790045009267012e-06,
      "loss": 0.3916,
      "step": 17570
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4872676134109497,
      "learning_rate": 8.789034087266733e-06,
      "loss": 0.4499,
      "step": 17571
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4725335836410522,
      "learning_rate": 8.788023177826105e-06,
      "loss": 0.4125,
      "step": 17572
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8654537200927734,
      "learning_rate": 8.787012280955616e-06,
      "loss": 0.4756,
      "step": 17573
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5642056465148926,
      "learning_rate": 8.786001396665752e-06,
      "loss": 0.4576,
      "step": 17574
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.666794776916504,
      "learning_rate": 8.784990524966993e-06,
      "loss": 0.4867,
      "step": 17575
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5928266048431396,
      "learning_rate": 8.783979665869824e-06,
      "loss": 0.4283,
      "step": 17576
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4968243837356567,
      "learning_rate": 8.782968819384731e-06,
      "loss": 0.4448,
      "step": 17577
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4713269472122192,
      "learning_rate": 8.7819579855222e-06,
      "loss": 0.402,
      "step": 17578
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5799607038497925,
      "learning_rate": 8.780947164292709e-06,
      "loss": 0.4379,
      "step": 17579
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5563561916351318,
      "learning_rate": 8.779936355706747e-06,
      "loss": 0.4433,
      "step": 17580
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5998789072036743,
      "learning_rate": 8.778925559774793e-06,
      "loss": 0.4736,
      "step": 17581
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4915834665298462,
      "learning_rate": 8.777914776507339e-06,
      "loss": 0.4182,
      "step": 17582
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5448627471923828,
      "learning_rate": 8.77690400591486e-06,
      "loss": 0.4584,
      "step": 17583
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.581015706062317,
      "learning_rate": 8.77589324800784e-06,
      "loss": 0.3946,
      "step": 17584
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6673611402511597,
      "learning_rate": 8.774882502796767e-06,
      "loss": 0.4513,
      "step": 17585
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.43796706199646,
      "learning_rate": 8.773871770292118e-06,
      "loss": 0.3693,
      "step": 17586
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4957852363586426,
      "learning_rate": 8.772861050504386e-06,
      "loss": 0.4304,
      "step": 17587
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.55524742603302,
      "learning_rate": 8.771850343444043e-06,
      "loss": 0.426,
      "step": 17588
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6417096853256226,
      "learning_rate": 8.770839649121576e-06,
      "loss": 0.4606,
      "step": 17589
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5537036657333374,
      "learning_rate": 8.769828967547467e-06,
      "loss": 0.3963,
      "step": 17590
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6509121656417847,
      "learning_rate": 8.768818298732202e-06,
      "loss": 0.4342,
      "step": 17591
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5153210163116455,
      "learning_rate": 8.767807642686256e-06,
      "loss": 0.3987,
      "step": 17592
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5644702911376953,
      "learning_rate": 8.766796999420117e-06,
      "loss": 0.4839,
      "step": 17593
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.527243971824646,
      "learning_rate": 8.765786368944264e-06,
      "loss": 0.4069,
      "step": 17594
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5466705560684204,
      "learning_rate": 8.764775751269184e-06,
      "loss": 0.4337,
      "step": 17595
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.548532485961914,
      "learning_rate": 8.76376514640535e-06,
      "loss": 0.392,
      "step": 17596
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6733956336975098,
      "learning_rate": 8.762754554363248e-06,
      "loss": 0.4679,
      "step": 17597
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5058844089508057,
      "learning_rate": 8.761743975153362e-06,
      "loss": 0.3873,
      "step": 17598
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6130659580230713,
      "learning_rate": 8.760733408786174e-06,
      "loss": 0.4668,
      "step": 17599
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6899603605270386,
      "learning_rate": 8.759722855272157e-06,
      "loss": 0.4212,
      "step": 17600
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.565712332725525,
      "learning_rate": 8.758712314621799e-06,
      "loss": 0.4668,
      "step": 17601
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.508553147315979,
      "learning_rate": 8.757701786845576e-06,
      "loss": 0.4206,
      "step": 17602
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.620926022529602,
      "learning_rate": 8.756691271953979e-06,
      "loss": 0.51,
      "step": 17603
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4746391773223877,
      "learning_rate": 8.75568076995748e-06,
      "loss": 0.4146,
      "step": 17604
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.543367624282837,
      "learning_rate": 8.754670280866556e-06,
      "loss": 0.4278,
      "step": 17605
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4266875982284546,
      "learning_rate": 8.753659804691696e-06,
      "loss": 0.3954,
      "step": 17606
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.536265254020691,
      "learning_rate": 8.75264934144338e-06,
      "loss": 0.4149,
      "step": 17607
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.663851261138916,
      "learning_rate": 8.75163889113208e-06,
      "loss": 0.4228,
      "step": 17608
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5950757265090942,
      "learning_rate": 8.750628453768281e-06,
      "loss": 0.4373,
      "step": 17609
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4858272075653076,
      "learning_rate": 8.749618029362462e-06,
      "loss": 0.433,
      "step": 17610
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5166141986846924,
      "learning_rate": 8.74860761792511e-06,
      "loss": 0.4398,
      "step": 17611
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5029834508895874,
      "learning_rate": 8.747597219466693e-06,
      "loss": 0.4062,
      "step": 17612
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.928711175918579,
      "learning_rate": 8.746586833997695e-06,
      "loss": 1.0047,
      "step": 17613
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5209074020385742,
      "learning_rate": 8.745576461528597e-06,
      "loss": 1.0289,
      "step": 17614
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.586356520652771,
      "learning_rate": 8.74456610206988e-06,
      "loss": 0.5009,
      "step": 17615
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6011292934417725,
      "learning_rate": 8.743555755632018e-06,
      "loss": 0.4082,
      "step": 17616
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6180322170257568,
      "learning_rate": 8.74254542222549e-06,
      "loss": 0.482,
      "step": 17617
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6169582605361938,
      "learning_rate": 8.741535101860777e-06,
      "loss": 0.3911,
      "step": 17618
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5815454721450806,
      "learning_rate": 8.740524794548361e-06,
      "loss": 0.4971,
      "step": 17619
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5752779245376587,
      "learning_rate": 8.739514500298713e-06,
      "loss": 0.4434,
      "step": 17620
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5788689851760864,
      "learning_rate": 8.738504219122316e-06,
      "loss": 0.4807,
      "step": 17621
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.511642336845398,
      "learning_rate": 8.737493951029647e-06,
      "loss": 0.3946,
      "step": 17622
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5915818214416504,
      "learning_rate": 8.736483696031186e-06,
      "loss": 0.5274,
      "step": 17623
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5138366222381592,
      "learning_rate": 8.735473454137412e-06,
      "loss": 0.4089,
      "step": 17624
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7998754978179932,
      "learning_rate": 8.734463225358795e-06,
      "loss": 0.4422,
      "step": 17625
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4662519693374634,
      "learning_rate": 8.73345300970582e-06,
      "loss": 0.3921,
      "step": 17626
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5345103740692139,
      "learning_rate": 8.732442807188966e-06,
      "loss": 0.4692,
      "step": 17627
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.58292555809021,
      "learning_rate": 8.731432617818703e-06,
      "loss": 0.4429,
      "step": 17628
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6022791862487793,
      "learning_rate": 8.730422441605513e-06,
      "loss": 0.4847,
      "step": 17629
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.599550485610962,
      "learning_rate": 8.72941227855987e-06,
      "loss": 0.4558,
      "step": 17630
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5514003038406372,
      "learning_rate": 8.72840212869226e-06,
      "loss": 0.4709,
      "step": 17631
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6253217458724976,
      "learning_rate": 8.72739199201315e-06,
      "loss": 0.4136,
      "step": 17632
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4848805665969849,
      "learning_rate": 8.726381868533018e-06,
      "loss": 0.4838,
      "step": 17633
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.551416277885437,
      "learning_rate": 8.725371758262344e-06,
      "loss": 0.3919,
      "step": 17634
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5624017715454102,
      "learning_rate": 8.724361661211607e-06,
      "loss": 0.5044,
      "step": 17635
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5993915796279907,
      "learning_rate": 8.723351577391272e-06,
      "loss": 0.4156,
      "step": 17636
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5112608671188354,
      "learning_rate": 8.722341506811828e-06,
      "loss": 0.4651,
      "step": 17637
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4593801498413086,
      "learning_rate": 8.721331449483741e-06,
      "loss": 0.4539,
      "step": 17638
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5027821063995361,
      "learning_rate": 8.720321405417498e-06,
      "loss": 0.4907,
      "step": 17639
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5899773836135864,
      "learning_rate": 8.719311374623565e-06,
      "loss": 0.4457,
      "step": 17640
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5681759119033813,
      "learning_rate": 8.71830135711242e-06,
      "loss": 0.5032,
      "step": 17641
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4446581602096558,
      "learning_rate": 8.71729135289454e-06,
      "loss": 0.396,
      "step": 17642
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6020797491073608,
      "learning_rate": 8.716281361980404e-06,
      "loss": 0.4553,
      "step": 17643
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.483760118484497,
      "learning_rate": 8.715271384380477e-06,
      "loss": 0.3813,
      "step": 17644
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.502186894416809,
      "learning_rate": 8.714261420105242e-06,
      "loss": 0.4758,
      "step": 17645
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4939519166946411,
      "learning_rate": 8.71325146916517e-06,
      "loss": 0.4096,
      "step": 17646
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.579003095626831,
      "learning_rate": 8.712241531570744e-06,
      "loss": 0.4545,
      "step": 17647
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6727405786514282,
      "learning_rate": 8.711231607332429e-06,
      "loss": 0.3936,
      "step": 17648
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6356253623962402,
      "learning_rate": 8.7102216964607e-06,
      "loss": 0.5011,
      "step": 17649
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.513860821723938,
      "learning_rate": 8.709211798966038e-06,
      "loss": 0.4082,
      "step": 17650
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.622527837753296,
      "learning_rate": 8.708201914858918e-06,
      "loss": 0.4863,
      "step": 17651
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5778864622116089,
      "learning_rate": 8.707192044149803e-06,
      "loss": 0.4421,
      "step": 17652
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6738601922988892,
      "learning_rate": 8.706182186849175e-06,
      "loss": 0.4157,
      "step": 17653
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5016096830368042,
      "learning_rate": 8.705172342967507e-06,
      "loss": 0.4078,
      "step": 17654
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5445297956466675,
      "learning_rate": 8.704162512515276e-06,
      "loss": 0.4694,
      "step": 17655
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5112245082855225,
      "learning_rate": 8.703152695502946e-06,
      "loss": 0.399,
      "step": 17656
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.559295654296875,
      "learning_rate": 8.702142891941e-06,
      "loss": 0.5105,
      "step": 17657
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4585574865341187,
      "learning_rate": 8.701133101839905e-06,
      "loss": 0.4104,
      "step": 17658
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6390267610549927,
      "learning_rate": 8.700123325210141e-06,
      "loss": 0.5601,
      "step": 17659
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5022369623184204,
      "learning_rate": 8.699113562062173e-06,
      "loss": 0.3974,
      "step": 17660
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.546233892440796,
      "learning_rate": 8.698103812406478e-06,
      "loss": 0.425,
      "step": 17661
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5640376806259155,
      "learning_rate": 8.69709407625353e-06,
      "loss": 0.4167,
      "step": 17662
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5341438055038452,
      "learning_rate": 8.6960843536138e-06,
      "loss": 0.4439,
      "step": 17663
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5605446100234985,
      "learning_rate": 8.695074644497757e-06,
      "loss": 0.3963,
      "step": 17664
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6315251588821411,
      "learning_rate": 8.69406494891588e-06,
      "loss": 0.5044,
      "step": 17665
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5085283517837524,
      "learning_rate": 8.693055266878633e-06,
      "loss": 0.3829,
      "step": 17666
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.601057767868042,
      "learning_rate": 8.6920455983965e-06,
      "loss": 0.502,
      "step": 17667
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5133939981460571,
      "learning_rate": 8.691035943479943e-06,
      "loss": 0.408,
      "step": 17668
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5253772735595703,
      "learning_rate": 8.690026302139433e-06,
      "loss": 0.4458,
      "step": 17669
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.527355670928955,
      "learning_rate": 8.689016674385447e-06,
      "loss": 0.4068,
      "step": 17670
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5754896402359009,
      "learning_rate": 8.688007060228459e-06,
      "loss": 0.4871,
      "step": 17671
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6212644577026367,
      "learning_rate": 8.68699745967893e-06,
      "loss": 0.4454,
      "step": 17672
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.562309741973877,
      "learning_rate": 8.685987872747338e-06,
      "loss": 0.4497,
      "step": 17673
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4888575077056885,
      "learning_rate": 8.684978299444152e-06,
      "loss": 0.4033,
      "step": 17674
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.62721586227417,
      "learning_rate": 8.683968739779848e-06,
      "loss": 0.4931,
      "step": 17675
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5815616846084595,
      "learning_rate": 8.682959193764891e-06,
      "loss": 0.4077,
      "step": 17676
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5920149087905884,
      "learning_rate": 8.681949661409749e-06,
      "loss": 0.4708,
      "step": 17677
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6347788572311401,
      "learning_rate": 8.6809401427249e-06,
      "loss": 0.4019,
      "step": 17678
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6070289611816406,
      "learning_rate": 8.679930637720813e-06,
      "loss": 0.4572,
      "step": 17679
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6315505504608154,
      "learning_rate": 8.678921146407953e-06,
      "loss": 0.4201,
      "step": 17680
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.51032292842865,
      "learning_rate": 8.677911668796795e-06,
      "loss": 0.4468,
      "step": 17681
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5221868753433228,
      "learning_rate": 8.676902204897804e-06,
      "loss": 0.4077,
      "step": 17682
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5552676916122437,
      "learning_rate": 8.675892754721458e-06,
      "loss": 0.4421,
      "step": 17683
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5708436965942383,
      "learning_rate": 8.67488331827822e-06,
      "loss": 0.4409,
      "step": 17684
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5554783344268799,
      "learning_rate": 8.673873895578556e-06,
      "loss": 0.4749,
      "step": 17685
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5716619491577148,
      "learning_rate": 8.672864486632945e-06,
      "loss": 0.3988,
      "step": 17686
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6490750312805176,
      "learning_rate": 8.671855091451853e-06,
      "loss": 0.4909,
      "step": 17687
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.488065242767334,
      "learning_rate": 8.670845710045744e-06,
      "loss": 0.3797,
      "step": 17688
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.527658224105835,
      "learning_rate": 8.669836342425091e-06,
      "loss": 0.4659,
      "step": 17689
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.558870553970337,
      "learning_rate": 8.66882698860036e-06,
      "loss": 0.4251,
      "step": 17690
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.634140968322754,
      "learning_rate": 8.66781764858203e-06,
      "loss": 0.4953,
      "step": 17691
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5251847505569458,
      "learning_rate": 8.666808322380555e-06,
      "loss": 0.4121,
      "step": 17692
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.590196132659912,
      "learning_rate": 8.66579901000641e-06,
      "loss": 0.4417,
      "step": 17693
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5163027048110962,
      "learning_rate": 8.664789711470061e-06,
      "loss": 0.413,
      "step": 17694
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7590340375900269,
      "learning_rate": 8.663780426781982e-06,
      "loss": 0.4557,
      "step": 17695
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5821692943572998,
      "learning_rate": 8.662771155952635e-06,
      "loss": 0.4316,
      "step": 17696
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5447821617126465,
      "learning_rate": 8.661761898992487e-06,
      "loss": 0.4648,
      "step": 17697
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.489575743675232,
      "learning_rate": 8.66075265591201e-06,
      "loss": 0.3975,
      "step": 17698
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5148886442184448,
      "learning_rate": 8.659743426721672e-06,
      "loss": 0.4381,
      "step": 17699
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.530776858329773,
      "learning_rate": 8.658734211431935e-06,
      "loss": 0.4201,
      "step": 17700
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5231846570968628,
      "learning_rate": 8.657725010053268e-06,
      "loss": 0.4352,
      "step": 17701
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5438662767410278,
      "learning_rate": 8.656715822596139e-06,
      "loss": 0.4141,
      "step": 17702
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.532288908958435,
      "learning_rate": 8.655706649071018e-06,
      "loss": 0.4321,
      "step": 17703
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.556119680404663,
      "learning_rate": 8.654697489488369e-06,
      "loss": 0.3884,
      "step": 17704
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5353596210479736,
      "learning_rate": 8.653688343858655e-06,
      "loss": 0.4507,
      "step": 17705
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.525076150894165,
      "learning_rate": 8.652679212192347e-06,
      "loss": 0.411,
      "step": 17706
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5722347497940063,
      "learning_rate": 8.651670094499912e-06,
      "loss": 0.4371,
      "step": 17707
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.578414797782898,
      "learning_rate": 8.65066099079181e-06,
      "loss": 0.3914,
      "step": 17708
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5954169034957886,
      "learning_rate": 8.649651901078513e-06,
      "loss": 0.4757,
      "step": 17709
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4861104488372803,
      "learning_rate": 8.648642825370485e-06,
      "loss": 0.4006,
      "step": 17710
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5258535146713257,
      "learning_rate": 8.647633763678195e-06,
      "loss": 0.4452,
      "step": 17711
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5379449129104614,
      "learning_rate": 8.646624716012104e-06,
      "loss": 0.4029,
      "step": 17712
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4855871200561523,
      "learning_rate": 8.645615682382675e-06,
      "loss": 0.4147,
      "step": 17713
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5419448614120483,
      "learning_rate": 8.64460666280038e-06,
      "loss": 0.4114,
      "step": 17714
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5239976644515991,
      "learning_rate": 8.643597657275685e-06,
      "loss": 0.4589,
      "step": 17715
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5566911697387695,
      "learning_rate": 8.642588665819045e-06,
      "loss": 0.4233,
      "step": 17716
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.543088674545288,
      "learning_rate": 8.641579688440934e-06,
      "loss": 0.412,
      "step": 17717
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8071532249450684,
      "learning_rate": 8.64057072515181e-06,
      "loss": 0.392,
      "step": 17718
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6233831644058228,
      "learning_rate": 8.639561775962149e-06,
      "loss": 0.4596,
      "step": 17719
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5305006504058838,
      "learning_rate": 8.638552840882403e-06,
      "loss": 0.3878,
      "step": 17720
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6351783275604248,
      "learning_rate": 8.63754391992304e-06,
      "loss": 0.4862,
      "step": 17721
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5663983821868896,
      "learning_rate": 8.636535013094528e-06,
      "loss": 0.4616,
      "step": 17722
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6382423639297485,
      "learning_rate": 8.63552612040733e-06,
      "loss": 0.454,
      "step": 17723
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.626213550567627,
      "learning_rate": 8.634517241871903e-06,
      "loss": 0.3952,
      "step": 17724
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6589370965957642,
      "learning_rate": 8.633508377498718e-06,
      "loss": 0.4343,
      "step": 17725
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5185034275054932,
      "learning_rate": 8.632499527298235e-06,
      "loss": 0.3906,
      "step": 17726
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5048978328704834,
      "learning_rate": 8.631490691280925e-06,
      "loss": 0.4173,
      "step": 17727
    },
    {
      "epoch": 0.56,
      "grad_norm": 3.7145237922668457,
      "learning_rate": 8.630481869457238e-06,
      "loss": 0.4213,
      "step": 17728
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.56480872631073,
      "learning_rate": 8.629473061837642e-06,
      "loss": 0.5075,
      "step": 17729
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5598795413970947,
      "learning_rate": 8.628464268432609e-06,
      "loss": 0.412,
      "step": 17730
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5788145065307617,
      "learning_rate": 8.627455489252592e-06,
      "loss": 0.4649,
      "step": 17731
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4584561586380005,
      "learning_rate": 8.626446724308056e-06,
      "loss": 0.3757,
      "step": 17732
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5601212978363037,
      "learning_rate": 8.625437973609463e-06,
      "loss": 0.4614,
      "step": 17733
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5068351030349731,
      "learning_rate": 8.624429237167276e-06,
      "loss": 0.4095,
      "step": 17734
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.439996600151062,
      "learning_rate": 8.623420514991962e-06,
      "loss": 0.4418,
      "step": 17735
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4769744873046875,
      "learning_rate": 8.622411807093974e-06,
      "loss": 0.3962,
      "step": 17736
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.497426986694336,
      "learning_rate": 8.621403113483779e-06,
      "loss": 0.4438,
      "step": 17737
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6307520866394043,
      "learning_rate": 8.620394434171836e-06,
      "loss": 0.4373,
      "step": 17738
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6105469465255737,
      "learning_rate": 8.619385769168615e-06,
      "loss": 0.4553,
      "step": 17739
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5301845073699951,
      "learning_rate": 8.618377118484567e-06,
      "loss": 0.4091,
      "step": 17740
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.572582483291626,
      "learning_rate": 8.617368482130158e-06,
      "loss": 0.48,
      "step": 17741
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.495552659034729,
      "learning_rate": 8.616359860115848e-06,
      "loss": 0.3958,
      "step": 17742
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5783214569091797,
      "learning_rate": 8.615351252452102e-06,
      "loss": 0.4331,
      "step": 17743
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5724201202392578,
      "learning_rate": 8.614342659149376e-06,
      "loss": 0.3925,
      "step": 17744
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6035094261169434,
      "learning_rate": 8.61333408021813e-06,
      "loss": 0.4272,
      "step": 17745
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5369576215744019,
      "learning_rate": 8.612325515668827e-06,
      "loss": 0.3956,
      "step": 17746
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4864963293075562,
      "learning_rate": 8.611316965511934e-06,
      "loss": 0.4278,
      "step": 17747
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.498785138130188,
      "learning_rate": 8.610308429757898e-06,
      "loss": 0.3972,
      "step": 17748
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.542736530303955,
      "learning_rate": 8.609299908417187e-06,
      "loss": 0.4589,
      "step": 17749
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.594995379447937,
      "learning_rate": 8.608291401500262e-06,
      "loss": 0.4156,
      "step": 17750
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5505253076553345,
      "learning_rate": 8.607282909017582e-06,
      "loss": 0.4951,
      "step": 17751
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5625269412994385,
      "learning_rate": 8.6062744309796e-06,
      "loss": 0.3912,
      "step": 17752
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.567302942276001,
      "learning_rate": 8.605265967396783e-06,
      "loss": 0.4657,
      "step": 17753
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.484884262084961,
      "learning_rate": 8.604257518279588e-06,
      "loss": 0.4234,
      "step": 17754
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5618174076080322,
      "learning_rate": 8.603249083638476e-06,
      "loss": 0.4663,
      "step": 17755
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4790425300598145,
      "learning_rate": 8.602240663483906e-06,
      "loss": 0.4166,
      "step": 17756
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5898363590240479,
      "learning_rate": 8.60123225782633e-06,
      "loss": 0.461,
      "step": 17757
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.736840844154358,
      "learning_rate": 8.600223866676216e-06,
      "loss": 0.4169,
      "step": 17758
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6893184185028076,
      "learning_rate": 8.59921549004402e-06,
      "loss": 0.5144,
      "step": 17759
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5320175886154175,
      "learning_rate": 8.598207127940196e-06,
      "loss": 0.4216,
      "step": 17760
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6191388368606567,
      "learning_rate": 8.597198780375206e-06,
      "loss": 0.4487,
      "step": 17761
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6043046712875366,
      "learning_rate": 8.596190447359508e-06,
      "loss": 0.4328,
      "step": 17762
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5223355293273926,
      "learning_rate": 8.595182128903564e-06,
      "loss": 0.4343,
      "step": 17763
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4735535383224487,
      "learning_rate": 8.594173825017823e-06,
      "loss": 0.3986,
      "step": 17764
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4796535968780518,
      "learning_rate": 8.593165535712749e-06,
      "loss": 0.4031,
      "step": 17765
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5377254486083984,
      "learning_rate": 8.592157260998798e-06,
      "loss": 0.4145,
      "step": 17766
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5803052186965942,
      "learning_rate": 8.591149000886429e-06,
      "loss": 0.4375,
      "step": 17767
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5588048696517944,
      "learning_rate": 8.590140755386099e-06,
      "loss": 0.4099,
      "step": 17768
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5820624828338623,
      "learning_rate": 8.589132524508257e-06,
      "loss": 0.4958,
      "step": 17769
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6257091760635376,
      "learning_rate": 8.588124308263373e-06,
      "loss": 0.4047,
      "step": 17770
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6465545892715454,
      "learning_rate": 8.587116106661897e-06,
      "loss": 0.4617,
      "step": 17771
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5267382860183716,
      "learning_rate": 8.586107919714285e-06,
      "loss": 0.401,
      "step": 17772
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5141329765319824,
      "learning_rate": 8.585099747430994e-06,
      "loss": 0.4352,
      "step": 17773
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.483988642692566,
      "learning_rate": 8.584091589822479e-06,
      "loss": 0.4174,
      "step": 17774
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5848499536514282,
      "learning_rate": 8.583083446899204e-06,
      "loss": 0.4639,
      "step": 17775
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.491809368133545,
      "learning_rate": 8.582075318671617e-06,
      "loss": 0.3937,
      "step": 17776
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5841737985610962,
      "learning_rate": 8.581067205150173e-06,
      "loss": 0.463,
      "step": 17777
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4736900329589844,
      "learning_rate": 8.580059106345336e-06,
      "loss": 0.3918,
      "step": 17778
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4984415769577026,
      "learning_rate": 8.579051022267556e-06,
      "loss": 0.4514,
      "step": 17779
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5533382892608643,
      "learning_rate": 8.578042952927285e-06,
      "loss": 0.4017,
      "step": 17780
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5802019834518433,
      "learning_rate": 8.577034898334986e-06,
      "loss": 0.4784,
      "step": 17781
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.487644076347351,
      "learning_rate": 8.57602685850111e-06,
      "loss": 0.4273,
      "step": 17782
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6720247268676758,
      "learning_rate": 8.575018833436113e-06,
      "loss": 0.4535,
      "step": 17783
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.549295425415039,
      "learning_rate": 8.574010823150449e-06,
      "loss": 0.4058,
      "step": 17784
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5598489046096802,
      "learning_rate": 8.573002827654571e-06,
      "loss": 0.4187,
      "step": 17785
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.505662441253662,
      "learning_rate": 8.571994846958937e-06,
      "loss": 0.386,
      "step": 17786
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6582473516464233,
      "learning_rate": 8.570986881074003e-06,
      "loss": 0.5115,
      "step": 17787
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7502340078353882,
      "learning_rate": 8.569978930010215e-06,
      "loss": 0.4018,
      "step": 17788
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5086368322372437,
      "learning_rate": 8.568970993778036e-06,
      "loss": 0.4359,
      "step": 17789
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4958364963531494,
      "learning_rate": 8.567963072387912e-06,
      "loss": 0.3813,
      "step": 17790
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5488297939300537,
      "learning_rate": 8.566955165850306e-06,
      "loss": 0.4386,
      "step": 17791
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5490970611572266,
      "learning_rate": 8.565947274175665e-06,
      "loss": 0.4074,
      "step": 17792
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5492713451385498,
      "learning_rate": 8.56493939737444e-06,
      "loss": 0.4213,
      "step": 17793
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4818288087844849,
      "learning_rate": 8.563931535457091e-06,
      "loss": 0.4072,
      "step": 17794
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.60587477684021,
      "learning_rate": 8.562923688434072e-06,
      "loss": 0.4216,
      "step": 17795
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5356441736221313,
      "learning_rate": 8.561915856315826e-06,
      "loss": 0.3979,
      "step": 17796
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.606688141822815,
      "learning_rate": 8.560908039112817e-06,
      "loss": 0.4966,
      "step": 17797
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4990942478179932,
      "learning_rate": 8.559900236835488e-06,
      "loss": 0.4211,
      "step": 17798
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5957403182983398,
      "learning_rate": 8.558892449494302e-06,
      "loss": 0.4455,
      "step": 17799
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.567629098892212,
      "learning_rate": 8.557884677099704e-06,
      "loss": 0.413,
      "step": 17800
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6332207918167114,
      "learning_rate": 8.556876919662145e-06,
      "loss": 0.4828,
      "step": 17801
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4999768733978271,
      "learning_rate": 8.555869177192082e-06,
      "loss": 0.4133,
      "step": 17802
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5878444910049438,
      "learning_rate": 8.554861449699968e-06,
      "loss": 0.4561,
      "step": 17803
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5093486309051514,
      "learning_rate": 8.553853737196246e-06,
      "loss": 0.3878,
      "step": 17804
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6132854223251343,
      "learning_rate": 8.552846039691376e-06,
      "loss": 0.4408,
      "step": 17805
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5928974151611328,
      "learning_rate": 8.551838357195804e-06,
      "loss": 0.4186,
      "step": 17806
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6343334913253784,
      "learning_rate": 8.550830689719991e-06,
      "loss": 0.4449,
      "step": 17807
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5581698417663574,
      "learning_rate": 8.549823037274375e-06,
      "loss": 0.4289,
      "step": 17808
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5335144996643066,
      "learning_rate": 8.548815399869414e-06,
      "loss": 0.4156,
      "step": 17809
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5755870342254639,
      "learning_rate": 8.547807777515558e-06,
      "loss": 0.4229,
      "step": 17810
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5324653387069702,
      "learning_rate": 8.54680017022326e-06,
      "loss": 0.471,
      "step": 17811
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5775917768478394,
      "learning_rate": 8.545792578002968e-06,
      "loss": 0.4143,
      "step": 17812
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6027066707611084,
      "learning_rate": 8.544785000865128e-06,
      "loss": 0.4969,
      "step": 17813
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.654019832611084,
      "learning_rate": 8.543777438820197e-06,
      "loss": 0.4038,
      "step": 17814
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5273367166519165,
      "learning_rate": 8.542769891878625e-06,
      "loss": 0.4634,
      "step": 17815
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.573457956314087,
      "learning_rate": 8.541762360050854e-06,
      "loss": 0.3994,
      "step": 17816
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5422613620758057,
      "learning_rate": 8.540754843347343e-06,
      "loss": 0.4391,
      "step": 17817
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5694780349731445,
      "learning_rate": 8.539747341778535e-06,
      "loss": 0.417,
      "step": 17818
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5681957006454468,
      "learning_rate": 8.538739855354887e-06,
      "loss": 0.4818,
      "step": 17819
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5612224340438843,
      "learning_rate": 8.53773238408684e-06,
      "loss": 0.4313,
      "step": 17820
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5863280296325684,
      "learning_rate": 8.536724927984845e-06,
      "loss": 0.4739,
      "step": 17821
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5461069345474243,
      "learning_rate": 8.535717487059354e-06,
      "loss": 0.4345,
      "step": 17822
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5769423246383667,
      "learning_rate": 8.534710061320819e-06,
      "loss": 0.4427,
      "step": 17823
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6112957000732422,
      "learning_rate": 8.533702650779675e-06,
      "loss": 0.4012,
      "step": 17824
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6750938892364502,
      "learning_rate": 8.532695255446384e-06,
      "loss": 0.4405,
      "step": 17825
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5925312042236328,
      "learning_rate": 8.531687875331387e-06,
      "loss": 0.4099,
      "step": 17826
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5420087575912476,
      "learning_rate": 8.530680510445137e-06,
      "loss": 0.4632,
      "step": 17827
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.558881402015686,
      "learning_rate": 8.529673160798082e-06,
      "loss": 0.3849,
      "step": 17828
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5881816148757935,
      "learning_rate": 8.528665826400662e-06,
      "loss": 0.4479,
      "step": 17829
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5432708263397217,
      "learning_rate": 8.527658507263334e-06,
      "loss": 0.3771,
      "step": 17830
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.506179690361023,
      "learning_rate": 8.52665120339654e-06,
      "loss": 0.4236,
      "step": 17831
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5914177894592285,
      "learning_rate": 8.52564391481073e-06,
      "loss": 0.3977,
      "step": 17832
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6633621454238892,
      "learning_rate": 8.524636641516352e-06,
      "loss": 0.9918,
      "step": 17833
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5019338130950928,
      "learning_rate": 8.523629383523845e-06,
      "loss": 0.9835,
      "step": 17834
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6536669731140137,
      "learning_rate": 8.522622140843668e-06,
      "loss": 0.4726,
      "step": 17835
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5161237716674805,
      "learning_rate": 8.521614913486263e-06,
      "loss": 0.3585,
      "step": 17836
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.652029275894165,
      "learning_rate": 8.52060770146207e-06,
      "loss": 0.4667,
      "step": 17837
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5319902896881104,
      "learning_rate": 8.519600504781544e-06,
      "loss": 0.4065,
      "step": 17838
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5588314533233643,
      "learning_rate": 8.518593323455126e-06,
      "loss": 0.4595,
      "step": 17839
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5243996381759644,
      "learning_rate": 8.517586157493267e-06,
      "loss": 0.3907,
      "step": 17840
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5734611749649048,
      "learning_rate": 8.516579006906408e-06,
      "loss": 0.4566,
      "step": 17841
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5237287282943726,
      "learning_rate": 8.515571871704997e-06,
      "loss": 0.3961,
      "step": 17842
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8395028114318848,
      "learning_rate": 8.51456475189948e-06,
      "loss": 0.9938,
      "step": 17843
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4913523197174072,
      "learning_rate": 8.513557647500303e-06,
      "loss": 1.0083,
      "step": 17844
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5330528020858765,
      "learning_rate": 8.51255055851791e-06,
      "loss": 0.4398,
      "step": 17845
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5484291315078735,
      "learning_rate": 8.511543484962743e-06,
      "loss": 0.3904,
      "step": 17846
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.607250452041626,
      "learning_rate": 8.510536426845253e-06,
      "loss": 0.4563,
      "step": 17847
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4891855716705322,
      "learning_rate": 8.509529384175882e-06,
      "loss": 0.4152,
      "step": 17848
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6617038249969482,
      "learning_rate": 8.508522356965073e-06,
      "loss": 0.4878,
      "step": 17849
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.575213074684143,
      "learning_rate": 8.507515345223274e-06,
      "loss": 0.401,
      "step": 17850
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4909510612487793,
      "learning_rate": 8.506508348960924e-06,
      "loss": 0.4591,
      "step": 17851
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6050775051116943,
      "learning_rate": 8.505501368188476e-06,
      "loss": 0.4287,
      "step": 17852
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5532468557357788,
      "learning_rate": 8.504494402916366e-06,
      "loss": 0.4324,
      "step": 17853
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5292589664459229,
      "learning_rate": 8.503487453155038e-06,
      "loss": 0.3887,
      "step": 17854
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5809327363967896,
      "learning_rate": 8.502480518914942e-06,
      "loss": 0.4389,
      "step": 17855
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.484717607498169,
      "learning_rate": 8.501473600206517e-06,
      "loss": 0.3917,
      "step": 17856
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6204924583435059,
      "learning_rate": 8.500466697040205e-06,
      "loss": 0.5269,
      "step": 17857
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5476850271224976,
      "learning_rate": 8.499459809426453e-06,
      "loss": 0.4229,
      "step": 17858
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5774610042572021,
      "learning_rate": 8.4984529373757e-06,
      "loss": 0.4646,
      "step": 17859
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6014950275421143,
      "learning_rate": 8.497446080898396e-06,
      "loss": 0.4123,
      "step": 17860
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.546743631362915,
      "learning_rate": 8.496439240004975e-06,
      "loss": 0.4243,
      "step": 17861
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5129625797271729,
      "learning_rate": 8.495432414705882e-06,
      "loss": 0.4241,
      "step": 17862
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4897315502166748,
      "learning_rate": 8.494425605011563e-06,
      "loss": 0.4398,
      "step": 17863
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5472899675369263,
      "learning_rate": 8.49341881093246e-06,
      "loss": 0.404,
      "step": 17864
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4949041604995728,
      "learning_rate": 8.492412032479009e-06,
      "loss": 0.4628,
      "step": 17865
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4832533597946167,
      "learning_rate": 8.491405269661658e-06,
      "loss": 0.4032,
      "step": 17866
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.676316261291504,
      "learning_rate": 8.490398522490844e-06,
      "loss": 0.4616,
      "step": 17867
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5654469728469849,
      "learning_rate": 8.489391790977017e-06,
      "loss": 0.4086,
      "step": 17868
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5922247171401978,
      "learning_rate": 8.48838507513061e-06,
      "loss": 0.4531,
      "step": 17869
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5039827823638916,
      "learning_rate": 8.487378374962066e-06,
      "loss": 0.4005,
      "step": 17870
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5353641510009766,
      "learning_rate": 8.486371690481827e-06,
      "loss": 0.4559,
      "step": 17871
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.513585090637207,
      "learning_rate": 8.485365021700336e-06,
      "loss": 0.406,
      "step": 17872
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5253700017929077,
      "learning_rate": 8.48435836862803e-06,
      "loss": 0.4552,
      "step": 17873
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6572374105453491,
      "learning_rate": 8.483351731275351e-06,
      "loss": 0.468,
      "step": 17874
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5016772747039795,
      "learning_rate": 8.482345109652738e-06,
      "loss": 0.4453,
      "step": 17875
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5667575597763062,
      "learning_rate": 8.48133850377064e-06,
      "loss": 0.4251,
      "step": 17876
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5416474342346191,
      "learning_rate": 8.480331913639485e-06,
      "loss": 0.4645,
      "step": 17877
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5944229364395142,
      "learning_rate": 8.479325339269716e-06,
      "loss": 0.4379,
      "step": 17878
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.605886697769165,
      "learning_rate": 8.478318780671779e-06,
      "loss": 0.4727,
      "step": 17879
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4336583614349365,
      "learning_rate": 8.477312237856112e-06,
      "loss": 0.3915,
      "step": 17880
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6652132272720337,
      "learning_rate": 8.476305710833145e-06,
      "loss": 0.5057,
      "step": 17881
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4817980527877808,
      "learning_rate": 8.475299199613328e-06,
      "loss": 0.4016,
      "step": 17882
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5615133047103882,
      "learning_rate": 8.474292704207095e-06,
      "loss": 0.5202,
      "step": 17883
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5297880172729492,
      "learning_rate": 8.473286224624892e-06,
      "loss": 0.4092,
      "step": 17884
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5478891134262085,
      "learning_rate": 8.472279760877145e-06,
      "loss": 0.4539,
      "step": 17885
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4434387683868408,
      "learning_rate": 8.471273312974302e-06,
      "loss": 0.4166,
      "step": 17886
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.624405860900879,
      "learning_rate": 8.470266880926799e-06,
      "loss": 0.4541,
      "step": 17887
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.728301763534546,
      "learning_rate": 8.469260464745077e-06,
      "loss": 0.4244,
      "step": 17888
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.616629958152771,
      "learning_rate": 8.46825406443957e-06,
      "loss": 0.4796,
      "step": 17889
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4776546955108643,
      "learning_rate": 8.467247680020718e-06,
      "loss": 0.398,
      "step": 17890
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6185754537582397,
      "learning_rate": 8.466241311498957e-06,
      "loss": 0.4862,
      "step": 17891
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.548805594444275,
      "learning_rate": 8.46523495888473e-06,
      "loss": 0.4008,
      "step": 17892
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.59040105342865,
      "learning_rate": 8.464228622188467e-06,
      "loss": 0.5111,
      "step": 17893
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5553101301193237,
      "learning_rate": 8.46322230142061e-06,
      "loss": 0.4508,
      "step": 17894
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5269750356674194,
      "learning_rate": 8.462215996591594e-06,
      "loss": 0.4247,
      "step": 17895
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5222065448760986,
      "learning_rate": 8.461209707711861e-06,
      "loss": 0.414,
      "step": 17896
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5044410228729248,
      "learning_rate": 8.460203434791843e-06,
      "loss": 0.4354,
      "step": 17897
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.567142367362976,
      "learning_rate": 8.459197177841974e-06,
      "loss": 0.4117,
      "step": 17898
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5752232074737549,
      "learning_rate": 8.458190936872696e-06,
      "loss": 0.4574,
      "step": 17899
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5025112628936768,
      "learning_rate": 8.457184711894447e-06,
      "loss": 0.4111,
      "step": 17900
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.644531011581421,
      "learning_rate": 8.456178502917654e-06,
      "loss": 0.4679,
      "step": 17901
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5999661684036255,
      "learning_rate": 8.455172309952762e-06,
      "loss": 0.4562,
      "step": 17902
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.676957130432129,
      "learning_rate": 8.4541661330102e-06,
      "loss": 0.4475,
      "step": 17903
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.589082956314087,
      "learning_rate": 8.453159972100411e-06,
      "loss": 0.4004,
      "step": 17904
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6335495710372925,
      "learning_rate": 8.452153827233826e-06,
      "loss": 0.52,
      "step": 17905
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5378336906433105,
      "learning_rate": 8.451147698420879e-06,
      "loss": 0.4163,
      "step": 17906
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5262807607650757,
      "learning_rate": 8.45014158567201e-06,
      "loss": 0.4251,
      "step": 17907
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5155853033065796,
      "learning_rate": 8.44913548899765e-06,
      "loss": 0.4051,
      "step": 17908
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6522958278656006,
      "learning_rate": 8.448129408408234e-06,
      "loss": 0.5006,
      "step": 17909
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7055013179779053,
      "learning_rate": 8.447123343914198e-06,
      "loss": 0.4283,
      "step": 17910
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6047672033309937,
      "learning_rate": 8.446117295525974e-06,
      "loss": 0.5013,
      "step": 17911
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.8374860286712646,
      "learning_rate": 8.445111263254003e-06,
      "loss": 0.446,
      "step": 17912
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6308797597885132,
      "learning_rate": 8.444105247108711e-06,
      "loss": 0.4534,
      "step": 17913
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5505181550979614,
      "learning_rate": 8.443099247100535e-06,
      "loss": 0.4238,
      "step": 17914
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6670807600021362,
      "learning_rate": 8.442093263239913e-06,
      "loss": 0.4342,
      "step": 17915
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6187407970428467,
      "learning_rate": 8.441087295537275e-06,
      "loss": 0.4339,
      "step": 17916
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.512725830078125,
      "learning_rate": 8.44008134400305e-06,
      "loss": 0.4453,
      "step": 17917
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5781974792480469,
      "learning_rate": 8.43907540864768e-06,
      "loss": 0.4387,
      "step": 17918
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6265372037887573,
      "learning_rate": 8.43806948948159e-06,
      "loss": 0.4539,
      "step": 17919
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.627238154411316,
      "learning_rate": 8.437063586515225e-06,
      "loss": 0.4144,
      "step": 17920
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5790317058563232,
      "learning_rate": 8.436057699759001e-06,
      "loss": 0.4817,
      "step": 17921
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5816051959991455,
      "learning_rate": 8.435051829223362e-06,
      "loss": 0.43,
      "step": 17922
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.524797797203064,
      "learning_rate": 8.43404597491874e-06,
      "loss": 0.441,
      "step": 17923
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5286885499954224,
      "learning_rate": 8.433040136855567e-06,
      "loss": 0.3831,
      "step": 17924
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5961089134216309,
      "learning_rate": 8.43203431504427e-06,
      "loss": 0.4658,
      "step": 17925
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5729215145111084,
      "learning_rate": 8.431028509495284e-06,
      "loss": 0.4817,
      "step": 17926
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6993207931518555,
      "learning_rate": 8.430022720219043e-06,
      "loss": 0.476,
      "step": 17927
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5860660076141357,
      "learning_rate": 8.42901694722598e-06,
      "loss": 0.4296,
      "step": 17928
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.534533977508545,
      "learning_rate": 8.428011190526517e-06,
      "loss": 0.445,
      "step": 17929
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5058529376983643,
      "learning_rate": 8.427005450131095e-06,
      "loss": 0.3682,
      "step": 17930
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.518179178237915,
      "learning_rate": 8.42599972605014e-06,
      "loss": 0.4479,
      "step": 17931
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5644845962524414,
      "learning_rate": 8.424994018294089e-06,
      "loss": 0.4296,
      "step": 17932
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5681359767913818,
      "learning_rate": 8.423988326873367e-06,
      "loss": 0.4679,
      "step": 17933
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.467241883277893,
      "learning_rate": 8.422982651798402e-06,
      "loss": 0.432,
      "step": 17934
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5783783197402954,
      "learning_rate": 8.421976993079633e-06,
      "loss": 0.4399,
      "step": 17935
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6225858926773071,
      "learning_rate": 8.420971350727487e-06,
      "loss": 0.4154,
      "step": 17936
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5766377449035645,
      "learning_rate": 8.419965724752388e-06,
      "loss": 0.5149,
      "step": 17937
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5706108808517456,
      "learning_rate": 8.418960115164775e-06,
      "loss": 0.4052,
      "step": 17938
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6319490671157837,
      "learning_rate": 8.41795452197507e-06,
      "loss": 0.5056,
      "step": 17939
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4947752952575684,
      "learning_rate": 8.416948945193712e-06,
      "loss": 0.3992,
      "step": 17940
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.558608055114746,
      "learning_rate": 8.415943384831121e-06,
      "loss": 0.4565,
      "step": 17941
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5463175773620605,
      "learning_rate": 8.414937840897731e-06,
      "loss": 0.4271,
      "step": 17942
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5518640279769897,
      "learning_rate": 8.413932313403971e-06,
      "loss": 0.433,
      "step": 17943
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4386800527572632,
      "learning_rate": 8.41292680236027e-06,
      "loss": 0.3742,
      "step": 17944
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5446679592132568,
      "learning_rate": 8.411921307777054e-06,
      "loss": 0.4747,
      "step": 17945
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5361026525497437,
      "learning_rate": 8.410915829664752e-06,
      "loss": 0.4235,
      "step": 17946
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4964944124221802,
      "learning_rate": 8.409910368033796e-06,
      "loss": 0.4035,
      "step": 17947
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.540183186531067,
      "learning_rate": 8.408904922894613e-06,
      "loss": 0.401,
      "step": 17948
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5084779262542725,
      "learning_rate": 8.40789949425763e-06,
      "loss": 0.4485,
      "step": 17949
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6190892457962036,
      "learning_rate": 8.406894082133272e-06,
      "loss": 0.4,
      "step": 17950
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5502265691757202,
      "learning_rate": 8.405888686531973e-06,
      "loss": 0.426,
      "step": 17951
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.490643858909607,
      "learning_rate": 8.40488330746416e-06,
      "loss": 0.4149,
      "step": 17952
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5655707120895386,
      "learning_rate": 8.403877944940252e-06,
      "loss": 0.4629,
      "step": 17953
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4856915473937988,
      "learning_rate": 8.402872598970686e-06,
      "loss": 0.3848,
      "step": 17954
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5638729333877563,
      "learning_rate": 8.40186726956588e-06,
      "loss": 0.4605,
      "step": 17955
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.546239972114563,
      "learning_rate": 8.400861956736273e-06,
      "loss": 0.4138,
      "step": 17956
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5562971830368042,
      "learning_rate": 8.399856660492281e-06,
      "loss": 0.4572,
      "step": 17957
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4870810508728027,
      "learning_rate": 8.398851380844332e-06,
      "loss": 0.3808,
      "step": 17958
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.590132474899292,
      "learning_rate": 8.397846117802857e-06,
      "loss": 0.4401,
      "step": 17959
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6184918880462646,
      "learning_rate": 8.396840871378284e-06,
      "loss": 0.4261,
      "step": 17960
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.701210856437683,
      "learning_rate": 8.395835641581031e-06,
      "loss": 0.4476,
      "step": 17961
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.59363853931427,
      "learning_rate": 8.394830428421525e-06,
      "loss": 0.4306,
      "step": 17962
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5668143033981323,
      "learning_rate": 8.393825231910195e-06,
      "loss": 0.4332,
      "step": 17963
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4506287574768066,
      "learning_rate": 8.392820052057471e-06,
      "loss": 0.3901,
      "step": 17964
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5622164011001587,
      "learning_rate": 8.391814888873767e-06,
      "loss": 0.4608,
      "step": 17965
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7423690557479858,
      "learning_rate": 8.390809742369517e-06,
      "loss": 0.4531,
      "step": 17966
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8165620565414429,
      "learning_rate": 8.38980461255514e-06,
      "loss": 0.5273,
      "step": 17967
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5514732599258423,
      "learning_rate": 8.388799499441069e-06,
      "loss": 0.4095,
      "step": 17968
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.823508858680725,
      "learning_rate": 8.38779440303772e-06,
      "loss": 0.4765,
      "step": 17969
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5583654642105103,
      "learning_rate": 8.386789323355521e-06,
      "loss": 0.403,
      "step": 17970
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6149916648864746,
      "learning_rate": 8.385784260404899e-06,
      "loss": 0.4294,
      "step": 17971
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6043474674224854,
      "learning_rate": 8.384779214196276e-06,
      "loss": 0.4333,
      "step": 17972
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4797426462173462,
      "learning_rate": 8.383774184740073e-06,
      "loss": 0.4087,
      "step": 17973
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.557706594467163,
      "learning_rate": 8.382769172046716e-06,
      "loss": 0.408,
      "step": 17974
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5110969543457031,
      "learning_rate": 8.381764176126628e-06,
      "loss": 0.4559,
      "step": 17975
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5913379192352295,
      "learning_rate": 8.380759196990237e-06,
      "loss": 0.4144,
      "step": 17976
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.532326102256775,
      "learning_rate": 8.37975423464796e-06,
      "loss": 0.3729,
      "step": 17977
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6491731405258179,
      "learning_rate": 8.378749289110221e-06,
      "loss": 0.4873,
      "step": 17978
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.616644263267517,
      "learning_rate": 8.377744360387447e-06,
      "loss": 0.4321,
      "step": 17979
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.671684741973877,
      "learning_rate": 8.376739448490058e-06,
      "loss": 0.4111,
      "step": 17980
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5620872974395752,
      "learning_rate": 8.375734553428475e-06,
      "loss": 0.474,
      "step": 17981
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5690512657165527,
      "learning_rate": 8.374729675213123e-06,
      "loss": 0.4353,
      "step": 17982
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7333555221557617,
      "learning_rate": 8.373724813854421e-06,
      "loss": 0.4172,
      "step": 17983
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.513497233390808,
      "learning_rate": 8.372719969362796e-06,
      "loss": 0.3591,
      "step": 17984
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5677586793899536,
      "learning_rate": 8.371715141748669e-06,
      "loss": 0.4339,
      "step": 17985
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.520240306854248,
      "learning_rate": 8.370710331022453e-06,
      "loss": 0.3723,
      "step": 17986
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.574902057647705,
      "learning_rate": 8.369705537194581e-06,
      "loss": 0.446,
      "step": 17987
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5811280012130737,
      "learning_rate": 8.368700760275472e-06,
      "loss": 0.4285,
      "step": 17988
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5865418910980225,
      "learning_rate": 8.367696000275538e-06,
      "loss": 0.4716,
      "step": 17989
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6081798076629639,
      "learning_rate": 8.366691257205211e-06,
      "loss": 0.446,
      "step": 17990
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5227288007736206,
      "learning_rate": 8.365686531074906e-06,
      "loss": 0.4434,
      "step": 17991
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6442785263061523,
      "learning_rate": 8.364681821895048e-06,
      "loss": 0.4019,
      "step": 17992
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6184242963790894,
      "learning_rate": 8.363677129676053e-06,
      "loss": 0.5027,
      "step": 17993
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.539773941040039,
      "learning_rate": 8.36267245442834e-06,
      "loss": 0.416,
      "step": 17994
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5426270961761475,
      "learning_rate": 8.361667796162333e-06,
      "loss": 0.4865,
      "step": 17995
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.559360384941101,
      "learning_rate": 8.360663154888455e-06,
      "loss": 0.4287,
      "step": 17996
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5601136684417725,
      "learning_rate": 8.359658530617116e-06,
      "loss": 0.4476,
      "step": 17997
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6722073554992676,
      "learning_rate": 8.358653923358744e-06,
      "loss": 0.3919,
      "step": 17998
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6651818752288818,
      "learning_rate": 8.357649333123752e-06,
      "loss": 0.5203,
      "step": 17999
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4053521156311035,
      "learning_rate": 8.35664475992257e-06,
      "loss": 0.3718,
      "step": 18000
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5540390014648438,
      "learning_rate": 8.355640203765603e-06,
      "loss": 0.4472,
      "step": 18001
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5039849281311035,
      "learning_rate": 8.35463566466328e-06,
      "loss": 0.4053,
      "step": 18002
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6372530460357666,
      "learning_rate": 8.353631142626012e-06,
      "loss": 0.4309,
      "step": 18003
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4531607627868652,
      "learning_rate": 8.352626637664227e-06,
      "loss": 0.3831,
      "step": 18004
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.555916666984558,
      "learning_rate": 8.351622149788336e-06,
      "loss": 0.467,
      "step": 18005
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5684866905212402,
      "learning_rate": 8.350617679008757e-06,
      "loss": 0.3987,
      "step": 18006
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.555974006652832,
      "learning_rate": 8.349613225335912e-06,
      "loss": 0.4578,
      "step": 18007
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.584007740020752,
      "learning_rate": 8.34860878878022e-06,
      "loss": 0.406,
      "step": 18008
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5955021381378174,
      "learning_rate": 8.34760436935209e-06,
      "loss": 0.4432,
      "step": 18009
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5450847148895264,
      "learning_rate": 8.346599967061947e-06,
      "loss": 0.418,
      "step": 18010
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.602512240409851,
      "learning_rate": 8.345595581920205e-06,
      "loss": 0.4837,
      "step": 18011
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5509198904037476,
      "learning_rate": 8.344591213937287e-06,
      "loss": 0.4032,
      "step": 18012
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.571509599685669,
      "learning_rate": 8.343586863123603e-06,
      "loss": 0.4207,
      "step": 18013
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7433748245239258,
      "learning_rate": 8.34258252948957e-06,
      "loss": 0.4493,
      "step": 18014
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5006788969039917,
      "learning_rate": 8.341578213045606e-06,
      "loss": 0.4215,
      "step": 18015
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.59700608253479,
      "learning_rate": 8.340573913802134e-06,
      "loss": 0.4375,
      "step": 18016
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4684643745422363,
      "learning_rate": 8.339569631769557e-06,
      "loss": 0.4231,
      "step": 18017
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.57634437084198,
      "learning_rate": 8.338565366958301e-06,
      "loss": 0.4565,
      "step": 18018
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5930190086364746,
      "learning_rate": 8.337561119378777e-06,
      "loss": 0.4637,
      "step": 18019
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4616934061050415,
      "learning_rate": 8.336556889041408e-06,
      "loss": 0.3993,
      "step": 18020
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5092594623565674,
      "learning_rate": 8.335552675956599e-06,
      "loss": 0.441,
      "step": 18021
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.490870714187622,
      "learning_rate": 8.334548480134769e-06,
      "loss": 0.4172,
      "step": 18022
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6541415452957153,
      "learning_rate": 8.333544301586338e-06,
      "loss": 0.4908,
      "step": 18023
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.522631287574768,
      "learning_rate": 8.33254014032172e-06,
      "loss": 0.4023,
      "step": 18024
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.544808268547058,
      "learning_rate": 8.33153599635132e-06,
      "loss": 0.424,
      "step": 18025
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5349171161651611,
      "learning_rate": 8.330531869685565e-06,
      "loss": 0.3925,
      "step": 18026
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5396963357925415,
      "learning_rate": 8.32952776033486e-06,
      "loss": 0.4434,
      "step": 18027
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5625495910644531,
      "learning_rate": 8.328523668309629e-06,
      "loss": 0.4406,
      "step": 18028
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6486480236053467,
      "learning_rate": 8.327519593620278e-06,
      "loss": 0.9696,
      "step": 18029
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4643878936767578,
      "learning_rate": 8.32651553627722e-06,
      "loss": 0.995,
      "step": 18030
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5108634233474731,
      "learning_rate": 8.325511496290875e-06,
      "loss": 0.4306,
      "step": 18031
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5477955341339111,
      "learning_rate": 8.324507473671655e-06,
      "loss": 0.4213,
      "step": 18032
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6128944158554077,
      "learning_rate": 8.323503468429968e-06,
      "loss": 0.4803,
      "step": 18033
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.529036283493042,
      "learning_rate": 8.322499480576232e-06,
      "loss": 0.3975,
      "step": 18034
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5677109956741333,
      "learning_rate": 8.321495510120859e-06,
      "loss": 0.466,
      "step": 18035
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4797565937042236,
      "learning_rate": 8.320491557074267e-06,
      "loss": 0.4064,
      "step": 18036
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5371495485305786,
      "learning_rate": 8.319487621446855e-06,
      "loss": 0.4623,
      "step": 18037
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5274276733398438,
      "learning_rate": 8.318483703249047e-06,
      "loss": 0.3999,
      "step": 18038
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4938037395477295,
      "learning_rate": 8.317479802491251e-06,
      "loss": 0.4136,
      "step": 18039
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4755874872207642,
      "learning_rate": 8.316475919183882e-06,
      "loss": 0.378,
      "step": 18040
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5836061239242554,
      "learning_rate": 8.315472053337348e-06,
      "loss": 0.4717,
      "step": 18041
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.52174711227417,
      "learning_rate": 8.314468204962062e-06,
      "loss": 0.4154,
      "step": 18042
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.474286675453186,
      "learning_rate": 8.313464374068438e-06,
      "loss": 0.413,
      "step": 18043
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6241264343261719,
      "learning_rate": 8.312460560666886e-06,
      "loss": 0.4348,
      "step": 18044
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.585656762123108,
      "learning_rate": 8.311456764767813e-06,
      "loss": 0.4699,
      "step": 18045
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5210574865341187,
      "learning_rate": 8.310452986381636e-06,
      "loss": 0.3997,
      "step": 18046
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.560838222503662,
      "learning_rate": 8.309449225518759e-06,
      "loss": 0.4623,
      "step": 18047
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.576854944229126,
      "learning_rate": 8.308445482189604e-06,
      "loss": 0.4063,
      "step": 18048
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6457639932632446,
      "learning_rate": 8.30744175640457e-06,
      "loss": 0.4665,
      "step": 18049
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.597559928894043,
      "learning_rate": 8.306438048174072e-06,
      "loss": 0.4387,
      "step": 18050
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.569085717201233,
      "learning_rate": 8.305434357508519e-06,
      "loss": 0.4319,
      "step": 18051
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6280425786972046,
      "learning_rate": 8.304430684418325e-06,
      "loss": 0.4089,
      "step": 18052
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5022671222686768,
      "learning_rate": 8.303427028913891e-06,
      "loss": 0.4388,
      "step": 18053
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.460074782371521,
      "learning_rate": 8.302423391005635e-06,
      "loss": 0.374,
      "step": 18054
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.583067774772644,
      "learning_rate": 8.30141977070396e-06,
      "loss": 0.4586,
      "step": 18055
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.567548394203186,
      "learning_rate": 8.300416168019283e-06,
      "loss": 0.4267,
      "step": 18056
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7689927816390991,
      "learning_rate": 8.299412582962005e-06,
      "loss": 0.5105,
      "step": 18057
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.569922685623169,
      "learning_rate": 8.298409015542538e-06,
      "loss": 0.4174,
      "step": 18058
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6622214317321777,
      "learning_rate": 8.29740546577129e-06,
      "loss": 0.4775,
      "step": 18059
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.520309567451477,
      "learning_rate": 8.296401933658673e-06,
      "loss": 0.3835,
      "step": 18060
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4797120094299316,
      "learning_rate": 8.295398419215088e-06,
      "loss": 0.4178,
      "step": 18061
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4800989627838135,
      "learning_rate": 8.294394922450948e-06,
      "loss": 0.406,
      "step": 18062
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6601637601852417,
      "learning_rate": 8.29339144337666e-06,
      "loss": 0.5694,
      "step": 18063
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5868510007858276,
      "learning_rate": 8.292387982002636e-06,
      "loss": 0.4016,
      "step": 18064
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5771664381027222,
      "learning_rate": 8.291384538339275e-06,
      "loss": 0.4626,
      "step": 18065
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5659395456314087,
      "learning_rate": 8.290381112396989e-06,
      "loss": 0.414,
      "step": 18066
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6000252962112427,
      "learning_rate": 8.289377704186185e-06,
      "loss": 0.491,
      "step": 18067
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.456385850906372,
      "learning_rate": 8.288374313717267e-06,
      "loss": 0.4212,
      "step": 18068
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.463695526123047,
      "learning_rate": 8.28737094100065e-06,
      "loss": 0.4729,
      "step": 18069
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.533960223197937,
      "learning_rate": 8.286367586046733e-06,
      "loss": 0.4117,
      "step": 18070
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5023902654647827,
      "learning_rate": 8.285364248865922e-06,
      "loss": 0.4441,
      "step": 18071
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5080885887145996,
      "learning_rate": 8.284360929468627e-06,
      "loss": 0.4243,
      "step": 18072
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5254868268966675,
      "learning_rate": 8.283357627865254e-06,
      "loss": 0.5704,
      "step": 18073
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.56479811668396,
      "learning_rate": 8.282354344066206e-06,
      "loss": 0.4192,
      "step": 18074
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5918848514556885,
      "learning_rate": 8.28135107808189e-06,
      "loss": 0.4778,
      "step": 18075
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5504149198532104,
      "learning_rate": 8.28034782992271e-06,
      "loss": 0.4279,
      "step": 18076
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6909089088439941,
      "learning_rate": 8.27934459959908e-06,
      "loss": 0.456,
      "step": 18077
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6332873106002808,
      "learning_rate": 8.27834138712139e-06,
      "loss": 0.4256,
      "step": 18078
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5763434171676636,
      "learning_rate": 8.277338192500057e-06,
      "loss": 0.4697,
      "step": 18079
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6646884679794312,
      "learning_rate": 8.276335015745478e-06,
      "loss": 0.4058,
      "step": 18080
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5652844905853271,
      "learning_rate": 8.275331856868065e-06,
      "loss": 0.4543,
      "step": 18081
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.576577067375183,
      "learning_rate": 8.274328715878218e-06,
      "loss": 0.4312,
      "step": 18082
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.571128487586975,
      "learning_rate": 8.27332559278634e-06,
      "loss": 0.4387,
      "step": 18083
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5499660968780518,
      "learning_rate": 8.272322487602836e-06,
      "loss": 0.4173,
      "step": 18084
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.555873990058899,
      "learning_rate": 8.271319400338115e-06,
      "loss": 0.4266,
      "step": 18085
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5129961967468262,
      "learning_rate": 8.270316331002571e-06,
      "loss": 0.3921,
      "step": 18086
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5695955753326416,
      "learning_rate": 8.269313279606614e-06,
      "loss": 0.4361,
      "step": 18087
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4572254419326782,
      "learning_rate": 8.268310246160645e-06,
      "loss": 0.3829,
      "step": 18088
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5489871501922607,
      "learning_rate": 8.267307230675071e-06,
      "loss": 0.517,
      "step": 18089
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5244585275650024,
      "learning_rate": 8.266304233160291e-06,
      "loss": 0.394,
      "step": 18090
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6100473403930664,
      "learning_rate": 8.265301253626706e-06,
      "loss": 0.5185,
      "step": 18091
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.505976676940918,
      "learning_rate": 8.264298292084722e-06,
      "loss": 0.4112,
      "step": 18092
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6898105144500732,
      "learning_rate": 8.263295348544743e-06,
      "loss": 0.5367,
      "step": 18093
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4769257307052612,
      "learning_rate": 8.262292423017165e-06,
      "loss": 0.3797,
      "step": 18094
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.509426236152649,
      "learning_rate": 8.261289515512393e-06,
      "loss": 0.4653,
      "step": 18095
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.532691478729248,
      "learning_rate": 8.260286626040828e-06,
      "loss": 0.4161,
      "step": 18096
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.048102378845215,
      "learning_rate": 8.259283754612878e-06,
      "loss": 0.4822,
      "step": 18097
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5097846984863281,
      "learning_rate": 8.258280901238935e-06,
      "loss": 0.3809,
      "step": 18098
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5924835205078125,
      "learning_rate": 8.257278065929404e-06,
      "loss": 0.4156,
      "step": 18099
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5152661800384521,
      "learning_rate": 8.256275248694686e-06,
      "loss": 0.4053,
      "step": 18100
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7322691679000854,
      "learning_rate": 8.255272449545187e-06,
      "loss": 0.4389,
      "step": 18101
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4641700983047485,
      "learning_rate": 8.254269668491297e-06,
      "loss": 0.3967,
      "step": 18102
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5192121267318726,
      "learning_rate": 8.253266905543425e-06,
      "loss": 0.4137,
      "step": 18103
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5034483671188354,
      "learning_rate": 8.252264160711964e-06,
      "loss": 0.4025,
      "step": 18104
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4878367185592651,
      "learning_rate": 8.251261434007326e-06,
      "loss": 0.4435,
      "step": 18105
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5190829038619995,
      "learning_rate": 8.250258725439899e-06,
      "loss": 0.4157,
      "step": 18106
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5934735536575317,
      "learning_rate": 8.249256035020086e-06,
      "loss": 0.4548,
      "step": 18107
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5028825998306274,
      "learning_rate": 8.248253362758289e-06,
      "loss": 0.4145,
      "step": 18108
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7072417736053467,
      "learning_rate": 8.247250708664907e-06,
      "loss": 0.4449,
      "step": 18109
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5667457580566406,
      "learning_rate": 8.246248072750335e-06,
      "loss": 0.4078,
      "step": 18110
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5341930389404297,
      "learning_rate": 8.245245455024976e-06,
      "loss": 0.4445,
      "step": 18111
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.572497844696045,
      "learning_rate": 8.244242855499226e-06,
      "loss": 0.4396,
      "step": 18112
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5390902757644653,
      "learning_rate": 8.243240274183493e-06,
      "loss": 0.435,
      "step": 18113
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.605149745941162,
      "learning_rate": 8.242237711088159e-06,
      "loss": 0.4127,
      "step": 18114
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5112059116363525,
      "learning_rate": 8.241235166223632e-06,
      "loss": 0.4581,
      "step": 18115
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5147651433944702,
      "learning_rate": 8.24023263960031e-06,
      "loss": 0.4054,
      "step": 18116
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5632950067520142,
      "learning_rate": 8.239230131228591e-06,
      "loss": 0.4672,
      "step": 18117
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4879988431930542,
      "learning_rate": 8.238227641118871e-06,
      "loss": 0.3889,
      "step": 18118
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5776240825653076,
      "learning_rate": 8.237225169281543e-06,
      "loss": 0.4505,
      "step": 18119
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5463249683380127,
      "learning_rate": 8.236222715727014e-06,
      "loss": 0.4251,
      "step": 18120
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6127675771713257,
      "learning_rate": 8.235220280465677e-06,
      "loss": 0.3895,
      "step": 18121
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7244648933410645,
      "learning_rate": 8.234217863507922e-06,
      "loss": 0.422,
      "step": 18122
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5495667457580566,
      "learning_rate": 8.233215464864155e-06,
      "loss": 0.463,
      "step": 18123
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6741243600845337,
      "learning_rate": 8.232213084544765e-06,
      "loss": 0.3931,
      "step": 18124
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6628069877624512,
      "learning_rate": 8.231210722560159e-06,
      "loss": 0.4886,
      "step": 18125
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5179576873779297,
      "learning_rate": 8.230208378920722e-06,
      "loss": 0.3853,
      "step": 18126
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5347132682800293,
      "learning_rate": 8.229206053636854e-06,
      "loss": 0.4627,
      "step": 18127
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.564785361289978,
      "learning_rate": 8.22820374671895e-06,
      "loss": 0.4236,
      "step": 18128
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5740488767623901,
      "learning_rate": 8.22720145817741e-06,
      "loss": 0.4391,
      "step": 18129
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7013405561447144,
      "learning_rate": 8.226199188022624e-06,
      "loss": 0.4132,
      "step": 18130
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.630002737045288,
      "learning_rate": 8.225196936264988e-06,
      "loss": 0.5291,
      "step": 18131
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.571201205253601,
      "learning_rate": 8.224194702914896e-06,
      "loss": 0.4152,
      "step": 18132
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.603454351425171,
      "learning_rate": 8.22319248798275e-06,
      "loss": 0.4931,
      "step": 18133
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5497561693191528,
      "learning_rate": 8.222190291478935e-06,
      "loss": 0.4339,
      "step": 18134
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5579862594604492,
      "learning_rate": 8.22118811341385e-06,
      "loss": 0.4736,
      "step": 18135
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5415229797363281,
      "learning_rate": 8.220185953797888e-06,
      "loss": 0.4262,
      "step": 18136
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7767242193222046,
      "learning_rate": 8.219183812641447e-06,
      "loss": 0.489,
      "step": 18137
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5717440843582153,
      "learning_rate": 8.218181689954916e-06,
      "loss": 0.4111,
      "step": 18138
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4879682064056396,
      "learning_rate": 8.21717958574869e-06,
      "loss": 0.4271,
      "step": 18139
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5273163318634033,
      "learning_rate": 8.216177500033159e-06,
      "loss": 0.4292,
      "step": 18140
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.556941270828247,
      "learning_rate": 8.215175432818725e-06,
      "loss": 0.4289,
      "step": 18141
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5231117010116577,
      "learning_rate": 8.214173384115775e-06,
      "loss": 0.4279,
      "step": 18142
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.537132740020752,
      "learning_rate": 8.2131713539347e-06,
      "loss": 0.424,
      "step": 18143
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5494625568389893,
      "learning_rate": 8.212169342285896e-06,
      "loss": 0.4066,
      "step": 18144
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5626178979873657,
      "learning_rate": 8.211167349179758e-06,
      "loss": 0.4747,
      "step": 18145
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6611443758010864,
      "learning_rate": 8.210165374626671e-06,
      "loss": 0.4154,
      "step": 18146
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5769402980804443,
      "learning_rate": 8.209163418637033e-06,
      "loss": 0.4737,
      "step": 18147
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5657145977020264,
      "learning_rate": 8.208161481221232e-06,
      "loss": 0.4165,
      "step": 18148
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.608207106590271,
      "learning_rate": 8.207159562389668e-06,
      "loss": 0.5223,
      "step": 18149
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.451328158378601,
      "learning_rate": 8.206157662152721e-06,
      "loss": 0.4048,
      "step": 18150
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5789412260055542,
      "learning_rate": 8.205155780520787e-06,
      "loss": 0.508,
      "step": 18151
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.534253716468811,
      "learning_rate": 8.204153917504259e-06,
      "loss": 0.3986,
      "step": 18152
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6714308261871338,
      "learning_rate": 8.20315207311353e-06,
      "loss": 0.512,
      "step": 18153
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5645737648010254,
      "learning_rate": 8.202150247358985e-06,
      "loss": 0.3889,
      "step": 18154
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6004291772842407,
      "learning_rate": 8.201148440251014e-06,
      "loss": 0.4306,
      "step": 18155
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5431345701217651,
      "learning_rate": 8.200146651800012e-06,
      "loss": 0.4377,
      "step": 18156
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.571168303489685,
      "learning_rate": 8.19914488201637e-06,
      "loss": 0.477,
      "step": 18157
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5610703229904175,
      "learning_rate": 8.19814313091047e-06,
      "loss": 0.4299,
      "step": 18158
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6214287281036377,
      "learning_rate": 8.19714139849271e-06,
      "loss": 0.4676,
      "step": 18159
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5217169523239136,
      "learning_rate": 8.196139684773474e-06,
      "loss": 0.3819,
      "step": 18160
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5410640239715576,
      "learning_rate": 8.19513798976316e-06,
      "loss": 0.408,
      "step": 18161
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4346563816070557,
      "learning_rate": 8.194136313472147e-06,
      "loss": 0.4004,
      "step": 18162
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6871453523635864,
      "learning_rate": 8.193134655910828e-06,
      "loss": 0.425,
      "step": 18163
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5103263854980469,
      "learning_rate": 8.192133017089592e-06,
      "loss": 0.4176,
      "step": 18164
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6330809593200684,
      "learning_rate": 8.19113139701883e-06,
      "loss": 0.5042,
      "step": 18165
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4771676063537598,
      "learning_rate": 8.190129795708925e-06,
      "loss": 0.4023,
      "step": 18166
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5964206457138062,
      "learning_rate": 8.18912821317027e-06,
      "loss": 0.4462,
      "step": 18167
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5798254013061523,
      "learning_rate": 8.188126649413249e-06,
      "loss": 0.4044,
      "step": 18168
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7286853790283203,
      "learning_rate": 8.187125104448254e-06,
      "loss": 0.4828,
      "step": 18169
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.50408935546875,
      "learning_rate": 8.18612357828567e-06,
      "loss": 0.4079,
      "step": 18170
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4879043102264404,
      "learning_rate": 8.185122070935885e-06,
      "loss": 0.4439,
      "step": 18171
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5399150848388672,
      "learning_rate": 8.184120582409285e-06,
      "loss": 0.4014,
      "step": 18172
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.53092622756958,
      "learning_rate": 8.183119112716262e-06,
      "loss": 0.4446,
      "step": 18173
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5253459215164185,
      "learning_rate": 8.182117661867195e-06,
      "loss": 0.4059,
      "step": 18174
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4754729270935059,
      "learning_rate": 8.181116229872476e-06,
      "loss": 0.4397,
      "step": 18175
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6978684663772583,
      "learning_rate": 8.18011481674249e-06,
      "loss": 0.4149,
      "step": 18176
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7378851175308228,
      "learning_rate": 8.179113422487627e-06,
      "loss": 1.0386,
      "step": 18177
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6118178367614746,
      "learning_rate": 8.178112047118268e-06,
      "loss": 1.0338,
      "step": 18178
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4429848194122314,
      "learning_rate": 8.177110690644796e-06,
      "loss": 0.4145,
      "step": 18179
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5431456565856934,
      "learning_rate": 8.176109353077604e-06,
      "loss": 0.4318,
      "step": 18180
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6000622510910034,
      "learning_rate": 8.175108034427079e-06,
      "loss": 0.4442,
      "step": 18181
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.846828579902649,
      "learning_rate": 8.174106734703595e-06,
      "loss": 0.4158,
      "step": 18182
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6988400220870972,
      "learning_rate": 8.173105453917547e-06,
      "loss": 0.9538,
      "step": 18183
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4138609170913696,
      "learning_rate": 8.172104192079314e-06,
      "loss": 1.0009,
      "step": 18184
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5385899543762207,
      "learning_rate": 8.171102949199289e-06,
      "loss": 0.437,
      "step": 18185
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.587458848953247,
      "learning_rate": 8.170101725287847e-06,
      "loss": 0.434,
      "step": 18186
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6638078689575195,
      "learning_rate": 8.169100520355375e-06,
      "loss": 0.4662,
      "step": 18187
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.498693585395813,
      "learning_rate": 8.168099334412261e-06,
      "loss": 0.3907,
      "step": 18188
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6217637062072754,
      "learning_rate": 8.16709816746889e-06,
      "loss": 0.4578,
      "step": 18189
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5915236473083496,
      "learning_rate": 8.166097019535636e-06,
      "loss": 0.4031,
      "step": 18190
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6102302074432373,
      "learning_rate": 8.16509589062289e-06,
      "loss": 0.4707,
      "step": 18191
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5080515146255493,
      "learning_rate": 8.16409478074103e-06,
      "loss": 0.428,
      "step": 18192
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5610883235931396,
      "learning_rate": 8.163093689900452e-06,
      "loss": 0.4591,
      "step": 18193
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5327017307281494,
      "learning_rate": 8.162092618111523e-06,
      "loss": 0.4223,
      "step": 18194
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5783134698867798,
      "learning_rate": 8.161091565384634e-06,
      "loss": 0.4419,
      "step": 18195
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5501984357833862,
      "learning_rate": 8.160090531730165e-06,
      "loss": 0.425,
      "step": 18196
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6094779968261719,
      "learning_rate": 8.159089517158503e-06,
      "loss": 0.4615,
      "step": 18197
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5167150497436523,
      "learning_rate": 8.158088521680024e-06,
      "loss": 0.3877,
      "step": 18198
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6028176546096802,
      "learning_rate": 8.157087545305112e-06,
      "loss": 0.5281,
      "step": 18199
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5438532829284668,
      "learning_rate": 8.156086588044148e-06,
      "loss": 0.4144,
      "step": 18200
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6238555908203125,
      "learning_rate": 8.15508564990752e-06,
      "loss": 0.4584,
      "step": 18201
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5670530796051025,
      "learning_rate": 8.154084730905598e-06,
      "loss": 0.3932,
      "step": 18202
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5382858514785767,
      "learning_rate": 8.153083831048772e-06,
      "loss": 0.4428,
      "step": 18203
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4779245853424072,
      "learning_rate": 8.152082950347416e-06,
      "loss": 0.395,
      "step": 18204
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6175100803375244,
      "learning_rate": 8.151082088811921e-06,
      "loss": 0.491,
      "step": 18205
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5427706241607666,
      "learning_rate": 8.150081246452659e-06,
      "loss": 0.4099,
      "step": 18206
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5559008121490479,
      "learning_rate": 8.149080423280009e-06,
      "loss": 0.4472,
      "step": 18207
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4859575033187866,
      "learning_rate": 8.148079619304358e-06,
      "loss": 0.4023,
      "step": 18208
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.549737572669983,
      "learning_rate": 8.147078834536086e-06,
      "loss": 0.4771,
      "step": 18209
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5265289545059204,
      "learning_rate": 8.146078068985563e-06,
      "loss": 0.399,
      "step": 18210
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5893454551696777,
      "learning_rate": 8.14507732266318e-06,
      "loss": 0.4822,
      "step": 18211
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.616162896156311,
      "learning_rate": 8.144076595579307e-06,
      "loss": 0.4426,
      "step": 18212
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8631205558776855,
      "learning_rate": 8.143075887744332e-06,
      "loss": 0.577,
      "step": 18213
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.482518196105957,
      "learning_rate": 8.142075199168626e-06,
      "loss": 0.3945,
      "step": 18214
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5481853485107422,
      "learning_rate": 8.141074529862571e-06,
      "loss": 0.459,
      "step": 18215
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6104530096054077,
      "learning_rate": 8.140073879836548e-06,
      "loss": 0.437,
      "step": 18216
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6787337064743042,
      "learning_rate": 8.139073249100934e-06,
      "loss": 0.5237,
      "step": 18217
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5265017747879028,
      "learning_rate": 8.138072637666102e-06,
      "loss": 0.43,
      "step": 18218
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.846286654472351,
      "learning_rate": 8.137072045542436e-06,
      "loss": 0.9435,
      "step": 18219
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.533323049545288,
      "learning_rate": 8.136071472740312e-06,
      "loss": 1.0485,
      "step": 18220
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6418055295944214,
      "learning_rate": 8.13507091927011e-06,
      "loss": 0.4631,
      "step": 18221
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.644563913345337,
      "learning_rate": 8.134070385142202e-06,
      "loss": 0.4138,
      "step": 18222
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4981046915054321,
      "learning_rate": 8.133069870366967e-06,
      "loss": 0.4967,
      "step": 18223
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6913480758666992,
      "learning_rate": 8.132069374954784e-06,
      "loss": 0.409,
      "step": 18224
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.723348617553711,
      "learning_rate": 8.131068898916031e-06,
      "loss": 0.5199,
      "step": 18225
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.559874176979065,
      "learning_rate": 8.130068442261078e-06,
      "loss": 0.4049,
      "step": 18226
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5676133632659912,
      "learning_rate": 8.129068005000307e-06,
      "loss": 0.4825,
      "step": 18227
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4531326293945312,
      "learning_rate": 8.12806758714409e-06,
      "loss": 0.3862,
      "step": 18228
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5520367622375488,
      "learning_rate": 8.127067188702813e-06,
      "loss": 0.4986,
      "step": 18229
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4792873859405518,
      "learning_rate": 8.12606680968684e-06,
      "loss": 0.4007,
      "step": 18230
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.489526629447937,
      "learning_rate": 8.125066450106549e-06,
      "loss": 0.4584,
      "step": 18231
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4960813522338867,
      "learning_rate": 8.124066109972316e-06,
      "loss": 0.383,
      "step": 18232
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5945056676864624,
      "learning_rate": 8.123065789294522e-06,
      "loss": 0.4924,
      "step": 18233
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5517140626907349,
      "learning_rate": 8.122065488083534e-06,
      "loss": 0.4051,
      "step": 18234
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.807675838470459,
      "learning_rate": 8.12106520634973e-06,
      "loss": 0.4498,
      "step": 18235
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6284387111663818,
      "learning_rate": 8.120064944103486e-06,
      "loss": 0.3966,
      "step": 18236
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5346256494522095,
      "learning_rate": 8.119064701355174e-06,
      "loss": 0.4736,
      "step": 18237
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5256030559539795,
      "learning_rate": 8.118064478115168e-06,
      "loss": 0.4111,
      "step": 18238
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4768571853637695,
      "learning_rate": 8.117064274393842e-06,
      "loss": 0.419,
      "step": 18239
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5953342914581299,
      "learning_rate": 8.11606409020157e-06,
      "loss": 0.4422,
      "step": 18240
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5926138162612915,
      "learning_rate": 8.115063925548729e-06,
      "loss": 0.4592,
      "step": 18241
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5974313020706177,
      "learning_rate": 8.11406378044569e-06,
      "loss": 0.3963,
      "step": 18242
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5688508749008179,
      "learning_rate": 8.11306365490282e-06,
      "loss": 0.475,
      "step": 18243
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.555828332901001,
      "learning_rate": 8.112063548930501e-06,
      "loss": 0.4157,
      "step": 18244
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4910160303115845,
      "learning_rate": 8.111063462539103e-06,
      "loss": 0.4455,
      "step": 18245
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.52703058719635,
      "learning_rate": 8.110063395738995e-06,
      "loss": 0.4049,
      "step": 18246
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.650899887084961,
      "learning_rate": 8.109063348540553e-06,
      "loss": 0.4816,
      "step": 18247
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4918314218521118,
      "learning_rate": 8.108063320954145e-06,
      "loss": 0.3998,
      "step": 18248
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5748393535614014,
      "learning_rate": 8.10706331299015e-06,
      "loss": 0.4293,
      "step": 18249
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5305219888687134,
      "learning_rate": 8.106063324658934e-06,
      "loss": 0.4048,
      "step": 18250
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5370604991912842,
      "learning_rate": 8.105063355970868e-06,
      "loss": 0.4582,
      "step": 18251
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9171884059906006,
      "learning_rate": 8.104063406936327e-06,
      "loss": 0.4015,
      "step": 18252
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.612899661064148,
      "learning_rate": 8.10306347756568e-06,
      "loss": 0.4644,
      "step": 18253
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5020561218261719,
      "learning_rate": 8.102063567869298e-06,
      "loss": 0.4016,
      "step": 18254
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9803828001022339,
      "learning_rate": 8.101063677857552e-06,
      "loss": 0.9708,
      "step": 18255
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.533260464668274,
      "learning_rate": 8.10006380754081e-06,
      "loss": 0.9655,
      "step": 18256
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6398215293884277,
      "learning_rate": 8.099063956929449e-06,
      "loss": 0.4565,
      "step": 18257
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.560638427734375,
      "learning_rate": 8.098064126033831e-06,
      "loss": 0.4099,
      "step": 18258
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5951822996139526,
      "learning_rate": 8.097064314864328e-06,
      "loss": 0.4514,
      "step": 18259
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5014407634735107,
      "learning_rate": 8.096064523431315e-06,
      "loss": 0.3957,
      "step": 18260
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5680978298187256,
      "learning_rate": 8.095064751745156e-06,
      "loss": 0.4301,
      "step": 18261
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5400636196136475,
      "learning_rate": 8.09406499981622e-06,
      "loss": 0.3914,
      "step": 18262
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5367064476013184,
      "learning_rate": 8.09306526765488e-06,
      "loss": 0.4521,
      "step": 18263
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5417189598083496,
      "learning_rate": 8.092065555271499e-06,
      "loss": 0.4085,
      "step": 18264
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5690709352493286,
      "learning_rate": 8.091065862676454e-06,
      "loss": 0.4317,
      "step": 18265
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4315634965896606,
      "learning_rate": 8.090066189880105e-06,
      "loss": 0.3911,
      "step": 18266
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5450465679168701,
      "learning_rate": 8.089066536892825e-06,
      "loss": 0.4417,
      "step": 18267
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.631066083908081,
      "learning_rate": 8.088066903724981e-06,
      "loss": 0.3863,
      "step": 18268
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5254477262496948,
      "learning_rate": 8.087067290386942e-06,
      "loss": 0.4314,
      "step": 18269
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6590685844421387,
      "learning_rate": 8.086067696889073e-06,
      "loss": 0.4265,
      "step": 18270
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5370744466781616,
      "learning_rate": 8.08506812324174e-06,
      "loss": 0.4303,
      "step": 18271
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5278639793395996,
      "learning_rate": 8.084068569455314e-06,
      "loss": 0.4489,
      "step": 18272
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4919116497039795,
      "learning_rate": 8.083069035540165e-06,
      "loss": 0.4321,
      "step": 18273
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.519909143447876,
      "learning_rate": 8.08206952150665e-06,
      "loss": 0.4514,
      "step": 18274
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4918121099472046,
      "learning_rate": 8.081070027365142e-06,
      "loss": 0.4516,
      "step": 18275
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5886679887771606,
      "learning_rate": 8.080070553126005e-06,
      "loss": 0.4244,
      "step": 18276
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.716538906097412,
      "learning_rate": 8.079071098799611e-06,
      "loss": 0.4955,
      "step": 18277
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4492032527923584,
      "learning_rate": 8.078071664396319e-06,
      "loss": 0.3847,
      "step": 18278
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6294152736663818,
      "learning_rate": 8.077072249926495e-06,
      "loss": 0.4793,
      "step": 18279
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.627095103263855,
      "learning_rate": 8.07607285540051e-06,
      "loss": 0.435,
      "step": 18280
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5733239650726318,
      "learning_rate": 8.075073480828726e-06,
      "loss": 0.47,
      "step": 18281
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4776092767715454,
      "learning_rate": 8.074074126221506e-06,
      "loss": 0.385,
      "step": 18282
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4737459421157837,
      "learning_rate": 8.073074791589218e-06,
      "loss": 0.4549,
      "step": 18283
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5122510194778442,
      "learning_rate": 8.072075476942223e-06,
      "loss": 0.4121,
      "step": 18284
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.552567958831787,
      "learning_rate": 8.071076182290893e-06,
      "loss": 0.4563,
      "step": 18285
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.499032974243164,
      "learning_rate": 8.070076907645585e-06,
      "loss": 0.4044,
      "step": 18286
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5771727561950684,
      "learning_rate": 8.069077653016665e-06,
      "loss": 0.4515,
      "step": 18287
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4869837760925293,
      "learning_rate": 8.068078418414499e-06,
      "loss": 0.4036,
      "step": 18288
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7363176345825195,
      "learning_rate": 8.06707920384945e-06,
      "loss": 0.4403,
      "step": 18289
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4566981792449951,
      "learning_rate": 8.066080009331878e-06,
      "loss": 0.3926,
      "step": 18290
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4915508031845093,
      "learning_rate": 8.06508083487215e-06,
      "loss": 0.4228,
      "step": 18291
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4846919775009155,
      "learning_rate": 8.064081680480626e-06,
      "loss": 0.4125,
      "step": 18292
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5274076461791992,
      "learning_rate": 8.063082546167674e-06,
      "loss": 0.426,
      "step": 18293
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4632962942123413,
      "learning_rate": 8.06208343194365e-06,
      "loss": 0.4153,
      "step": 18294
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5452489852905273,
      "learning_rate": 8.06108433781892e-06,
      "loss": 0.4288,
      "step": 18295
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5341875553131104,
      "learning_rate": 8.06008526380385e-06,
      "loss": 0.4242,
      "step": 18296
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6243805885314941,
      "learning_rate": 8.059086209908797e-06,
      "loss": 0.5048,
      "step": 18297
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5312318801879883,
      "learning_rate": 8.058087176144118e-06,
      "loss": 0.4003,
      "step": 18298
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5532495975494385,
      "learning_rate": 8.057088162520186e-06,
      "loss": 0.4339,
      "step": 18299
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4564379453659058,
      "learning_rate": 8.056089169047352e-06,
      "loss": 0.3852,
      "step": 18300
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5971578359603882,
      "learning_rate": 8.055090195735986e-06,
      "loss": 0.4279,
      "step": 18301
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5272177457809448,
      "learning_rate": 8.054091242596445e-06,
      "loss": 0.4086,
      "step": 18302
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.619186282157898,
      "learning_rate": 8.053092309639085e-06,
      "loss": 0.4524,
      "step": 18303
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5177422761917114,
      "learning_rate": 8.052093396874275e-06,
      "loss": 0.4029,
      "step": 18304
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.231008291244507,
      "learning_rate": 8.051094504312372e-06,
      "loss": 0.4105,
      "step": 18305
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5680105686187744,
      "learning_rate": 8.050095631963732e-06,
      "loss": 0.4219,
      "step": 18306
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5412025451660156,
      "learning_rate": 8.04909677983872e-06,
      "loss": 0.4546,
      "step": 18307
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5673096179962158,
      "learning_rate": 8.04809794794769e-06,
      "loss": 0.4185,
      "step": 18308
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.927677035331726,
      "learning_rate": 8.04709913630101e-06,
      "loss": 0.4553,
      "step": 18309
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.515837550163269,
      "learning_rate": 8.046100344909037e-06,
      "loss": 0.4207,
      "step": 18310
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.518584966659546,
      "learning_rate": 8.045101573782124e-06,
      "loss": 0.4569,
      "step": 18311
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.536662220954895,
      "learning_rate": 8.044102822930633e-06,
      "loss": 0.3873,
      "step": 18312
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6550604104995728,
      "learning_rate": 8.043104092364927e-06,
      "loss": 0.4283,
      "step": 18313
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5340365171432495,
      "learning_rate": 8.04210538209536e-06,
      "loss": 0.4026,
      "step": 18314
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6324732303619385,
      "learning_rate": 8.04110669213229e-06,
      "loss": 0.4943,
      "step": 18315
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4731693267822266,
      "learning_rate": 8.040108022486075e-06,
      "loss": 0.3864,
      "step": 18316
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6574068069458008,
      "learning_rate": 8.039109373167074e-06,
      "loss": 0.4985,
      "step": 18317
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.481994390487671,
      "learning_rate": 8.038110744185647e-06,
      "loss": 0.4032,
      "step": 18318
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6348237991333008,
      "learning_rate": 8.037112135552147e-06,
      "loss": 0.4878,
      "step": 18319
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5132688283920288,
      "learning_rate": 8.036113547276932e-06,
      "loss": 0.3907,
      "step": 18320
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.655178189277649,
      "learning_rate": 8.03511497937036e-06,
      "loss": 0.5047,
      "step": 18321
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.467248558998108,
      "learning_rate": 8.034116431842791e-06,
      "loss": 0.3724,
      "step": 18322
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5950865745544434,
      "learning_rate": 8.033117904704574e-06,
      "loss": 0.4656,
      "step": 18323
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5171911716461182,
      "learning_rate": 8.032119397966071e-06,
      "loss": 0.428,
      "step": 18324
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5864795446395874,
      "learning_rate": 8.031120911637636e-06,
      "loss": 0.4467,
      "step": 18325
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5972671508789062,
      "learning_rate": 8.03012244572963e-06,
      "loss": 0.4219,
      "step": 18326
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5993421077728271,
      "learning_rate": 8.0291240002524e-06,
      "loss": 0.4761,
      "step": 18327
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5048706531524658,
      "learning_rate": 8.028125575216304e-06,
      "loss": 0.4111,
      "step": 18328
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5534902811050415,
      "learning_rate": 8.027127170631701e-06,
      "loss": 0.4573,
      "step": 18329
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.494134783744812,
      "learning_rate": 8.026128786508948e-06,
      "loss": 0.3888,
      "step": 18330
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5176403522491455,
      "learning_rate": 8.02513042285839e-06,
      "loss": 0.4393,
      "step": 18331
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5233572721481323,
      "learning_rate": 8.02413207969039e-06,
      "loss": 0.4065,
      "step": 18332
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5131937265396118,
      "learning_rate": 8.023133757015298e-06,
      "loss": 0.4171,
      "step": 18333
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4929184913635254,
      "learning_rate": 8.022135454843474e-06,
      "loss": 0.4092,
      "step": 18334
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5805981159210205,
      "learning_rate": 8.021137173185266e-06,
      "loss": 0.4531,
      "step": 18335
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4366346597671509,
      "learning_rate": 8.020138912051027e-06,
      "loss": 0.3984,
      "step": 18336
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5491520166397095,
      "learning_rate": 8.019140671451119e-06,
      "loss": 0.4537,
      "step": 18337
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4553197622299194,
      "learning_rate": 8.018142451395889e-06,
      "loss": 0.3962,
      "step": 18338
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8836356401443481,
      "learning_rate": 8.017144251895686e-06,
      "loss": 0.9733,
      "step": 18339
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6149592399597168,
      "learning_rate": 8.01614607296087e-06,
      "loss": 1.0183,
      "step": 18340
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.607072114944458,
      "learning_rate": 8.015147914601791e-06,
      "loss": 0.4559,
      "step": 18341
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5966010093688965,
      "learning_rate": 8.014149776828808e-06,
      "loss": 0.4,
      "step": 18342
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6403250694274902,
      "learning_rate": 8.013151659652262e-06,
      "loss": 0.5372,
      "step": 18343
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5135926008224487,
      "learning_rate": 8.012153563082508e-06,
      "loss": 0.3945,
      "step": 18344
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.615449070930481,
      "learning_rate": 8.011155487129904e-06,
      "loss": 0.4539,
      "step": 18345
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5270556211471558,
      "learning_rate": 8.0101574318048e-06,
      "loss": 0.3926,
      "step": 18346
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6957054138183594,
      "learning_rate": 8.009159397117542e-06,
      "loss": 0.4266,
      "step": 18347
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5079576969146729,
      "learning_rate": 8.008161383078486e-06,
      "loss": 0.4188,
      "step": 18348
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6600053310394287,
      "learning_rate": 8.00716338969798e-06,
      "loss": 0.4803,
      "step": 18349
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.547598958015442,
      "learning_rate": 8.00616541698638e-06,
      "loss": 0.3793,
      "step": 18350
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.599169373512268,
      "learning_rate": 8.005167464954027e-06,
      "loss": 0.4617,
      "step": 18351
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6705288887023926,
      "learning_rate": 8.00416953361128e-06,
      "loss": 0.4049,
      "step": 18352
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4413124322891235,
      "learning_rate": 8.003171622968486e-06,
      "loss": 0.4284,
      "step": 18353
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5947128534317017,
      "learning_rate": 8.002173733036e-06,
      "loss": 0.4023,
      "step": 18354
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.664416790008545,
      "learning_rate": 8.001175863824163e-06,
      "loss": 0.4442,
      "step": 18355
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5146856307983398,
      "learning_rate": 8.000178015343327e-06,
      "loss": 0.3749,
      "step": 18356
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5758203268051147,
      "learning_rate": 7.999180187603845e-06,
      "loss": 0.4454,
      "step": 18357
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5440844297409058,
      "learning_rate": 7.998182380616066e-06,
      "loss": 0.411,
      "step": 18358
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5304924249649048,
      "learning_rate": 7.997184594390332e-06,
      "loss": 0.4397,
      "step": 18359
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5812225341796875,
      "learning_rate": 7.996186828937e-06,
      "loss": 0.3908,
      "step": 18360
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.534888505935669,
      "learning_rate": 7.995189084266413e-06,
      "loss": 0.4323,
      "step": 18361
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5924458503723145,
      "learning_rate": 7.994191360388923e-06,
      "loss": 0.4276,
      "step": 18362
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6007925271987915,
      "learning_rate": 7.993193657314874e-06,
      "loss": 0.4576,
      "step": 18363
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.510547161102295,
      "learning_rate": 7.992195975054616e-06,
      "loss": 0.4011,
      "step": 18364
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6423259973526,
      "learning_rate": 7.991198313618497e-06,
      "loss": 0.5175,
      "step": 18365
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8687403202056885,
      "learning_rate": 7.990200673016867e-06,
      "loss": 0.435,
      "step": 18366
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5800552368164062,
      "learning_rate": 7.989203053260064e-06,
      "loss": 0.4434,
      "step": 18367
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.591313362121582,
      "learning_rate": 7.988205454358444e-06,
      "loss": 0.4148,
      "step": 18368
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.548796534538269,
      "learning_rate": 7.987207876322348e-06,
      "loss": 0.4725,
      "step": 18369
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4875956773757935,
      "learning_rate": 7.986210319162129e-06,
      "loss": 0.3988,
      "step": 18370
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5913251638412476,
      "learning_rate": 7.985212782888127e-06,
      "loss": 0.4614,
      "step": 18371
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6083794832229614,
      "learning_rate": 7.984215267510689e-06,
      "loss": 0.3973,
      "step": 18372
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7070984840393066,
      "learning_rate": 7.983217773040164e-06,
      "loss": 0.4322,
      "step": 18373
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4760726690292358,
      "learning_rate": 7.982220299486897e-06,
      "loss": 0.3817,
      "step": 18374
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7551406621932983,
      "learning_rate": 7.981222846861228e-06,
      "loss": 1.0086,
      "step": 18375
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.529725432395935,
      "learning_rate": 7.980225415173508e-06,
      "loss": 1.0441,
      "step": 18376
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.597732663154602,
      "learning_rate": 7.979228004434078e-06,
      "loss": 0.4525,
      "step": 18377
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6779346466064453,
      "learning_rate": 7.97823061465329e-06,
      "loss": 0.4491,
      "step": 18378
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6050313711166382,
      "learning_rate": 7.977233245841483e-06,
      "loss": 0.4611,
      "step": 18379
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.50190007686615,
      "learning_rate": 7.976235898008997e-06,
      "loss": 0.3863,
      "step": 18380
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4858213663101196,
      "learning_rate": 7.975238571166183e-06,
      "loss": 0.4443,
      "step": 18381
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6716820001602173,
      "learning_rate": 7.974241265323389e-06,
      "loss": 0.4126,
      "step": 18382
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.636031150817871,
      "learning_rate": 7.973243980490943e-06,
      "loss": 0.4224,
      "step": 18383
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5583003759384155,
      "learning_rate": 7.972246716679202e-06,
      "loss": 0.3873,
      "step": 18384
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7124512195587158,
      "learning_rate": 7.971249473898503e-06,
      "loss": 0.4432,
      "step": 18385
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6232703924179077,
      "learning_rate": 7.970252252159199e-06,
      "loss": 0.481,
      "step": 18386
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6551616191864014,
      "learning_rate": 7.969255051471616e-06,
      "loss": 0.4799,
      "step": 18387
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5201376676559448,
      "learning_rate": 7.968257871846109e-06,
      "loss": 0.3893,
      "step": 18388
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7196974754333496,
      "learning_rate": 7.967260713293014e-06,
      "loss": 0.4525,
      "step": 18389
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6092002391815186,
      "learning_rate": 7.96626357582268e-06,
      "loss": 0.438,
      "step": 18390
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5943453311920166,
      "learning_rate": 7.965266459445445e-06,
      "loss": 0.4749,
      "step": 18391
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4596748352050781,
      "learning_rate": 7.964269364171646e-06,
      "loss": 0.3973,
      "step": 18392
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5368199348449707,
      "learning_rate": 7.96327229001163e-06,
      "loss": 0.4719,
      "step": 18393
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5134423971176147,
      "learning_rate": 7.962275236975743e-06,
      "loss": 0.3983,
      "step": 18394
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4450509548187256,
      "learning_rate": 7.961278205074313e-06,
      "loss": 0.4005,
      "step": 18395
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4594531059265137,
      "learning_rate": 7.960281194317692e-06,
      "loss": 0.3986,
      "step": 18396
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5375792980194092,
      "learning_rate": 7.959284204716213e-06,
      "loss": 0.4739,
      "step": 18397
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4976980686187744,
      "learning_rate": 7.958287236280226e-06,
      "loss": 0.4034,
      "step": 18398
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6846954822540283,
      "learning_rate": 7.957290289020061e-06,
      "loss": 0.466,
      "step": 18399
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4785884618759155,
      "learning_rate": 7.956293362946062e-06,
      "loss": 0.4123,
      "step": 18400
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.50922429561615,
      "learning_rate": 7.95529645806857e-06,
      "loss": 0.422,
      "step": 18401
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5490976572036743,
      "learning_rate": 7.954299574397926e-06,
      "loss": 0.431,
      "step": 18402
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7941185235977173,
      "learning_rate": 7.953302711944461e-06,
      "loss": 1.0164,
      "step": 18403
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4992488622665405,
      "learning_rate": 7.952305870718525e-06,
      "loss": 0.9717,
      "step": 18404
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6601619720458984,
      "learning_rate": 7.951309050730447e-06,
      "loss": 0.4941,
      "step": 18405
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5699366331100464,
      "learning_rate": 7.950312251990574e-06,
      "loss": 0.4059,
      "step": 18406
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.661551833152771,
      "learning_rate": 7.949315474509239e-06,
      "loss": 0.4734,
      "step": 18407
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6856452226638794,
      "learning_rate": 7.948318718296781e-06,
      "loss": 0.4191,
      "step": 18408
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5649105310440063,
      "learning_rate": 7.947321983363539e-06,
      "loss": 0.4362,
      "step": 18409
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4945212602615356,
      "learning_rate": 7.946325269719853e-06,
      "loss": 0.3841,
      "step": 18410
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8946897983551025,
      "learning_rate": 7.945328577376054e-06,
      "loss": 0.9934,
      "step": 18411
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5004984140396118,
      "learning_rate": 7.944331906342487e-06,
      "loss": 1.0416,
      "step": 18412
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6146169900894165,
      "learning_rate": 7.94333525662948e-06,
      "loss": 0.4749,
      "step": 18413
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6232795715332031,
      "learning_rate": 7.942338628247382e-06,
      "loss": 0.4015,
      "step": 18414
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.647624135017395,
      "learning_rate": 7.94134202120652e-06,
      "loss": 0.4454,
      "step": 18415
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.548676609992981,
      "learning_rate": 7.940345435517229e-06,
      "loss": 0.4245,
      "step": 18416
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.495153784751892,
      "learning_rate": 7.939348871189854e-06,
      "loss": 0.4898,
      "step": 18417
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6094720363616943,
      "learning_rate": 7.938352328234727e-06,
      "loss": 0.4166,
      "step": 18418
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6641652584075928,
      "learning_rate": 7.937355806662181e-06,
      "loss": 0.4701,
      "step": 18419
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5413150787353516,
      "learning_rate": 7.936359306482553e-06,
      "loss": 0.3867,
      "step": 18420
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5352741479873657,
      "learning_rate": 7.93536282770618e-06,
      "loss": 0.4752,
      "step": 18421
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5458300113677979,
      "learning_rate": 7.9343663703434e-06,
      "loss": 0.444,
      "step": 18422
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5560071468353271,
      "learning_rate": 7.933369934404538e-06,
      "loss": 0.4362,
      "step": 18423
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4404327869415283,
      "learning_rate": 7.932373519899936e-06,
      "loss": 0.368,
      "step": 18424
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6715762615203857,
      "learning_rate": 7.931377126839925e-06,
      "loss": 0.5195,
      "step": 18425
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5292645692825317,
      "learning_rate": 7.930380755234845e-06,
      "loss": 0.4249,
      "step": 18426
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5654957294464111,
      "learning_rate": 7.929384405095025e-06,
      "loss": 0.4458,
      "step": 18427
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.505722999572754,
      "learning_rate": 7.928388076430797e-06,
      "loss": 0.4175,
      "step": 18428
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.571689486503601,
      "learning_rate": 7.9273917692525e-06,
      "loss": 0.4533,
      "step": 18429
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5572482347488403,
      "learning_rate": 7.926395483570466e-06,
      "loss": 0.4116,
      "step": 18430
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5139464139938354,
      "learning_rate": 7.925399219395024e-06,
      "loss": 0.481,
      "step": 18431
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5062350034713745,
      "learning_rate": 7.924402976736511e-06,
      "loss": 0.4025,
      "step": 18432
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5638985633850098,
      "learning_rate": 7.923406755605255e-06,
      "loss": 0.4356,
      "step": 18433
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.488733172416687,
      "learning_rate": 7.922410556011596e-06,
      "loss": 0.3903,
      "step": 18434
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7652404308319092,
      "learning_rate": 7.92141437796586e-06,
      "loss": 0.5524,
      "step": 18435
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4932348728179932,
      "learning_rate": 7.92041822147838e-06,
      "loss": 0.3844,
      "step": 18436
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5456067323684692,
      "learning_rate": 7.919422086559489e-06,
      "loss": 0.4394,
      "step": 18437
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4867526292800903,
      "learning_rate": 7.91842597321952e-06,
      "loss": 0.4209,
      "step": 18438
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5907984972000122,
      "learning_rate": 7.917429881468798e-06,
      "loss": 0.5228,
      "step": 18439
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.545509696006775,
      "learning_rate": 7.91643381131766e-06,
      "loss": 0.3861,
      "step": 18440
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6493808031082153,
      "learning_rate": 7.915437762776435e-06,
      "loss": 0.4746,
      "step": 18441
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.555630087852478,
      "learning_rate": 7.914441735855457e-06,
      "loss": 0.3847,
      "step": 18442
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7109882831573486,
      "learning_rate": 7.91344573056505e-06,
      "loss": 0.487,
      "step": 18443
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5111312866210938,
      "learning_rate": 7.912449746915547e-06,
      "loss": 0.3797,
      "step": 18444
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5129611492156982,
      "learning_rate": 7.91145378491728e-06,
      "loss": 0.4333,
      "step": 18445
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.64786696434021,
      "learning_rate": 7.910457844580579e-06,
      "loss": 0.4293,
      "step": 18446
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.568552017211914,
      "learning_rate": 7.909461925915766e-06,
      "loss": 0.4665,
      "step": 18447
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5855704545974731,
      "learning_rate": 7.90846602893318e-06,
      "loss": 0.4039,
      "step": 18448
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5616027116775513,
      "learning_rate": 7.907470153643142e-06,
      "loss": 0.4595,
      "step": 18449
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4861524105072021,
      "learning_rate": 7.906474300055989e-06,
      "loss": 0.4167,
      "step": 18450
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6034578084945679,
      "learning_rate": 7.905478468182045e-06,
      "loss": 0.4349,
      "step": 18451
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.512120246887207,
      "learning_rate": 7.904482658031634e-06,
      "loss": 0.452,
      "step": 18452
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.473982572555542,
      "learning_rate": 7.903486869615093e-06,
      "loss": 0.4887,
      "step": 18453
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4877989292144775,
      "learning_rate": 7.902491102942746e-06,
      "loss": 0.3892,
      "step": 18454
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5145595073699951,
      "learning_rate": 7.901495358024916e-06,
      "loss": 0.4191,
      "step": 18455
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.459256649017334,
      "learning_rate": 7.900499634871938e-06,
      "loss": 0.3904,
      "step": 18456
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5001963376998901,
      "learning_rate": 7.899503933494135e-06,
      "loss": 0.4481,
      "step": 18457
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.515339732170105,
      "learning_rate": 7.898508253901837e-06,
      "loss": 0.3974,
      "step": 18458
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5894114971160889,
      "learning_rate": 7.897512596105367e-06,
      "loss": 0.4603,
      "step": 18459
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5437581539154053,
      "learning_rate": 7.896516960115053e-06,
      "loss": 0.3633,
      "step": 18460
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5333236455917358,
      "learning_rate": 7.895521345941222e-06,
      "loss": 0.4338,
      "step": 18461
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5713828802108765,
      "learning_rate": 7.894525753594204e-06,
      "loss": 0.4093,
      "step": 18462
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5202504396438599,
      "learning_rate": 7.893530183084317e-06,
      "loss": 0.4727,
      "step": 18463
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5435978174209595,
      "learning_rate": 7.89253463442189e-06,
      "loss": 0.41,
      "step": 18464
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7018190622329712,
      "learning_rate": 7.89153910761725e-06,
      "loss": 0.4797,
      "step": 18465
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5208874940872192,
      "learning_rate": 7.890543602680723e-06,
      "loss": 0.4257,
      "step": 18466
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.539482593536377,
      "learning_rate": 7.889548119622627e-06,
      "loss": 0.4656,
      "step": 18467
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6112143993377686,
      "learning_rate": 7.888552658453294e-06,
      "loss": 0.3987,
      "step": 18468
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7805774211883545,
      "learning_rate": 7.887557219183046e-06,
      "loss": 1.0247,
      "step": 18469
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5237120389938354,
      "learning_rate": 7.88656180182221e-06,
      "loss": 1.006,
      "step": 18470
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5855450630187988,
      "learning_rate": 7.885566406381107e-06,
      "loss": 0.4763,
      "step": 18471
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5853902101516724,
      "learning_rate": 7.884571032870059e-06,
      "loss": 0.3917,
      "step": 18472
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.562329649925232,
      "learning_rate": 7.883575681299394e-06,
      "loss": 0.4684,
      "step": 18473
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5906504392623901,
      "learning_rate": 7.882580351679435e-06,
      "loss": 0.4148,
      "step": 18474
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5973949432373047,
      "learning_rate": 7.8815850440205e-06,
      "loss": 0.4851,
      "step": 18475
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5260224342346191,
      "learning_rate": 7.880589758332918e-06,
      "loss": 0.4109,
      "step": 18476
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4651321172714233,
      "learning_rate": 7.879594494627006e-06,
      "loss": 0.4147,
      "step": 18477
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.556847333908081,
      "learning_rate": 7.878599252913094e-06,
      "loss": 0.4308,
      "step": 18478
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.561055064201355,
      "learning_rate": 7.877604033201501e-06,
      "loss": 0.4443,
      "step": 18479
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9654796123504639,
      "learning_rate": 7.876608835502542e-06,
      "loss": 0.3963,
      "step": 18480
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5954902172088623,
      "learning_rate": 7.87561365982655e-06,
      "loss": 0.4647,
      "step": 18481
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5204343795776367,
      "learning_rate": 7.874618506183842e-06,
      "loss": 0.4087,
      "step": 18482
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4653019905090332,
      "learning_rate": 7.873623374584735e-06,
      "loss": 0.4353,
      "step": 18483
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.636710286140442,
      "learning_rate": 7.872628265039557e-06,
      "loss": 0.4166,
      "step": 18484
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9220333099365234,
      "learning_rate": 7.87163317755862e-06,
      "loss": 0.4839,
      "step": 18485
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.508431077003479,
      "learning_rate": 7.870638112152257e-06,
      "loss": 0.4039,
      "step": 18486
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5638519525527954,
      "learning_rate": 7.869643068830781e-06,
      "loss": 0.4871,
      "step": 18487
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5506300926208496,
      "learning_rate": 7.86864804760451e-06,
      "loss": 0.4363,
      "step": 18488
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8854292631149292,
      "learning_rate": 7.867653048483768e-06,
      "loss": 0.4518,
      "step": 18489
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4866490364074707,
      "learning_rate": 7.866658071478876e-06,
      "loss": 0.4173,
      "step": 18490
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5641705989837646,
      "learning_rate": 7.865663116600149e-06,
      "loss": 0.4892,
      "step": 18491
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5487141609191895,
      "learning_rate": 7.864668183857907e-06,
      "loss": 0.4118,
      "step": 18492
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7100629806518555,
      "learning_rate": 7.863673273262471e-06,
      "loss": 0.9855,
      "step": 18493
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5773589611053467,
      "learning_rate": 7.862678384824163e-06,
      "loss": 1.0045,
      "step": 18494
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5412838459014893,
      "learning_rate": 7.861683518553296e-06,
      "loss": 0.4322,
      "step": 18495
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5253775119781494,
      "learning_rate": 7.860688674460187e-06,
      "loss": 0.4198,
      "step": 18496
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6363129615783691,
      "learning_rate": 7.859693852555161e-06,
      "loss": 0.4979,
      "step": 18497
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5135611295700073,
      "learning_rate": 7.858699052848534e-06,
      "loss": 0.4011,
      "step": 18498
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.806251049041748,
      "learning_rate": 7.85770427535062e-06,
      "loss": 1.0491,
      "step": 18499
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4455616474151611,
      "learning_rate": 7.856709520071735e-06,
      "loss": 1.0171,
      "step": 18500
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4740439653396606,
      "learning_rate": 7.855714787022201e-06,
      "loss": 0.436,
      "step": 18501
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5026905536651611,
      "learning_rate": 7.85472007621234e-06,
      "loss": 0.4235,
      "step": 18502
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7317742109298706,
      "learning_rate": 7.853725387652454e-06,
      "loss": 0.4585,
      "step": 18503
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5231093168258667,
      "learning_rate": 7.85273072135287e-06,
      "loss": 0.4035,
      "step": 18504
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6904126405715942,
      "learning_rate": 7.851736077323901e-06,
      "loss": 0.4362,
      "step": 18505
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5817824602127075,
      "learning_rate": 7.850741455575869e-06,
      "loss": 0.4177,
      "step": 18506
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5864660739898682,
      "learning_rate": 7.849746856119083e-06,
      "loss": 0.4864,
      "step": 18507
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7731881141662598,
      "learning_rate": 7.848752278963857e-06,
      "loss": 0.4002,
      "step": 18508
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5949954986572266,
      "learning_rate": 7.847757724120513e-06,
      "loss": 0.4601,
      "step": 18509
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5987493991851807,
      "learning_rate": 7.846763191599366e-06,
      "loss": 0.392,
      "step": 18510
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.531419038772583,
      "learning_rate": 7.845768681410721e-06,
      "loss": 0.4578,
      "step": 18511
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6140857934951782,
      "learning_rate": 7.844774193564903e-06,
      "loss": 0.4012,
      "step": 18512
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5849045515060425,
      "learning_rate": 7.843779728072222e-06,
      "loss": 0.4928,
      "step": 18513
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5271415710449219,
      "learning_rate": 7.842785284942996e-06,
      "loss": 0.4003,
      "step": 18514
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6878679990768433,
      "learning_rate": 7.841790864187534e-06,
      "loss": 0.4821,
      "step": 18515
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.500903606414795,
      "learning_rate": 7.84079646581615e-06,
      "loss": 0.4152,
      "step": 18516
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6450414657592773,
      "learning_rate": 7.839802089839162e-06,
      "loss": 0.4474,
      "step": 18517
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.2589640617370605,
      "learning_rate": 7.838807736266882e-06,
      "loss": 0.4034,
      "step": 18518
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7060093879699707,
      "learning_rate": 7.83781340510962e-06,
      "loss": 0.5214,
      "step": 18519
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5071687698364258,
      "learning_rate": 7.83681909637769e-06,
      "loss": 0.3779,
      "step": 18520
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6178646087646484,
      "learning_rate": 7.835824810081402e-06,
      "loss": 0.4561,
      "step": 18521
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5056551694869995,
      "learning_rate": 7.834830546231079e-06,
      "loss": 0.4144,
      "step": 18522
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4967756271362305,
      "learning_rate": 7.833836304837022e-06,
      "loss": 0.4333,
      "step": 18523
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4901643991470337,
      "learning_rate": 7.832842085909543e-06,
      "loss": 0.4136,
      "step": 18524
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6050459146499634,
      "learning_rate": 7.83184788945896e-06,
      "loss": 0.4476,
      "step": 18525
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4776991605758667,
      "learning_rate": 7.830853715495583e-06,
      "loss": 0.3864,
      "step": 18526
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.548583745956421,
      "learning_rate": 7.829859564029718e-06,
      "loss": 0.4534,
      "step": 18527
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4632471799850464,
      "learning_rate": 7.828865435071682e-06,
      "loss": 0.4029,
      "step": 18528
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6268826723098755,
      "learning_rate": 7.827871328631781e-06,
      "loss": 0.4641,
      "step": 18529
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.439517617225647,
      "learning_rate": 7.826877244720333e-06,
      "loss": 0.3906,
      "step": 18530
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6252238750457764,
      "learning_rate": 7.825883183347638e-06,
      "loss": 0.4144,
      "step": 18531
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.502575159072876,
      "learning_rate": 7.82488914452401e-06,
      "loss": 0.4056,
      "step": 18532
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5177730321884155,
      "learning_rate": 7.823895128259763e-06,
      "loss": 0.4306,
      "step": 18533
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.499587059020996,
      "learning_rate": 7.822901134565206e-06,
      "loss": 0.3816,
      "step": 18534
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.651241660118103,
      "learning_rate": 7.82190716345064e-06,
      "loss": 0.4937,
      "step": 18535
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.503180742263794,
      "learning_rate": 7.820913214926381e-06,
      "loss": 0.4051,
      "step": 18536
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5774630308151245,
      "learning_rate": 7.819919289002735e-06,
      "loss": 0.4383,
      "step": 18537
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6106724739074707,
      "learning_rate": 7.81892538569002e-06,
      "loss": 0.4324,
      "step": 18538
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6588523387908936,
      "learning_rate": 7.817931504998529e-06,
      "loss": 0.4651,
      "step": 18539
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5608750581741333,
      "learning_rate": 7.816937646938579e-06,
      "loss": 0.419,
      "step": 18540
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6292940378189087,
      "learning_rate": 7.815943811520474e-06,
      "loss": 0.4976,
      "step": 18541
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6250040531158447,
      "learning_rate": 7.81494999875453e-06,
      "loss": 0.3953,
      "step": 18542
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7045987844467163,
      "learning_rate": 7.813956208651045e-06,
      "loss": 0.4719,
      "step": 18543
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4603569507598877,
      "learning_rate": 7.812962441220329e-06,
      "loss": 0.3932,
      "step": 18544
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4696294069290161,
      "learning_rate": 7.811968696472692e-06,
      "loss": 0.4348,
      "step": 18545
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6113648414611816,
      "learning_rate": 7.81097497441844e-06,
      "loss": 0.4469,
      "step": 18546
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5699180364608765,
      "learning_rate": 7.809981275067872e-06,
      "loss": 0.4311,
      "step": 18547
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5718514919281006,
      "learning_rate": 7.808987598431303e-06,
      "loss": 0.3965,
      "step": 18548
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5266190767288208,
      "learning_rate": 7.807993944519036e-06,
      "loss": 0.4557,
      "step": 18549
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5215535163879395,
      "learning_rate": 7.807000313341379e-06,
      "loss": 0.406,
      "step": 18550
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5751140117645264,
      "learning_rate": 7.806006704908636e-06,
      "loss": 0.4184,
      "step": 18551
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6131865978240967,
      "learning_rate": 7.805013119231108e-06,
      "loss": 0.4614,
      "step": 18552
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5499991178512573,
      "learning_rate": 7.804019556319107e-06,
      "loss": 0.4828,
      "step": 18553
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.556776762008667,
      "learning_rate": 7.803026016182931e-06,
      "loss": 0.4269,
      "step": 18554
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6945743560791016,
      "learning_rate": 7.802032498832895e-06,
      "loss": 0.454,
      "step": 18555
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4941891431808472,
      "learning_rate": 7.801039004279292e-06,
      "loss": 0.3952,
      "step": 18556
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6408212184906006,
      "learning_rate": 7.800045532532429e-06,
      "loss": 0.4457,
      "step": 18557
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5352569818496704,
      "learning_rate": 7.799052083602616e-06,
      "loss": 0.3938,
      "step": 18558
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5307306051254272,
      "learning_rate": 7.798058657500155e-06,
      "loss": 0.4587,
      "step": 18559
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5320677757263184,
      "learning_rate": 7.79706525423534e-06,
      "loss": 0.4149,
      "step": 18560
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5996001958847046,
      "learning_rate": 7.796071873818483e-06,
      "loss": 0.466,
      "step": 18561
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.55010187625885,
      "learning_rate": 7.795078516259884e-06,
      "loss": 0.3985,
      "step": 18562
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.516567349433899,
      "learning_rate": 7.79408518156985e-06,
      "loss": 0.4546,
      "step": 18563
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5233829021453857,
      "learning_rate": 7.793091869758679e-06,
      "loss": 0.4075,
      "step": 18564
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6263618469238281,
      "learning_rate": 7.792098580836673e-06,
      "loss": 0.4759,
      "step": 18565
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5489084720611572,
      "learning_rate": 7.791105314814136e-06,
      "loss": 0.4124,
      "step": 18566
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.510577917098999,
      "learning_rate": 7.79011207170137e-06,
      "loss": 0.4261,
      "step": 18567
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5052281618118286,
      "learning_rate": 7.789118851508673e-06,
      "loss": 0.3899,
      "step": 18568
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6443535089492798,
      "learning_rate": 7.788125654246352e-06,
      "loss": 0.454,
      "step": 18569
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4540514945983887,
      "learning_rate": 7.787132479924702e-06,
      "loss": 0.3713,
      "step": 18570
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6223007440567017,
      "learning_rate": 7.78613932855403e-06,
      "loss": 0.4655,
      "step": 18571
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.539405345916748,
      "learning_rate": 7.785146200144631e-06,
      "loss": 0.4265,
      "step": 18572
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5138130187988281,
      "learning_rate": 7.784153094706807e-06,
      "loss": 0.4723,
      "step": 18573
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5552254915237427,
      "learning_rate": 7.78316001225086e-06,
      "loss": 0.4139,
      "step": 18574
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6472030878067017,
      "learning_rate": 7.782166952787091e-06,
      "loss": 0.4793,
      "step": 18575
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7135075330734253,
      "learning_rate": 7.781173916325793e-06,
      "loss": 0.3959,
      "step": 18576
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5091321468353271,
      "learning_rate": 7.780180902877271e-06,
      "loss": 0.481,
      "step": 18577
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5287895202636719,
      "learning_rate": 7.779187912451822e-06,
      "loss": 0.4207,
      "step": 18578
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5625250339508057,
      "learning_rate": 7.77819494505975e-06,
      "loss": 0.4845,
      "step": 18579
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5153740644454956,
      "learning_rate": 7.777202000711345e-06,
      "loss": 0.3755,
      "step": 18580
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4997153282165527,
      "learning_rate": 7.77620907941691e-06,
      "loss": 0.4825,
      "step": 18581
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4413621425628662,
      "learning_rate": 7.775216181186743e-06,
      "loss": 0.3799,
      "step": 18582
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.454184889793396,
      "learning_rate": 7.774223306031144e-06,
      "loss": 0.4212,
      "step": 18583
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5217585563659668,
      "learning_rate": 7.773230453960407e-06,
      "loss": 0.3922,
      "step": 18584
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5484156608581543,
      "learning_rate": 7.772237624984828e-06,
      "loss": 0.4489,
      "step": 18585
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.507703185081482,
      "learning_rate": 7.77124481911471e-06,
      "loss": 0.3888,
      "step": 18586
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6612837314605713,
      "learning_rate": 7.77025203636035e-06,
      "loss": 0.5138,
      "step": 18587
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5714478492736816,
      "learning_rate": 7.769259276732037e-06,
      "loss": 0.3951,
      "step": 18588
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5185749530792236,
      "learning_rate": 7.768266540240075e-06,
      "loss": 0.4353,
      "step": 18589
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5259675979614258,
      "learning_rate": 7.767273826894756e-06,
      "loss": 0.4145,
      "step": 18590
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5030789375305176,
      "learning_rate": 7.766281136706381e-06,
      "loss": 0.4512,
      "step": 18591
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5010735988616943,
      "learning_rate": 7.76528846968524e-06,
      "loss": 0.4606,
      "step": 18592
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5696930885314941,
      "learning_rate": 7.76429582584163e-06,
      "loss": 0.4669,
      "step": 18593
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4969792366027832,
      "learning_rate": 7.76330320518585e-06,
      "loss": 0.3885,
      "step": 18594
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5464884042739868,
      "learning_rate": 7.762310607728194e-06,
      "loss": 0.4849,
      "step": 18595
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6738556623458862,
      "learning_rate": 7.761318033478949e-06,
      "loss": 0.4291,
      "step": 18596
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4607141017913818,
      "learning_rate": 7.76032548244842e-06,
      "loss": 0.458,
      "step": 18597
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5369778871536255,
      "learning_rate": 7.759332954646894e-06,
      "loss": 0.41,
      "step": 18598
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5176231861114502,
      "learning_rate": 7.758340450084674e-06,
      "loss": 0.4189,
      "step": 18599
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5579127073287964,
      "learning_rate": 7.757347968772044e-06,
      "loss": 0.4051,
      "step": 18600
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5376908779144287,
      "learning_rate": 7.756355510719301e-06,
      "loss": 0.4286,
      "step": 18601
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5340150594711304,
      "learning_rate": 7.755363075936742e-06,
      "loss": 0.4387,
      "step": 18602
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5154775381088257,
      "learning_rate": 7.754370664434658e-06,
      "loss": 0.4363,
      "step": 18603
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.568341851234436,
      "learning_rate": 7.753378276223339e-06,
      "loss": 0.4371,
      "step": 18604
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4597376585006714,
      "learning_rate": 7.752385911313078e-06,
      "loss": 0.3851,
      "step": 18605
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.564096212387085,
      "learning_rate": 7.75139356971417e-06,
      "loss": 0.4554,
      "step": 18606
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5256314277648926,
      "learning_rate": 7.75040125143691e-06,
      "loss": 0.44,
      "step": 18607
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5104329586029053,
      "learning_rate": 7.749408956491584e-06,
      "loss": 0.4036,
      "step": 18608
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4780669212341309,
      "learning_rate": 7.748416684888484e-06,
      "loss": 0.4143,
      "step": 18609
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4807621240615845,
      "learning_rate": 7.747424436637905e-06,
      "loss": 0.3914,
      "step": 18610
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6686900854110718,
      "learning_rate": 7.74643221175014e-06,
      "loss": 0.4762,
      "step": 18611
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5178922414779663,
      "learning_rate": 7.745440010235471e-06,
      "loss": 0.3874,
      "step": 18612
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8161107301712036,
      "learning_rate": 7.744447832104197e-06,
      "loss": 1.0126,
      "step": 18613
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4377161264419556,
      "learning_rate": 7.743455677366605e-06,
      "loss": 1.0117,
      "step": 18614
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6653156280517578,
      "learning_rate": 7.74246354603299e-06,
      "loss": 0.5421,
      "step": 18615
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5970983505249023,
      "learning_rate": 7.741471438113634e-06,
      "loss": 0.4141,
      "step": 18616
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6147149801254272,
      "learning_rate": 7.740479353618831e-06,
      "loss": 0.5097,
      "step": 18617
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5869263410568237,
      "learning_rate": 7.73948729255887e-06,
      "loss": 0.432,
      "step": 18618
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5546194314956665,
      "learning_rate": 7.738495254944042e-06,
      "loss": 0.4436,
      "step": 18619
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4853335618972778,
      "learning_rate": 7.737503240784634e-06,
      "loss": 0.42,
      "step": 18620
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.532287836074829,
      "learning_rate": 7.736511250090934e-06,
      "loss": 0.4505,
      "step": 18621
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6098072528839111,
      "learning_rate": 7.735519282873234e-06,
      "loss": 0.4322,
      "step": 18622
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5277541875839233,
      "learning_rate": 7.73452733914182e-06,
      "loss": 0.4445,
      "step": 18623
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5877430438995361,
      "learning_rate": 7.733535418906978e-06,
      "loss": 0.4296,
      "step": 18624
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5624686479568481,
      "learning_rate": 7.732543522178999e-06,
      "loss": 0.5242,
      "step": 18625
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5486822128295898,
      "learning_rate": 7.731551648968167e-06,
      "loss": 0.4494,
      "step": 18626
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5390626192092896,
      "learning_rate": 7.730559799284778e-06,
      "loss": 0.4875,
      "step": 18627
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.527782917022705,
      "learning_rate": 7.72956797313911e-06,
      "loss": 0.4157,
      "step": 18628
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.605536937713623,
      "learning_rate": 7.72857617054145e-06,
      "loss": 0.5069,
      "step": 18629
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5130536556243896,
      "learning_rate": 7.727584391502089e-06,
      "loss": 0.3838,
      "step": 18630
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5461028814315796,
      "learning_rate": 7.726592636031315e-06,
      "loss": 0.4218,
      "step": 18631
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5538932085037231,
      "learning_rate": 7.725600904139406e-06,
      "loss": 0.4318,
      "step": 18632
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5921452045440674,
      "learning_rate": 7.724609195836654e-06,
      "loss": 0.4543,
      "step": 18633
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5849050283432007,
      "learning_rate": 7.723617511133343e-06,
      "loss": 0.4196,
      "step": 18634
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5238368511199951,
      "learning_rate": 7.722625850039761e-06,
      "loss": 0.4344,
      "step": 18635
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5231432914733887,
      "learning_rate": 7.72163421256619e-06,
      "loss": 0.4161,
      "step": 18636
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5960758924484253,
      "learning_rate": 7.720642598722911e-06,
      "loss": 0.4973,
      "step": 18637
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6061292886734009,
      "learning_rate": 7.719651008520217e-06,
      "loss": 0.4327,
      "step": 18638
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.577960729598999,
      "learning_rate": 7.71865944196839e-06,
      "loss": 0.4911,
      "step": 18639
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5168148279190063,
      "learning_rate": 7.717667899077708e-06,
      "loss": 0.3985,
      "step": 18640
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5943629741668701,
      "learning_rate": 7.716676379858464e-06,
      "loss": 0.4966,
      "step": 18641
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5938974618911743,
      "learning_rate": 7.715684884320933e-06,
      "loss": 0.4371,
      "step": 18642
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.600322961807251,
      "learning_rate": 7.714693412475408e-06,
      "loss": 0.4854,
      "step": 18643
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.680785059928894,
      "learning_rate": 7.713701964332163e-06,
      "loss": 0.4234,
      "step": 18644
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5552643537521362,
      "learning_rate": 7.712710539901483e-06,
      "loss": 0.4302,
      "step": 18645
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.456484079360962,
      "learning_rate": 7.711719139193654e-06,
      "loss": 0.3955,
      "step": 18646
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4812901020050049,
      "learning_rate": 7.71072776221896e-06,
      "loss": 0.4312,
      "step": 18647
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5047470331192017,
      "learning_rate": 7.709736408987675e-06,
      "loss": 0.4059,
      "step": 18648
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.464357614517212,
      "learning_rate": 7.708745079510088e-06,
      "loss": 0.3948,
      "step": 18649
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4737972021102905,
      "learning_rate": 7.707753773796475e-06,
      "loss": 0.3927,
      "step": 18650
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.599880337715149,
      "learning_rate": 7.706762491857128e-06,
      "loss": 0.4416,
      "step": 18651
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5833680629730225,
      "learning_rate": 7.705771233702317e-06,
      "loss": 0.4029,
      "step": 18652
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6312873363494873,
      "learning_rate": 7.704779999342324e-06,
      "loss": 0.4537,
      "step": 18653
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4411636590957642,
      "learning_rate": 7.703788788787436e-06,
      "loss": 0.3759,
      "step": 18654
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5407752990722656,
      "learning_rate": 7.702797602047932e-06,
      "loss": 0.4208,
      "step": 18655
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.546501636505127,
      "learning_rate": 7.701806439134087e-06,
      "loss": 0.4388,
      "step": 18656
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5508863925933838,
      "learning_rate": 7.700815300056185e-06,
      "loss": 0.4664,
      "step": 18657
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5316892862319946,
      "learning_rate": 7.699824184824504e-06,
      "loss": 0.4004,
      "step": 18658
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5245517492294312,
      "learning_rate": 7.698833093449328e-06,
      "loss": 0.426,
      "step": 18659
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5171276330947876,
      "learning_rate": 7.697842025940929e-06,
      "loss": 0.3979,
      "step": 18660
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.644020676612854,
      "learning_rate": 7.69685098230959e-06,
      "loss": 0.4386,
      "step": 18661
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5477612018585205,
      "learning_rate": 7.695859962565588e-06,
      "loss": 0.3868,
      "step": 18662
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.486477255821228,
      "learning_rate": 7.694868966719206e-06,
      "loss": 0.464,
      "step": 18663
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4974298477172852,
      "learning_rate": 7.69387799478072e-06,
      "loss": 0.4335,
      "step": 18664
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.532070517539978,
      "learning_rate": 7.692887046760401e-06,
      "loss": 0.4435,
      "step": 18665
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5030248165130615,
      "learning_rate": 7.691896122668537e-06,
      "loss": 0.4012,
      "step": 18666
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1101298332214355,
      "learning_rate": 7.690905222515401e-06,
      "loss": 0.4579,
      "step": 18667
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6288440227508545,
      "learning_rate": 7.689914346311268e-06,
      "loss": 0.413,
      "step": 18668
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.530523419380188,
      "learning_rate": 7.688923494066418e-06,
      "loss": 0.4398,
      "step": 18669
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.571953296661377,
      "learning_rate": 7.687932665791126e-06,
      "loss": 0.4231,
      "step": 18670
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6586010456085205,
      "learning_rate": 7.686941861495672e-06,
      "loss": 0.4819,
      "step": 18671
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5437989234924316,
      "learning_rate": 7.685951081190329e-06,
      "loss": 0.4029,
      "step": 18672
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6602171659469604,
      "learning_rate": 7.684960324885371e-06,
      "loss": 0.4805,
      "step": 18673
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8749412298202515,
      "learning_rate": 7.68396959259108e-06,
      "loss": 0.4023,
      "step": 18674
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5590002536773682,
      "learning_rate": 7.682978884317727e-06,
      "loss": 0.4064,
      "step": 18675
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6099318265914917,
      "learning_rate": 7.681988200075587e-06,
      "loss": 0.4377,
      "step": 18676
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5519847869873047,
      "learning_rate": 7.680997539874936e-06,
      "loss": 0.4305,
      "step": 18677
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4925580024719238,
      "learning_rate": 7.680006903726047e-06,
      "loss": 0.3972,
      "step": 18678
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.606074333190918,
      "learning_rate": 7.679016291639201e-06,
      "loss": 0.4668,
      "step": 18679
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5440528392791748,
      "learning_rate": 7.678025703624665e-06,
      "loss": 0.4323,
      "step": 18680
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.610739827156067,
      "learning_rate": 7.677035139692715e-06,
      "loss": 0.4417,
      "step": 18681
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.45991051197052,
      "learning_rate": 7.676044599853625e-06,
      "loss": 0.3748,
      "step": 18682
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.528761625289917,
      "learning_rate": 7.675054084117673e-06,
      "loss": 0.4508,
      "step": 18683
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5593534708023071,
      "learning_rate": 7.67406359249512e-06,
      "loss": 0.4406,
      "step": 18684
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.540095567703247,
      "learning_rate": 7.673073124996253e-06,
      "loss": 0.4389,
      "step": 18685
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.508028507232666,
      "learning_rate": 7.672082681631334e-06,
      "loss": 0.4146,
      "step": 18686
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7564820051193237,
      "learning_rate": 7.671092262410646e-06,
      "loss": 0.5215,
      "step": 18687
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4807921648025513,
      "learning_rate": 7.670101867344452e-06,
      "loss": 0.382,
      "step": 18688
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5857559442520142,
      "learning_rate": 7.669111496443026e-06,
      "loss": 0.4527,
      "step": 18689
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4362236261367798,
      "learning_rate": 7.668121149716642e-06,
      "loss": 0.3948,
      "step": 18690
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6428619623184204,
      "learning_rate": 7.667130827175575e-06,
      "loss": 0.4904,
      "step": 18691
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5251227617263794,
      "learning_rate": 7.666140528830088e-06,
      "loss": 0.4066,
      "step": 18692
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5665960311889648,
      "learning_rate": 7.665150254690455e-06,
      "loss": 0.4836,
      "step": 18693
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5068883895874023,
      "learning_rate": 7.664160004766945e-06,
      "loss": 0.4118,
      "step": 18694
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9873284101486206,
      "learning_rate": 7.663169779069839e-06,
      "loss": 0.4328,
      "step": 18695
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5437992811203003,
      "learning_rate": 7.662179577609392e-06,
      "loss": 0.4135,
      "step": 18696
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6609517335891724,
      "learning_rate": 7.661189400395885e-06,
      "loss": 0.4596,
      "step": 18697
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5223584175109863,
      "learning_rate": 7.66019924743958e-06,
      "loss": 0.3865,
      "step": 18698
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4961284399032593,
      "learning_rate": 7.659209118750754e-06,
      "loss": 0.4723,
      "step": 18699
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.590131402015686,
      "learning_rate": 7.658219014339672e-06,
      "loss": 0.4166,
      "step": 18700
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4916483163833618,
      "learning_rate": 7.6572289342166e-06,
      "loss": 0.4536,
      "step": 18701
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0447030067443848,
      "learning_rate": 7.656238878391815e-06,
      "loss": 0.3915,
      "step": 18702
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5158932209014893,
      "learning_rate": 7.655248846875581e-06,
      "loss": 0.4019,
      "step": 18703
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6027079820632935,
      "learning_rate": 7.65425883967816e-06,
      "loss": 0.3998,
      "step": 18704
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5227162837982178,
      "learning_rate": 7.65326885680983e-06,
      "loss": 0.3948,
      "step": 18705
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5208065509796143,
      "learning_rate": 7.652278898280852e-06,
      "loss": 0.3966,
      "step": 18706
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.519763708114624,
      "learning_rate": 7.651288964101501e-06,
      "loss": 0.4207,
      "step": 18707
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.535927414894104,
      "learning_rate": 7.650299054282036e-06,
      "loss": 0.3847,
      "step": 18708
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5041106939315796,
      "learning_rate": 7.649309168832726e-06,
      "loss": 0.4124,
      "step": 18709
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5340005159378052,
      "learning_rate": 7.64831930776384e-06,
      "loss": 0.4055,
      "step": 18710
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4654057025909424,
      "learning_rate": 7.647329471085647e-06,
      "loss": 0.4435,
      "step": 18711
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5190913677215576,
      "learning_rate": 7.646339658808404e-06,
      "loss": 0.4299,
      "step": 18712
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5341756343841553,
      "learning_rate": 7.645349870942385e-06,
      "loss": 0.461,
      "step": 18713
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5687810182571411,
      "learning_rate": 7.64436010749785e-06,
      "loss": 0.4532,
      "step": 18714
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5404200553894043,
      "learning_rate": 7.643370368485072e-06,
      "loss": 0.4353,
      "step": 18715
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5337672233581543,
      "learning_rate": 7.642380653914313e-06,
      "loss": 0.3998,
      "step": 18716
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.626037836074829,
      "learning_rate": 7.64139096379583e-06,
      "loss": 0.5196,
      "step": 18717
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.575796127319336,
      "learning_rate": 7.640401298139898e-06,
      "loss": 0.4084,
      "step": 18718
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5478352308273315,
      "learning_rate": 7.63941165695678e-06,
      "loss": 0.4188,
      "step": 18719
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4857362508773804,
      "learning_rate": 7.638422040256733e-06,
      "loss": 0.4065,
      "step": 18720
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6346306800842285,
      "learning_rate": 7.637432448050029e-06,
      "loss": 0.4612,
      "step": 18721
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5301153659820557,
      "learning_rate": 7.636442880346926e-06,
      "loss": 0.3995,
      "step": 18722
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5196171998977661,
      "learning_rate": 7.635453337157695e-06,
      "loss": 0.4528,
      "step": 18723
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.465989351272583,
      "learning_rate": 7.63446381849259e-06,
      "loss": 0.4012,
      "step": 18724
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5472848415374756,
      "learning_rate": 7.633474324361876e-06,
      "loss": 0.4657,
      "step": 18725
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5272753238677979,
      "learning_rate": 7.632484854775821e-06,
      "loss": 0.4196,
      "step": 18726
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5551141500473022,
      "learning_rate": 7.631495409744685e-06,
      "loss": 0.4145,
      "step": 18727
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4973715543746948,
      "learning_rate": 7.630505989278725e-06,
      "loss": 0.3851,
      "step": 18728
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5860185623168945,
      "learning_rate": 7.629516593388209e-06,
      "loss": 0.4433,
      "step": 18729
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5645465850830078,
      "learning_rate": 7.628527222083395e-06,
      "loss": 0.4296,
      "step": 18730
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9282538890838623,
      "learning_rate": 7.627537875374551e-06,
      "loss": 0.9833,
      "step": 18731
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5557876825332642,
      "learning_rate": 7.626548553271927e-06,
      "loss": 1.0841,
      "step": 18732
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5881260633468628,
      "learning_rate": 7.625559255785793e-06,
      "loss": 0.4516,
      "step": 18733
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6164780855178833,
      "learning_rate": 7.624569982926403e-06,
      "loss": 0.4159,
      "step": 18734
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5863075256347656,
      "learning_rate": 7.623580734704026e-06,
      "loss": 0.3914,
      "step": 18735
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.603479027748108,
      "learning_rate": 7.622591511128914e-06,
      "loss": 0.4704,
      "step": 18736
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5244572162628174,
      "learning_rate": 7.621602312211328e-06,
      "loss": 0.4589,
      "step": 18737
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5118229389190674,
      "learning_rate": 7.620613137961532e-06,
      "loss": 0.404,
      "step": 18738
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5178431272506714,
      "learning_rate": 7.6196239883897834e-06,
      "loss": 0.4937,
      "step": 18739
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5808113813400269,
      "learning_rate": 7.618634863506338e-06,
      "loss": 0.426,
      "step": 18740
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5215976238250732,
      "learning_rate": 7.617645763321457e-06,
      "loss": 0.4358,
      "step": 18741
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6235030889511108,
      "learning_rate": 7.6166566878453986e-06,
      "loss": 0.4311,
      "step": 18742
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.543591022491455,
      "learning_rate": 7.615667637088425e-06,
      "loss": 0.4435,
      "step": 18743
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5526684522628784,
      "learning_rate": 7.614678611060789e-06,
      "loss": 0.4076,
      "step": 18744
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5504887104034424,
      "learning_rate": 7.613689609772748e-06,
      "loss": 0.4508,
      "step": 18745
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.513687252998352,
      "learning_rate": 7.612700633234563e-06,
      "loss": 0.3925,
      "step": 18746
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5140866041183472,
      "learning_rate": 7.611711681456493e-06,
      "loss": 0.446,
      "step": 18747
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5652157068252563,
      "learning_rate": 7.610722754448787e-06,
      "loss": 0.417,
      "step": 18748
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6890588998794556,
      "learning_rate": 7.609733852221709e-06,
      "loss": 0.427,
      "step": 18749
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5508363246917725,
      "learning_rate": 7.608744974785512e-06,
      "loss": 0.4395,
      "step": 18750
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.707048773765564,
      "learning_rate": 7.607756122150456e-06,
      "loss": 0.4682,
      "step": 18751
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5709861516952515,
      "learning_rate": 7.606767294326794e-06,
      "loss": 0.4305,
      "step": 18752
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6781481504440308,
      "learning_rate": 7.605778491324778e-06,
      "loss": 0.4729,
      "step": 18753
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5931341648101807,
      "learning_rate": 7.60478971315467e-06,
      "loss": 0.4087,
      "step": 18754
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8065513372421265,
      "learning_rate": 7.603800959826726e-06,
      "loss": 1.0668,
      "step": 18755
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6047557592391968,
      "learning_rate": 7.602812231351193e-06,
      "loss": 1.0458,
      "step": 18756
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6767886877059937,
      "learning_rate": 7.601823527738333e-06,
      "loss": 0.4423,
      "step": 18757
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5817828178405762,
      "learning_rate": 7.600834848998396e-06,
      "loss": 0.4027,
      "step": 18758
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4710789918899536,
      "learning_rate": 7.599846195141642e-06,
      "loss": 0.3915,
      "step": 18759
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6235249042510986,
      "learning_rate": 7.598857566178319e-06,
      "loss": 0.4333,
      "step": 18760
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5185579061508179,
      "learning_rate": 7.59786896211868e-06,
      "loss": 0.4212,
      "step": 18761
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5271179676055908,
      "learning_rate": 7.5968803829729845e-06,
      "loss": 0.4055,
      "step": 18762
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5299640893936157,
      "learning_rate": 7.5958918287514825e-06,
      "loss": 0.4712,
      "step": 18763
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5352203845977783,
      "learning_rate": 7.594903299464423e-06,
      "loss": 0.4075,
      "step": 18764
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5206977128982544,
      "learning_rate": 7.593914795122065e-06,
      "loss": 0.4508,
      "step": 18765
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4641032218933105,
      "learning_rate": 7.592926315734656e-06,
      "loss": 0.3826,
      "step": 18766
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3982794284820557,
      "learning_rate": 7.591937861312453e-06,
      "loss": 0.4163,
      "step": 18767
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5161001682281494,
      "learning_rate": 7.590949431865704e-06,
      "loss": 0.3926,
      "step": 18768
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5009715557098389,
      "learning_rate": 7.58996102740466e-06,
      "loss": 0.4391,
      "step": 18769
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5374523401260376,
      "learning_rate": 7.588972647939576e-06,
      "loss": 0.4246,
      "step": 18770
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5704518556594849,
      "learning_rate": 7.587984293480701e-06,
      "loss": 0.4414,
      "step": 18771
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4731165170669556,
      "learning_rate": 7.5869959640382864e-06,
      "loss": 0.4157,
      "step": 18772
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.535954475402832,
      "learning_rate": 7.586007659622579e-06,
      "loss": 0.4634,
      "step": 18773
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4980671405792236,
      "learning_rate": 7.585019380243835e-06,
      "loss": 0.4035,
      "step": 18774
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5634700059890747,
      "learning_rate": 7.5840311259123035e-06,
      "loss": 0.4341,
      "step": 18775
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5042933225631714,
      "learning_rate": 7.583042896638229e-06,
      "loss": 0.3851,
      "step": 18776
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9108833074569702,
      "learning_rate": 7.582054692431866e-06,
      "loss": 1.0716,
      "step": 18777
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5996776819229126,
      "learning_rate": 7.58106651330346e-06,
      "loss": 1.0391,
      "step": 18778
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5565588474273682,
      "learning_rate": 7.580078359263268e-06,
      "loss": 0.4728,
      "step": 18779
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5283657312393188,
      "learning_rate": 7.579090230321529e-06,
      "loss": 0.3827,
      "step": 18780
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5082509517669678,
      "learning_rate": 7.578102126488494e-06,
      "loss": 0.4113,
      "step": 18781
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5755335092544556,
      "learning_rate": 7.577114047774416e-06,
      "loss": 0.3969,
      "step": 18782
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5799275636672974,
      "learning_rate": 7.576125994189541e-06,
      "loss": 0.466,
      "step": 18783
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5481075048446655,
      "learning_rate": 7.57513796574411e-06,
      "loss": 0.3968,
      "step": 18784
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5898592472076416,
      "learning_rate": 7.574149962448379e-06,
      "loss": 0.5036,
      "step": 18785
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6952159404754639,
      "learning_rate": 7.573161984312591e-06,
      "loss": 0.441,
      "step": 18786
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.482109785079956,
      "learning_rate": 7.572174031346997e-06,
      "loss": 0.4492,
      "step": 18787
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4800410270690918,
      "learning_rate": 7.571186103561838e-06,
      "loss": 0.3917,
      "step": 18788
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6606673002243042,
      "learning_rate": 7.570198200967363e-06,
      "loss": 0.459,
      "step": 18789
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5435948371887207,
      "learning_rate": 7.569210323573818e-06,
      "loss": 0.3968,
      "step": 18790
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5860881805419922,
      "learning_rate": 7.568222471391447e-06,
      "loss": 0.4396,
      "step": 18791
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5342867374420166,
      "learning_rate": 7.567234644430502e-06,
      "loss": 0.4109,
      "step": 18792
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5784343481063843,
      "learning_rate": 7.566246842701223e-06,
      "loss": 0.5364,
      "step": 18793
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.491578221321106,
      "learning_rate": 7.565259066213853e-06,
      "loss": 0.384,
      "step": 18794
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7229171991348267,
      "learning_rate": 7.564271314978642e-06,
      "loss": 0.4339,
      "step": 18795
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5595800876617432,
      "learning_rate": 7.563283589005834e-06,
      "loss": 0.4116,
      "step": 18796
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5357565879821777,
      "learning_rate": 7.562295888305669e-06,
      "loss": 0.4971,
      "step": 18797
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5570781230926514,
      "learning_rate": 7.561308212888394e-06,
      "loss": 0.418,
      "step": 18798
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5263007879257202,
      "learning_rate": 7.560320562764252e-06,
      "loss": 0.42,
      "step": 18799
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5502965450286865,
      "learning_rate": 7.55933293794349e-06,
      "loss": 0.4043,
      "step": 18800
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6445125341415405,
      "learning_rate": 7.558345338436347e-06,
      "loss": 0.5194,
      "step": 18801
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4558956623077393,
      "learning_rate": 7.557357764253065e-06,
      "loss": 0.3745,
      "step": 18802
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6294748783111572,
      "learning_rate": 7.55637021540389e-06,
      "loss": 0.5261,
      "step": 18803
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.453456997871399,
      "learning_rate": 7.555382691899067e-06,
      "loss": 0.3932,
      "step": 18804
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6304900646209717,
      "learning_rate": 7.5543951937488295e-06,
      "loss": 0.5367,
      "step": 18805
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4432333707809448,
      "learning_rate": 7.553407720963427e-06,
      "loss": 0.3667,
      "step": 18806
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6175498962402344,
      "learning_rate": 7.552420273553096e-06,
      "loss": 0.4711,
      "step": 18807
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5163207054138184,
      "learning_rate": 7.551432851528089e-06,
      "loss": 0.4218,
      "step": 18808
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.524963140487671,
      "learning_rate": 7.5504454548986305e-06,
      "loss": 0.4421,
      "step": 18809
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6113735437393188,
      "learning_rate": 7.549458083674972e-06,
      "loss": 0.4105,
      "step": 18810
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5346577167510986,
      "learning_rate": 7.548470737867351e-06,
      "loss": 0.4224,
      "step": 18811
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.112220525741577,
      "learning_rate": 7.547483417486011e-06,
      "loss": 0.394,
      "step": 18812
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.533212423324585,
      "learning_rate": 7.546496122541188e-06,
      "loss": 0.4302,
      "step": 18813
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.494651198387146,
      "learning_rate": 7.545508853043122e-06,
      "loss": 0.3904,
      "step": 18814
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5124317407608032,
      "learning_rate": 7.544521609002056e-06,
      "loss": 0.4885,
      "step": 18815
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4930227994918823,
      "learning_rate": 7.54353439042823e-06,
      "loss": 0.4159,
      "step": 18816
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.516617774963379,
      "learning_rate": 7.542547197331875e-06,
      "loss": 0.399,
      "step": 18817
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5075984001159668,
      "learning_rate": 7.54156002972324e-06,
      "loss": 0.3918,
      "step": 18818
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.701093077659607,
      "learning_rate": 7.540572887612554e-06,
      "loss": 0.4753,
      "step": 18819
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4672445058822632,
      "learning_rate": 7.539585771010065e-06,
      "loss": 0.385,
      "step": 18820
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5108882188796997,
      "learning_rate": 7.538598679926004e-06,
      "loss": 0.4444,
      "step": 18821
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6078170537948608,
      "learning_rate": 7.537611614370608e-06,
      "loss": 0.433,
      "step": 18822
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5737624168395996,
      "learning_rate": 7.53662457435412e-06,
      "loss": 0.4207,
      "step": 18823
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.47230064868927,
      "learning_rate": 7.535637559886777e-06,
      "loss": 0.3871,
      "step": 18824
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6316001415252686,
      "learning_rate": 7.534650570978808e-06,
      "loss": 0.4791,
      "step": 18825
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5239405632019043,
      "learning_rate": 7.533663607640457e-06,
      "loss": 0.4058,
      "step": 18826
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.583060383796692,
      "learning_rate": 7.532676669881955e-06,
      "loss": 0.4636,
      "step": 18827
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4535987377166748,
      "learning_rate": 7.531689757713548e-06,
      "loss": 0.394,
      "step": 18828
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4807418584823608,
      "learning_rate": 7.530702871145461e-06,
      "loss": 0.4859,
      "step": 18829
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5472794771194458,
      "learning_rate": 7.529716010187933e-06,
      "loss": 0.4245,
      "step": 18830
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5011568069458008,
      "learning_rate": 7.528729174851203e-06,
      "loss": 0.5623,
      "step": 18831
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5730159282684326,
      "learning_rate": 7.5277423651455035e-06,
      "loss": 0.4008,
      "step": 18832
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6506609916687012,
      "learning_rate": 7.5267555810810655e-06,
      "loss": 0.4744,
      "step": 18833
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5385332107543945,
      "learning_rate": 7.525768822668129e-06,
      "loss": 0.403,
      "step": 18834
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3944140672683716,
      "learning_rate": 7.524782089916924e-06,
      "loss": 0.4419,
      "step": 18835
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4899146556854248,
      "learning_rate": 7.523795382837693e-06,
      "loss": 0.4114,
      "step": 18836
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5437836647033691,
      "learning_rate": 7.522808701440659e-06,
      "loss": 0.4511,
      "step": 18837
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.509514331817627,
      "learning_rate": 7.521822045736059e-06,
      "loss": 0.3921,
      "step": 18838
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5818305015563965,
      "learning_rate": 7.520835415734129e-06,
      "loss": 0.4483,
      "step": 18839
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8920050859451294,
      "learning_rate": 7.519848811445103e-06,
      "loss": 0.4673,
      "step": 18840
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5577566623687744,
      "learning_rate": 7.518862232879205e-06,
      "loss": 0.416,
      "step": 18841
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0989017486572266,
      "learning_rate": 7.517875680046677e-06,
      "loss": 0.3902,
      "step": 18842
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4707920551300049,
      "learning_rate": 7.516889152957744e-06,
      "loss": 0.4427,
      "step": 18843
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5076959133148193,
      "learning_rate": 7.515902651622646e-06,
      "loss": 0.4011,
      "step": 18844
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5888053178787231,
      "learning_rate": 7.514916176051607e-06,
      "loss": 0.4719,
      "step": 18845
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5144562721252441,
      "learning_rate": 7.513929726254859e-06,
      "loss": 0.4133,
      "step": 18846
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6384786367416382,
      "learning_rate": 7.512943302242638e-06,
      "loss": 0.4202,
      "step": 18847
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5447626113891602,
      "learning_rate": 7.511956904025174e-06,
      "loss": 0.399,
      "step": 18848
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5778132677078247,
      "learning_rate": 7.510970531612693e-06,
      "loss": 0.5027,
      "step": 18849
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.554756760597229,
      "learning_rate": 7.509984185015427e-06,
      "loss": 0.4226,
      "step": 18850
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5817875862121582,
      "learning_rate": 7.508997864243607e-06,
      "loss": 0.4089,
      "step": 18851
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6014132499694824,
      "learning_rate": 7.508011569307465e-06,
      "loss": 0.4416,
      "step": 18852
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5750408172607422,
      "learning_rate": 7.5070253002172235e-06,
      "loss": 0.4362,
      "step": 18853
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5917704105377197,
      "learning_rate": 7.506039056983119e-06,
      "loss": 0.4045,
      "step": 18854
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4787201881408691,
      "learning_rate": 7.505052839615375e-06,
      "loss": 0.4435,
      "step": 18855
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.460248351097107,
      "learning_rate": 7.504066648124227e-06,
      "loss": 0.4164,
      "step": 18856
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6095167398452759,
      "learning_rate": 7.5030804825198965e-06,
      "loss": 0.4615,
      "step": 18857
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4467401504516602,
      "learning_rate": 7.502094342812613e-06,
      "loss": 0.3984,
      "step": 18858
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.48812735080719,
      "learning_rate": 7.5011082290126066e-06,
      "loss": 0.4211,
      "step": 18859
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5855348110198975,
      "learning_rate": 7.500122141130106e-06,
      "loss": 0.4096,
      "step": 18860
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8170944452285767,
      "learning_rate": 7.499136079175333e-06,
      "loss": 0.9272,
      "step": 18861
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4370321035385132,
      "learning_rate": 7.4981500431585195e-06,
      "loss": 1.0558,
      "step": 18862
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.611578345298767,
      "learning_rate": 7.497164033089889e-06,
      "loss": 0.4081,
      "step": 18863
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6095887422561646,
      "learning_rate": 7.4961780489796735e-06,
      "loss": 0.4302,
      "step": 18864
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6747100353240967,
      "learning_rate": 7.495192090838092e-06,
      "loss": 0.5225,
      "step": 18865
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5996832847595215,
      "learning_rate": 7.494206158675375e-06,
      "loss": 0.4192,
      "step": 18866
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4956796169281006,
      "learning_rate": 7.4932202525017475e-06,
      "loss": 0.451,
      "step": 18867
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4705227613449097,
      "learning_rate": 7.492234372327436e-06,
      "loss": 0.4036,
      "step": 18868
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6690218448638916,
      "learning_rate": 7.49124851816266e-06,
      "loss": 0.4317,
      "step": 18869
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5756560564041138,
      "learning_rate": 7.490262690017651e-06,
      "loss": 0.4083,
      "step": 18870
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5518667697906494,
      "learning_rate": 7.48927688790263e-06,
      "loss": 0.4143,
      "step": 18871
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4601492881774902,
      "learning_rate": 7.488291111827826e-06,
      "loss": 0.3612,
      "step": 18872
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5366523265838623,
      "learning_rate": 7.487305361803457e-06,
      "loss": 0.4715,
      "step": 18873
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.563033938407898,
      "learning_rate": 7.486319637839747e-06,
      "loss": 0.4398,
      "step": 18874
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.58432936668396,
      "learning_rate": 7.485333939946926e-06,
      "loss": 0.4632,
      "step": 18875
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.519959568977356,
      "learning_rate": 7.484348268135214e-06,
      "loss": 0.3899,
      "step": 18876
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4963525533676147,
      "learning_rate": 7.48336262241483e-06,
      "loss": 0.4626,
      "step": 18877
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5397050380706787,
      "learning_rate": 7.482377002796e-06,
      "loss": 0.4084,
      "step": 18878
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5143554210662842,
      "learning_rate": 7.481391409288945e-06,
      "loss": 0.4363,
      "step": 18879
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5903663635253906,
      "learning_rate": 7.480405841903894e-06,
      "loss": 0.4245,
      "step": 18880
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.522169589996338,
      "learning_rate": 7.47942030065106e-06,
      "loss": 0.3976,
      "step": 18881
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7198175191879272,
      "learning_rate": 7.478434785540667e-06,
      "loss": 0.4186,
      "step": 18882
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.561529517173767,
      "learning_rate": 7.477449296582939e-06,
      "loss": 0.4442,
      "step": 18883
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.576093316078186,
      "learning_rate": 7.476463833788097e-06,
      "loss": 0.4131,
      "step": 18884
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5304851531982422,
      "learning_rate": 7.47547839716636e-06,
      "loss": 0.4128,
      "step": 18885
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5889267921447754,
      "learning_rate": 7.474492986727946e-06,
      "loss": 0.4836,
      "step": 18886
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7061331272125244,
      "learning_rate": 7.473507602483076e-06,
      "loss": 0.5121,
      "step": 18887
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5311226844787598,
      "learning_rate": 7.472522244441982e-06,
      "loss": 0.4071,
      "step": 18888
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5304359197616577,
      "learning_rate": 7.471536912614866e-06,
      "loss": 0.4709,
      "step": 18889
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.560955286026001,
      "learning_rate": 7.470551607011958e-06,
      "loss": 0.4325,
      "step": 18890
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4702181816101074,
      "learning_rate": 7.469566327643471e-06,
      "loss": 0.4226,
      "step": 18891
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4774800539016724,
      "learning_rate": 7.468581074519634e-06,
      "loss": 0.425,
      "step": 18892
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5897294282913208,
      "learning_rate": 7.467595847650657e-06,
      "loss": 0.4747,
      "step": 18893
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5405532121658325,
      "learning_rate": 7.466610647046757e-06,
      "loss": 0.4065,
      "step": 18894
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6550161838531494,
      "learning_rate": 7.465625472718159e-06,
      "loss": 0.5351,
      "step": 18895
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5315253734588623,
      "learning_rate": 7.464640324675078e-06,
      "loss": 0.3832,
      "step": 18896
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5054469108581543,
      "learning_rate": 7.463655202927729e-06,
      "loss": 0.3977,
      "step": 18897
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5024309158325195,
      "learning_rate": 7.462670107486331e-06,
      "loss": 0.4019,
      "step": 18898
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.504357099533081,
      "learning_rate": 7.461685038361101e-06,
      "loss": 0.4261,
      "step": 18899
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6078498363494873,
      "learning_rate": 7.460699995562259e-06,
      "loss": 0.4157,
      "step": 18900
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5963565111160278,
      "learning_rate": 7.459714979100019e-06,
      "loss": 0.4525,
      "step": 18901
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6671520471572876,
      "learning_rate": 7.458729988984592e-06,
      "loss": 0.4185,
      "step": 18902
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5227032899856567,
      "learning_rate": 7.457745025226203e-06,
      "loss": 0.466,
      "step": 18903
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4573179483413696,
      "learning_rate": 7.4567600878350644e-06,
      "loss": 0.3902,
      "step": 18904
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5413718223571777,
      "learning_rate": 7.455775176821386e-06,
      "loss": 0.4628,
      "step": 18905
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6102908849716187,
      "learning_rate": 7.454790292195391e-06,
      "loss": 0.4239,
      "step": 18906
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.569026231765747,
      "learning_rate": 7.453805433967287e-06,
      "loss": 0.4413,
      "step": 18907
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4978474378585815,
      "learning_rate": 7.452820602147297e-06,
      "loss": 0.3702,
      "step": 18908
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5864824056625366,
      "learning_rate": 7.451835796745628e-06,
      "loss": 0.4529,
      "step": 18909
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.530309796333313,
      "learning_rate": 7.450851017772494e-06,
      "loss": 0.3895,
      "step": 18910
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.58749520778656,
      "learning_rate": 7.449866265238115e-06,
      "loss": 0.4488,
      "step": 18911
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5000637769699097,
      "learning_rate": 7.448881539152701e-06,
      "loss": 0.3832,
      "step": 18912
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6827787160873413,
      "learning_rate": 7.447896839526462e-06,
      "loss": 0.4926,
      "step": 18913
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5282211303710938,
      "learning_rate": 7.446912166369616e-06,
      "loss": 0.4198,
      "step": 18914
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5141961574554443,
      "learning_rate": 7.4459275196923695e-06,
      "loss": 0.4459,
      "step": 18915
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5332309007644653,
      "learning_rate": 7.444942899504945e-06,
      "loss": 0.3919,
      "step": 18916
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5859986543655396,
      "learning_rate": 7.443958305817544e-06,
      "loss": 0.4388,
      "step": 18917
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5342918634414673,
      "learning_rate": 7.442973738640383e-06,
      "loss": 0.4016,
      "step": 18918
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5255568027496338,
      "learning_rate": 7.441989197983674e-06,
      "loss": 0.4505,
      "step": 18919
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4969751834869385,
      "learning_rate": 7.441004683857631e-06,
      "loss": 0.3936,
      "step": 18920
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7035279273986816,
      "learning_rate": 7.440020196272456e-06,
      "loss": 0.4285,
      "step": 18921
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6186550855636597,
      "learning_rate": 7.439035735238367e-06,
      "loss": 0.4222,
      "step": 18922
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5080821514129639,
      "learning_rate": 7.438051300765572e-06,
      "loss": 0.4058,
      "step": 18923
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5651487112045288,
      "learning_rate": 7.437066892864286e-06,
      "loss": 0.4003,
      "step": 18924
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5922831296920776,
      "learning_rate": 7.4360825115447096e-06,
      "loss": 0.4696,
      "step": 18925
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.512966513633728,
      "learning_rate": 7.43509815681706e-06,
      "loss": 0.3977,
      "step": 18926
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.469810128211975,
      "learning_rate": 7.434113828691542e-06,
      "loss": 0.4367,
      "step": 18927
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6026962995529175,
      "learning_rate": 7.43312952717837e-06,
      "loss": 0.4085,
      "step": 18928
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.522510290145874,
      "learning_rate": 7.432145252287748e-06,
      "loss": 0.443,
      "step": 18929
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.56946861743927,
      "learning_rate": 7.431161004029883e-06,
      "loss": 0.4035,
      "step": 18930
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6388710737228394,
      "learning_rate": 7.430176782414988e-06,
      "loss": 0.4949,
      "step": 18931
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4673283100128174,
      "learning_rate": 7.429192587453272e-06,
      "loss": 0.3825,
      "step": 18932
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6174969673156738,
      "learning_rate": 7.428208419154936e-06,
      "loss": 0.4437,
      "step": 18933
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5814141035079956,
      "learning_rate": 7.4272242775301915e-06,
      "loss": 0.4097,
      "step": 18934
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5851801633834839,
      "learning_rate": 7.426240162589245e-06,
      "loss": 0.4844,
      "step": 18935
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5401675701141357,
      "learning_rate": 7.4252560743423084e-06,
      "loss": 0.422,
      "step": 18936
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4825907945632935,
      "learning_rate": 7.42427201279958e-06,
      "loss": 0.4351,
      "step": 18937
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5346062183380127,
      "learning_rate": 7.423287977971268e-06,
      "loss": 0.4207,
      "step": 18938
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5372810363769531,
      "learning_rate": 7.4223039698675815e-06,
      "loss": 0.4203,
      "step": 18939
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5294057130813599,
      "learning_rate": 7.4213199884987276e-06,
      "loss": 0.4203,
      "step": 18940
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6509113311767578,
      "learning_rate": 7.420336033874904e-06,
      "loss": 0.4617,
      "step": 18941
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4930176734924316,
      "learning_rate": 7.419352106006323e-06,
      "loss": 0.3761,
      "step": 18942
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.559096097946167,
      "learning_rate": 7.418368204903187e-06,
      "loss": 0.4756,
      "step": 18943
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4262962341308594,
      "learning_rate": 7.417384330575703e-06,
      "loss": 0.3806,
      "step": 18944
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5061010122299194,
      "learning_rate": 7.416400483034073e-06,
      "loss": 0.4639,
      "step": 18945
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5818994045257568,
      "learning_rate": 7.4154166622884984e-06,
      "loss": 0.3997,
      "step": 18946
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.560813069343567,
      "learning_rate": 7.414432868349188e-06,
      "loss": 0.4224,
      "step": 18947
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6183243989944458,
      "learning_rate": 7.413449101226345e-06,
      "loss": 0.4607,
      "step": 18948
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4809536933898926,
      "learning_rate": 7.412465360930168e-06,
      "loss": 0.4409,
      "step": 18949
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5509287118911743,
      "learning_rate": 7.411481647470864e-06,
      "loss": 0.432,
      "step": 18950
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.53388249874115,
      "learning_rate": 7.410497960858631e-06,
      "loss": 0.4579,
      "step": 18951
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4906463623046875,
      "learning_rate": 7.409514301103682e-06,
      "loss": 0.3955,
      "step": 18952
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4574960470199585,
      "learning_rate": 7.408530668216209e-06,
      "loss": 0.4308,
      "step": 18953
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.642869234085083,
      "learning_rate": 7.407547062206415e-06,
      "loss": 0.4057,
      "step": 18954
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6235195398330688,
      "learning_rate": 7.406563483084504e-06,
      "loss": 0.4619,
      "step": 18955
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5927832126617432,
      "learning_rate": 7.40557993086068e-06,
      "loss": 0.4241,
      "step": 18956
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5269209146499634,
      "learning_rate": 7.404596405545137e-06,
      "loss": 0.4825,
      "step": 18957
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5427321195602417,
      "learning_rate": 7.403612907148081e-06,
      "loss": 0.4012,
      "step": 18958
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.564131736755371,
      "learning_rate": 7.4026294356797094e-06,
      "loss": 0.3938,
      "step": 18959
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5697822570800781,
      "learning_rate": 7.401645991150228e-06,
      "loss": 0.4077,
      "step": 18960
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.619175910949707,
      "learning_rate": 7.40066257356983e-06,
      "loss": 0.4672,
      "step": 18961
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5546233654022217,
      "learning_rate": 7.399679182948717e-06,
      "loss": 0.3911,
      "step": 18962
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6830980777740479,
      "learning_rate": 7.39869581929709e-06,
      "loss": 1.0298,
      "step": 18963
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4128918647766113,
      "learning_rate": 7.39771248262515e-06,
      "loss": 0.9673,
      "step": 18964
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5503458976745605,
      "learning_rate": 7.396729172943089e-06,
      "loss": 0.4524,
      "step": 18965
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7448704242706299,
      "learning_rate": 7.395745890261109e-06,
      "loss": 0.4248,
      "step": 18966
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6115087270736694,
      "learning_rate": 7.39476263458941e-06,
      "loss": 0.4523,
      "step": 18967
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5438114404678345,
      "learning_rate": 7.393779405938191e-06,
      "loss": 0.3799,
      "step": 18968
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5797079801559448,
      "learning_rate": 7.3927962043176416e-06,
      "loss": 0.4535,
      "step": 18969
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5276552438735962,
      "learning_rate": 7.391813029737968e-06,
      "loss": 0.3884,
      "step": 18970
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.584387183189392,
      "learning_rate": 7.3908298822093605e-06,
      "loss": 0.4999,
      "step": 18971
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5997378826141357,
      "learning_rate": 7.389846761742026e-06,
      "loss": 0.4066,
      "step": 18972
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5667035579681396,
      "learning_rate": 7.38886366834615e-06,
      "loss": 0.4486,
      "step": 18973
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5978639125823975,
      "learning_rate": 7.387880602031931e-06,
      "loss": 0.3808,
      "step": 18974
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5916870832443237,
      "learning_rate": 7.386897562809572e-06,
      "loss": 0.463,
      "step": 18975
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.524312973022461,
      "learning_rate": 7.385914550689264e-06,
      "loss": 0.3859,
      "step": 18976
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5207058191299438,
      "learning_rate": 7.384931565681197e-06,
      "loss": 0.4504,
      "step": 18977
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.770408034324646,
      "learning_rate": 7.383948607795574e-06,
      "loss": 0.4321,
      "step": 18978
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.547742247581482,
      "learning_rate": 7.382965677042587e-06,
      "loss": 0.473,
      "step": 18979
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6282906532287598,
      "learning_rate": 7.381982773432435e-06,
      "loss": 0.4227,
      "step": 18980
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.526906967163086,
      "learning_rate": 7.380999896975304e-06,
      "loss": 0.4554,
      "step": 18981
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6191291809082031,
      "learning_rate": 7.380017047681391e-06,
      "loss": 0.4251,
      "step": 18982
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5119491815567017,
      "learning_rate": 7.3790342255608925e-06,
      "loss": 0.4811,
      "step": 18983
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5556139945983887,
      "learning_rate": 7.378051430624002e-06,
      "loss": 0.42,
      "step": 18984
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5046080350875854,
      "learning_rate": 7.377068662880906e-06,
      "loss": 0.4307,
      "step": 18985
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4943387508392334,
      "learning_rate": 7.3760859223418065e-06,
      "loss": 0.4153,
      "step": 18986
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.582539439201355,
      "learning_rate": 7.375103209016888e-06,
      "loss": 0.4612,
      "step": 18987
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4611718654632568,
      "learning_rate": 7.374120522916352e-06,
      "loss": 0.4136,
      "step": 18988
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5175261497497559,
      "learning_rate": 7.373137864050382e-06,
      "loss": 0.4092,
      "step": 18989
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5595439672470093,
      "learning_rate": 7.372155232429173e-06,
      "loss": 0.4356,
      "step": 18990
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5059099197387695,
      "learning_rate": 7.371172628062916e-06,
      "loss": 0.395,
      "step": 18991
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.488616943359375,
      "learning_rate": 7.370190050961805e-06,
      "loss": 0.3994,
      "step": 18992
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5452295541763306,
      "learning_rate": 7.369207501136024e-06,
      "loss": 0.4564,
      "step": 18993
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5513525009155273,
      "learning_rate": 7.3682249785957714e-06,
      "loss": 0.4015,
      "step": 18994
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5266973972320557,
      "learning_rate": 7.367242483351231e-06,
      "loss": 0.4309,
      "step": 18995
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5458778142929077,
      "learning_rate": 7.366260015412602e-06,
      "loss": 0.392,
      "step": 18996
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5500967502593994,
      "learning_rate": 7.3652775747900655e-06,
      "loss": 0.4346,
      "step": 18997
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5393208265304565,
      "learning_rate": 7.3642951614938105e-06,
      "loss": 0.3893,
      "step": 18998
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5397891998291016,
      "learning_rate": 7.363312775534032e-06,
      "loss": 0.4611,
      "step": 18999
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5224575996398926,
      "learning_rate": 7.362330416920918e-06,
      "loss": 0.3946,
      "step": 19000
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.978296160697937,
      "learning_rate": 7.361348085664655e-06,
      "loss": 0.4103,
      "step": 19001
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5355348587036133,
      "learning_rate": 7.360365781775427e-06,
      "loss": 0.3815,
      "step": 19002
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5621662139892578,
      "learning_rate": 7.359383505263431e-06,
      "loss": 0.4265,
      "step": 19003
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5509798526763916,
      "learning_rate": 7.358401256138852e-06,
      "loss": 0.4207,
      "step": 19004
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6764456033706665,
      "learning_rate": 7.357419034411873e-06,
      "loss": 0.5027,
      "step": 19005
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6897484064102173,
      "learning_rate": 7.356436840092684e-06,
      "loss": 0.4577,
      "step": 19006
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6017054319381714,
      "learning_rate": 7.355454673191471e-06,
      "loss": 0.4648,
      "step": 19007
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5375211238861084,
      "learning_rate": 7.354472533718428e-06,
      "loss": 0.3973,
      "step": 19008
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6462299823760986,
      "learning_rate": 7.353490421683731e-06,
      "loss": 0.4717,
      "step": 19009
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4569060802459717,
      "learning_rate": 7.352508337097569e-06,
      "loss": 0.3824,
      "step": 19010
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.59465491771698,
      "learning_rate": 7.351526279970131e-06,
      "loss": 0.4478,
      "step": 19011
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5239620208740234,
      "learning_rate": 7.3505442503116035e-06,
      "loss": 0.3922,
      "step": 19012
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5517295598983765,
      "learning_rate": 7.349562248132163e-06,
      "loss": 0.4618,
      "step": 19013
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4997408390045166,
      "learning_rate": 7.348580273442003e-06,
      "loss": 0.4036,
      "step": 19014
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5228878259658813,
      "learning_rate": 7.347598326251304e-06,
      "loss": 0.4336,
      "step": 19015
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4475598335266113,
      "learning_rate": 7.346616406570256e-06,
      "loss": 0.3897,
      "step": 19016
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4991933107376099,
      "learning_rate": 7.3456345144090365e-06,
      "loss": 0.4186,
      "step": 19017
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5710147619247437,
      "learning_rate": 7.344652649777829e-06,
      "loss": 0.43,
      "step": 19018
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5017614364624023,
      "learning_rate": 7.343670812686823e-06,
      "loss": 0.4527,
      "step": 19019
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5530606508255005,
      "learning_rate": 7.3426890031462e-06,
      "loss": 0.4032,
      "step": 19020
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5414286851882935,
      "learning_rate": 7.3417072211661355e-06,
      "loss": 0.4474,
      "step": 19021
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4907461404800415,
      "learning_rate": 7.340725466756822e-06,
      "loss": 0.412,
      "step": 19022
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5140732526779175,
      "learning_rate": 7.339743739928433e-06,
      "loss": 0.4364,
      "step": 19023
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5084506273269653,
      "learning_rate": 7.338762040691163e-06,
      "loss": 0.3851,
      "step": 19024
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6208322048187256,
      "learning_rate": 7.337780369055182e-06,
      "loss": 0.4639,
      "step": 19025
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4527182579040527,
      "learning_rate": 7.336798725030674e-06,
      "loss": 0.3799,
      "step": 19026
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5909837484359741,
      "learning_rate": 7.335817108627824e-06,
      "loss": 0.4904,
      "step": 19027
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.553481101989746,
      "learning_rate": 7.334835519856813e-06,
      "loss": 0.4081,
      "step": 19028
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4784815311431885,
      "learning_rate": 7.333853958727816e-06,
      "loss": 0.4512,
      "step": 19029
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4892032146453857,
      "learning_rate": 7.332872425251017e-06,
      "loss": 0.4081,
      "step": 19030
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6333279609680176,
      "learning_rate": 7.331890919436595e-06,
      "loss": 0.4928,
      "step": 19031
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.575104832649231,
      "learning_rate": 7.330909441294734e-06,
      "loss": 0.3855,
      "step": 19032
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5710150003433228,
      "learning_rate": 7.32992799083561e-06,
      "loss": 0.4607,
      "step": 19033
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.524170994758606,
      "learning_rate": 7.3289465680694e-06,
      "loss": 0.419,
      "step": 19034
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5569164752960205,
      "learning_rate": 7.327965173006286e-06,
      "loss": 0.4555,
      "step": 19035
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5329880714416504,
      "learning_rate": 7.3269838056564445e-06,
      "loss": 0.407,
      "step": 19036
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5482664108276367,
      "learning_rate": 7.32600246603006e-06,
      "loss": 0.4107,
      "step": 19037
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5274808406829834,
      "learning_rate": 7.325021154137304e-06,
      "loss": 0.4048,
      "step": 19038
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7099171876907349,
      "learning_rate": 7.324039869988353e-06,
      "loss": 0.5009,
      "step": 19039
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5982146263122559,
      "learning_rate": 7.32305861359339e-06,
      "loss": 0.4571,
      "step": 19040
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6570401191711426,
      "learning_rate": 7.322077384962593e-06,
      "loss": 0.4689,
      "step": 19041
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4633451700210571,
      "learning_rate": 7.321096184106132e-06,
      "loss": 0.3971,
      "step": 19042
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5892183780670166,
      "learning_rate": 7.320115011034185e-06,
      "loss": 0.4454,
      "step": 19043
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5604400634765625,
      "learning_rate": 7.319133865756935e-06,
      "loss": 0.3996,
      "step": 19044
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5474554300308228,
      "learning_rate": 7.3181527482845546e-06,
      "loss": 0.4484,
      "step": 19045
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5894639492034912,
      "learning_rate": 7.317171658627214e-06,
      "loss": 0.3883,
      "step": 19046
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5669629573822021,
      "learning_rate": 7.316190596795096e-06,
      "loss": 0.4272,
      "step": 19047
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4771063327789307,
      "learning_rate": 7.315209562798371e-06,
      "loss": 0.3946,
      "step": 19048
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4976835250854492,
      "learning_rate": 7.3142285566472204e-06,
      "loss": 0.4249,
      "step": 19049
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5185199975967407,
      "learning_rate": 7.313247578351812e-06,
      "loss": 0.4092,
      "step": 19050
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5433627367019653,
      "learning_rate": 7.312266627922319e-06,
      "loss": 0.444,
      "step": 19051
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4769443273544312,
      "learning_rate": 7.311285705368924e-06,
      "loss": 0.3812,
      "step": 19052
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.787407636642456,
      "learning_rate": 7.310304810701796e-06,
      "loss": 0.466,
      "step": 19053
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5246704816818237,
      "learning_rate": 7.309323943931105e-06,
      "loss": 0.3808,
      "step": 19054
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5683073997497559,
      "learning_rate": 7.308343105067028e-06,
      "loss": 0.4539,
      "step": 19055
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.487522006034851,
      "learning_rate": 7.307362294119736e-06,
      "loss": 0.368,
      "step": 19056
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5051223039627075,
      "learning_rate": 7.306381511099407e-06,
      "loss": 0.4575,
      "step": 19057
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5139763355255127,
      "learning_rate": 7.305400756016207e-06,
      "loss": 0.371,
      "step": 19058
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8633406162261963,
      "learning_rate": 7.304420028880308e-06,
      "loss": 1.0447,
      "step": 19059
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5952856540679932,
      "learning_rate": 7.3034393297018855e-06,
      "loss": 1.1106,
      "step": 19060
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.701798677444458,
      "learning_rate": 7.302458658491112e-06,
      "loss": 0.4572,
      "step": 19061
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4721862077713013,
      "learning_rate": 7.30147801525815e-06,
      "loss": 0.3885,
      "step": 19062
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5306007862091064,
      "learning_rate": 7.30049740001318e-06,
      "loss": 0.4231,
      "step": 19063
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5302951335906982,
      "learning_rate": 7.2995168127663655e-06,
      "loss": 0.405,
      "step": 19064
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5511225461959839,
      "learning_rate": 7.298536253527883e-06,
      "loss": 0.4632,
      "step": 19065
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4735794067382812,
      "learning_rate": 7.297555722307901e-06,
      "loss": 0.4058,
      "step": 19066
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5389097929000854,
      "learning_rate": 7.296575219116582e-06,
      "loss": 0.4502,
      "step": 19067
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5617518424987793,
      "learning_rate": 7.295594743964105e-06,
      "loss": 0.4045,
      "step": 19068
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9659922122955322,
      "learning_rate": 7.294614296860636e-06,
      "loss": 0.9929,
      "step": 19069
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4100724458694458,
      "learning_rate": 7.29363387781634e-06,
      "loss": 0.9863,
      "step": 19070
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5751805305480957,
      "learning_rate": 7.292653486841388e-06,
      "loss": 0.4278,
      "step": 19071
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5779000520706177,
      "learning_rate": 7.291673123945948e-06,
      "loss": 0.3963,
      "step": 19072
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.628556728363037,
      "learning_rate": 7.290692789140194e-06,
      "loss": 0.4866,
      "step": 19073
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6257765293121338,
      "learning_rate": 7.289712482434285e-06,
      "loss": 0.4035,
      "step": 19074
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.567487359046936,
      "learning_rate": 7.288732203838389e-06,
      "loss": 0.4796,
      "step": 19075
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.489286184310913,
      "learning_rate": 7.287751953362679e-06,
      "loss": 0.3999,
      "step": 19076
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5101337432861328,
      "learning_rate": 7.286771731017319e-06,
      "loss": 0.4977,
      "step": 19077
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.539797306060791,
      "learning_rate": 7.285791536812472e-06,
      "loss": 0.4196,
      "step": 19078
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5239003896713257,
      "learning_rate": 7.284811370758307e-06,
      "loss": 0.4301,
      "step": 19079
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5996898412704468,
      "learning_rate": 7.283831232864989e-06,
      "loss": 0.4114,
      "step": 19080
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.679131031036377,
      "learning_rate": 7.282851123142691e-06,
      "loss": 0.4523,
      "step": 19081
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5182263851165771,
      "learning_rate": 7.2818710416015645e-06,
      "loss": 0.4021,
      "step": 19082
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5462546348571777,
      "learning_rate": 7.280890988251786e-06,
      "loss": 0.4188,
      "step": 19083
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4780265092849731,
      "learning_rate": 7.279910963103511e-06,
      "loss": 0.395,
      "step": 19084
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5640593767166138,
      "learning_rate": 7.278930966166915e-06,
      "loss": 0.438,
      "step": 19085
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5689828395843506,
      "learning_rate": 7.277950997452154e-06,
      "loss": 0.4345,
      "step": 19086
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5697802305221558,
      "learning_rate": 7.276971056969392e-06,
      "loss": 0.4544,
      "step": 19087
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.509652853012085,
      "learning_rate": 7.275991144728796e-06,
      "loss": 0.4296,
      "step": 19088
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5354468822479248,
      "learning_rate": 7.27501126074053e-06,
      "loss": 0.4513,
      "step": 19089
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6223042011260986,
      "learning_rate": 7.27403140501475e-06,
      "loss": 0.4176,
      "step": 19090
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5345770120620728,
      "learning_rate": 7.273051577561628e-06,
      "loss": 0.4322,
      "step": 19091
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5706334114074707,
      "learning_rate": 7.272071778391317e-06,
      "loss": 0.4528,
      "step": 19092
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5242676734924316,
      "learning_rate": 7.271092007513988e-06,
      "loss": 0.4493,
      "step": 19093
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6716612577438354,
      "learning_rate": 7.270112264939797e-06,
      "loss": 0.4112,
      "step": 19094
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.599934697151184,
      "learning_rate": 7.2691325506789055e-06,
      "loss": 0.4146,
      "step": 19095
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5804542303085327,
      "learning_rate": 7.268152864741478e-06,
      "loss": 0.3968,
      "step": 19096
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0913596153259277,
      "learning_rate": 7.267173207137676e-06,
      "loss": 0.4648,
      "step": 19097
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.526597499847412,
      "learning_rate": 7.266193577877654e-06,
      "loss": 0.3972,
      "step": 19098
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.600327968597412,
      "learning_rate": 7.265213976971578e-06,
      "loss": 0.5021,
      "step": 19099
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4768728017807007,
      "learning_rate": 7.264234404429605e-06,
      "loss": 0.3759,
      "step": 19100
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6423078775405884,
      "learning_rate": 7.263254860261898e-06,
      "loss": 0.4579,
      "step": 19101
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6672022342681885,
      "learning_rate": 7.262275344478613e-06,
      "loss": 0.4598,
      "step": 19102
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5935921669006348,
      "learning_rate": 7.261295857089909e-06,
      "loss": 0.4778,
      "step": 19103
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.462393879890442,
      "learning_rate": 7.260316398105949e-06,
      "loss": 0.3862,
      "step": 19104
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5159924030303955,
      "learning_rate": 7.259336967536889e-06,
      "loss": 0.4787,
      "step": 19105
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5344613790512085,
      "learning_rate": 7.258357565392886e-06,
      "loss": 0.3967,
      "step": 19106
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.59304678440094,
      "learning_rate": 7.257378191684099e-06,
      "loss": 0.5038,
      "step": 19107
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4667927026748657,
      "learning_rate": 7.256398846420685e-06,
      "loss": 0.3977,
      "step": 19108
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6095014810562134,
      "learning_rate": 7.255419529612805e-06,
      "loss": 0.4577,
      "step": 19109
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.504568338394165,
      "learning_rate": 7.254440241270613e-06,
      "loss": 0.3912,
      "step": 19110
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5832241773605347,
      "learning_rate": 7.253460981404261e-06,
      "loss": 0.4817,
      "step": 19111
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4731924533843994,
      "learning_rate": 7.252481750023916e-06,
      "loss": 0.39,
      "step": 19112
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5588064193725586,
      "learning_rate": 7.251502547139728e-06,
      "loss": 0.4558,
      "step": 19113
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5405250787734985,
      "learning_rate": 7.250523372761852e-06,
      "loss": 0.3897,
      "step": 19114
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5875298976898193,
      "learning_rate": 7.249544226900447e-06,
      "loss": 0.4615,
      "step": 19115
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5389599800109863,
      "learning_rate": 7.248565109565663e-06,
      "loss": 0.4126,
      "step": 19116
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6377681493759155,
      "learning_rate": 7.2475860207676655e-06,
      "loss": 0.4846,
      "step": 19117
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.575454592704773,
      "learning_rate": 7.246606960516595e-06,
      "loss": 0.41,
      "step": 19118
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5868061780929565,
      "learning_rate": 7.2456279288226174e-06,
      "loss": 0.4376,
      "step": 19119
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.568477988243103,
      "learning_rate": 7.2446489256958795e-06,
      "loss": 0.3918,
      "step": 19120
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.825939655303955,
      "learning_rate": 7.243669951146542e-06,
      "loss": 0.485,
      "step": 19121
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.523767113685608,
      "learning_rate": 7.242691005184754e-06,
      "loss": 0.408,
      "step": 19122
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.657045602798462,
      "learning_rate": 7.241712087820666e-06,
      "loss": 0.4887,
      "step": 19123
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4912703037261963,
      "learning_rate": 7.240733199064437e-06,
      "loss": 0.3847,
      "step": 19124
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5798145532608032,
      "learning_rate": 7.239754338926219e-06,
      "loss": 0.4451,
      "step": 19125
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7494909763336182,
      "learning_rate": 7.238775507416158e-06,
      "loss": 0.4314,
      "step": 19126
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5677646398544312,
      "learning_rate": 7.237796704544413e-06,
      "loss": 0.4468,
      "step": 19127
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5423239469528198,
      "learning_rate": 7.236817930321132e-06,
      "loss": 0.437,
      "step": 19128
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.488348126411438,
      "learning_rate": 7.23583918475647e-06,
      "loss": 0.4207,
      "step": 19129
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4803439378738403,
      "learning_rate": 7.2348604678605734e-06,
      "loss": 0.3888,
      "step": 19130
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6001380681991577,
      "learning_rate": 7.233881779643595e-06,
      "loss": 0.454,
      "step": 19131
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6306215524673462,
      "learning_rate": 7.232903120115686e-06,
      "loss": 0.3948,
      "step": 19132
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5544556379318237,
      "learning_rate": 7.231924489287001e-06,
      "loss": 0.4507,
      "step": 19133
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.631542444229126,
      "learning_rate": 7.230945887167679e-06,
      "loss": 0.4146,
      "step": 19134
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6409207582473755,
      "learning_rate": 7.229967313767879e-06,
      "loss": 0.4831,
      "step": 19135
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5726659297943115,
      "learning_rate": 7.228988769097746e-06,
      "loss": 0.3886,
      "step": 19136
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6108250617980957,
      "learning_rate": 7.228010253167435e-06,
      "loss": 0.4755,
      "step": 19137
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6565731763839722,
      "learning_rate": 7.227031765987087e-06,
      "loss": 0.4075,
      "step": 19138
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.462558627128601,
      "learning_rate": 7.226053307566852e-06,
      "loss": 0.4214,
      "step": 19139
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5377556085586548,
      "learning_rate": 7.225074877916881e-06,
      "loss": 0.3883,
      "step": 19140
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.544076919555664,
      "learning_rate": 7.224096477047324e-06,
      "loss": 0.4397,
      "step": 19141
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.53434157371521,
      "learning_rate": 7.223118104968322e-06,
      "loss": 0.417,
      "step": 19142
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5730005502700806,
      "learning_rate": 7.2221397616900265e-06,
      "loss": 0.47,
      "step": 19143
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4785394668579102,
      "learning_rate": 7.2211614472225825e-06,
      "loss": 0.3984,
      "step": 19144
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.625781536102295,
      "learning_rate": 7.220183161576142e-06,
      "loss": 0.4425,
      "step": 19145
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.498403787612915,
      "learning_rate": 7.219204904760845e-06,
      "loss": 0.4049,
      "step": 19146
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.566192865371704,
      "learning_rate": 7.218226676786837e-06,
      "loss": 0.4529,
      "step": 19147
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5101556777954102,
      "learning_rate": 7.217248477664269e-06,
      "loss": 0.4023,
      "step": 19148
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5905643701553345,
      "learning_rate": 7.216270307403288e-06,
      "loss": 0.4592,
      "step": 19149
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6696854829788208,
      "learning_rate": 7.215292166014029e-06,
      "loss": 0.4247,
      "step": 19150
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9351593255996704,
      "learning_rate": 7.214314053506646e-06,
      "loss": 0.9849,
      "step": 19151
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5290571451187134,
      "learning_rate": 7.21333596989128e-06,
      "loss": 1.065,
      "step": 19152
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.585889220237732,
      "learning_rate": 7.212357915178079e-06,
      "loss": 0.4709,
      "step": 19153
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.64478600025177,
      "learning_rate": 7.2113798893771815e-06,
      "loss": 0.4186,
      "step": 19154
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.514949083328247,
      "learning_rate": 7.210401892498733e-06,
      "loss": 0.3908,
      "step": 19155
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4760202169418335,
      "learning_rate": 7.209423924552879e-06,
      "loss": 0.3847,
      "step": 19156
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.546170949935913,
      "learning_rate": 7.208445985549764e-06,
      "loss": 0.4402,
      "step": 19157
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5525691509246826,
      "learning_rate": 7.2074680754995245e-06,
      "loss": 0.4093,
      "step": 19158
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5346944332122803,
      "learning_rate": 7.206490194412305e-06,
      "loss": 0.4959,
      "step": 19159
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4928745031356812,
      "learning_rate": 7.205512342298251e-06,
      "loss": 0.3867,
      "step": 19160
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6049901247024536,
      "learning_rate": 7.204534519167506e-06,
      "loss": 0.4953,
      "step": 19161
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5249541997909546,
      "learning_rate": 7.203556725030203e-06,
      "loss": 0.3772,
      "step": 19162
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6790399551391602,
      "learning_rate": 7.202578959896491e-06,
      "loss": 0.4475,
      "step": 19163
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5128331184387207,
      "learning_rate": 7.201601223776506e-06,
      "loss": 0.4016,
      "step": 19164
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6827359199523926,
      "learning_rate": 7.2006235166803965e-06,
      "loss": 0.4812,
      "step": 19165
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.572587251663208,
      "learning_rate": 7.199645838618294e-06,
      "loss": 0.406,
      "step": 19166
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7607102394104004,
      "learning_rate": 7.198668189600342e-06,
      "loss": 0.4896,
      "step": 19167
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.628546953201294,
      "learning_rate": 7.1976905696366805e-06,
      "loss": 0.414,
      "step": 19168
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.544435739517212,
      "learning_rate": 7.196712978737453e-06,
      "loss": 0.4505,
      "step": 19169
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.519234538078308,
      "learning_rate": 7.195735416912788e-06,
      "loss": 0.3763,
      "step": 19170
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5790205001831055,
      "learning_rate": 7.194757884172834e-06,
      "loss": 0.4914,
      "step": 19171
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7443273067474365,
      "learning_rate": 7.193780380527725e-06,
      "loss": 0.403,
      "step": 19172
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5269891023635864,
      "learning_rate": 7.192802905987605e-06,
      "loss": 0.4458,
      "step": 19173
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6628650426864624,
      "learning_rate": 7.191825460562605e-06,
      "loss": 0.4091,
      "step": 19174
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5908600091934204,
      "learning_rate": 7.190848044262864e-06,
      "loss": 0.4385,
      "step": 19175
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.558013677597046,
      "learning_rate": 7.189870657098524e-06,
      "loss": 0.4086,
      "step": 19176
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6332353353500366,
      "learning_rate": 7.188893299079718e-06,
      "loss": 0.4357,
      "step": 19177
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.60708749294281,
      "learning_rate": 7.187915970216582e-06,
      "loss": 0.4073,
      "step": 19178
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4931505918502808,
      "learning_rate": 7.1869386705192545e-06,
      "loss": 0.4382,
      "step": 19179
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4976493120193481,
      "learning_rate": 7.185961399997871e-06,
      "loss": 0.3848,
      "step": 19180
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.571587324142456,
      "learning_rate": 7.184984158662571e-06,
      "loss": 0.4619,
      "step": 19181
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5657058954238892,
      "learning_rate": 7.184006946523486e-06,
      "loss": 0.3959,
      "step": 19182
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6444472074508667,
      "learning_rate": 7.183029763590747e-06,
      "loss": 0.4432,
      "step": 19183
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.778502106666565,
      "learning_rate": 7.182052609874497e-06,
      "loss": 0.4546,
      "step": 19184
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5339356660842896,
      "learning_rate": 7.181075485384871e-06,
      "loss": 0.4421,
      "step": 19185
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4628742933273315,
      "learning_rate": 7.180098390131994e-06,
      "loss": 0.4044,
      "step": 19186
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5662727355957031,
      "learning_rate": 7.179121324126009e-06,
      "loss": 0.4703,
      "step": 19187
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4878973960876465,
      "learning_rate": 7.178144287377044e-06,
      "loss": 0.3859,
      "step": 19188
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5320498943328857,
      "learning_rate": 7.177167279895241e-06,
      "loss": 0.4476,
      "step": 19189
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5251761674880981,
      "learning_rate": 7.176190301690724e-06,
      "loss": 0.3986,
      "step": 19190
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6149476766586304,
      "learning_rate": 7.175213352773627e-06,
      "loss": 0.4854,
      "step": 19191
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.587715983390808,
      "learning_rate": 7.174236433154087e-06,
      "loss": 0.3934,
      "step": 19192
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5996589660644531,
      "learning_rate": 7.173259542842235e-06,
      "loss": 0.4787,
      "step": 19193
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5367627143859863,
      "learning_rate": 7.172282681848199e-06,
      "loss": 0.401,
      "step": 19194
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6236789226531982,
      "learning_rate": 7.171305850182113e-06,
      "loss": 0.4514,
      "step": 19195
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6485503911972046,
      "learning_rate": 7.17032904785411e-06,
      "loss": 0.4141,
      "step": 19196
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5714789628982544,
      "learning_rate": 7.1693522748743225e-06,
      "loss": 0.411,
      "step": 19197
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.493470311164856,
      "learning_rate": 7.168375531252873e-06,
      "loss": 0.3875,
      "step": 19198
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5569112300872803,
      "learning_rate": 7.1673988169999e-06,
      "loss": 0.4594,
      "step": 19199
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6784749031066895,
      "learning_rate": 7.166422132125529e-06,
      "loss": 0.3999,
      "step": 19200
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.556514024734497,
      "learning_rate": 7.165445476639896e-06,
      "loss": 0.4106,
      "step": 19201
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5286073684692383,
      "learning_rate": 7.164468850553124e-06,
      "loss": 0.3711,
      "step": 19202
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4805691242218018,
      "learning_rate": 7.163492253875341e-06,
      "loss": 0.4096,
      "step": 19203
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5044066905975342,
      "learning_rate": 7.1625156866166815e-06,
      "loss": 0.4161,
      "step": 19204
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.593511939048767,
      "learning_rate": 7.161539148787274e-06,
      "loss": 0.4342,
      "step": 19205
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5692341327667236,
      "learning_rate": 7.16056264039724e-06,
      "loss": 0.3987,
      "step": 19206
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6171423196792603,
      "learning_rate": 7.159586161456715e-06,
      "loss": 0.4694,
      "step": 19207
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5884648561477661,
      "learning_rate": 7.1586097119758215e-06,
      "loss": 0.4367,
      "step": 19208
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.514862060546875,
      "learning_rate": 7.157633291964692e-06,
      "loss": 0.4575,
      "step": 19209
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4967442750930786,
      "learning_rate": 7.156656901433449e-06,
      "loss": 0.3939,
      "step": 19210
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6025638580322266,
      "learning_rate": 7.155680540392218e-06,
      "loss": 0.3984,
      "step": 19211
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.487644076347351,
      "learning_rate": 7.154704208851131e-06,
      "loss": 0.395,
      "step": 19212
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.580609679222107,
      "learning_rate": 7.153727906820312e-06,
      "loss": 0.4556,
      "step": 19213
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.514778733253479,
      "learning_rate": 7.152751634309883e-06,
      "loss": 0.4097,
      "step": 19214
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5311716794967651,
      "learning_rate": 7.151775391329974e-06,
      "loss": 0.3975,
      "step": 19215
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5298075675964355,
      "learning_rate": 7.150799177890707e-06,
      "loss": 0.4172,
      "step": 19216
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5682622194290161,
      "learning_rate": 7.149822994002213e-06,
      "loss": 0.4805,
      "step": 19217
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.655288577079773,
      "learning_rate": 7.14884683967461e-06,
      "loss": 0.3982,
      "step": 19218
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6051816940307617,
      "learning_rate": 7.147870714918021e-06,
      "loss": 0.4855,
      "step": 19219
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.661693811416626,
      "learning_rate": 7.146894619742577e-06,
      "loss": 0.3993,
      "step": 19220
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5497256517410278,
      "learning_rate": 7.145918554158401e-06,
      "loss": 0.4717,
      "step": 19221
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5436265468597412,
      "learning_rate": 7.144942518175608e-06,
      "loss": 0.4329,
      "step": 19222
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.666059970855713,
      "learning_rate": 7.14396651180433e-06,
      "loss": 0.4274,
      "step": 19223
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5497004985809326,
      "learning_rate": 7.142990535054682e-06,
      "loss": 0.3951,
      "step": 19224
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4784857034683228,
      "learning_rate": 7.1420145879367965e-06,
      "loss": 0.4337,
      "step": 19225
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.662999153137207,
      "learning_rate": 7.141038670460787e-06,
      "loss": 0.4364,
      "step": 19226
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6241470575332642,
      "learning_rate": 7.140062782636778e-06,
      "loss": 0.5015,
      "step": 19227
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.626557469367981,
      "learning_rate": 7.139086924474892e-06,
      "loss": 0.4366,
      "step": 19228
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6105425357818604,
      "learning_rate": 7.1381110959852505e-06,
      "loss": 0.5022,
      "step": 19229
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4802262783050537,
      "learning_rate": 7.13713529717797e-06,
      "loss": 0.3841,
      "step": 19230
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5816888809204102,
      "learning_rate": 7.136159528063177e-06,
      "loss": 0.4486,
      "step": 19231
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5526946783065796,
      "learning_rate": 7.1351837886509854e-06,
      "loss": 0.4176,
      "step": 19232
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5163928270339966,
      "learning_rate": 7.134208078951528e-06,
      "loss": 0.4395,
      "step": 19233
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5301793813705444,
      "learning_rate": 7.1332323989749076e-06,
      "loss": 0.401,
      "step": 19234
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6148681640625,
      "learning_rate": 7.132256748731253e-06,
      "loss": 0.4412,
      "step": 19235
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5581223964691162,
      "learning_rate": 7.13128112823068e-06,
      "loss": 0.41,
      "step": 19236
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5131303071975708,
      "learning_rate": 7.130305537483313e-06,
      "loss": 0.4263,
      "step": 19237
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5610946416854858,
      "learning_rate": 7.129329976499267e-06,
      "loss": 0.4058,
      "step": 19238
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5761898756027222,
      "learning_rate": 7.1283544452886545e-06,
      "loss": 0.4532,
      "step": 19239
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4811623096466064,
      "learning_rate": 7.1273789438616e-06,
      "loss": 0.3868,
      "step": 19240
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6778823137283325,
      "learning_rate": 7.126403472228223e-06,
      "loss": 0.4939,
      "step": 19241
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5549135208129883,
      "learning_rate": 7.125428030398633e-06,
      "loss": 0.3865,
      "step": 19242
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.553253173828125,
      "learning_rate": 7.124452618382953e-06,
      "loss": 0.4446,
      "step": 19243
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.610164761543274,
      "learning_rate": 7.123477236191294e-06,
      "loss": 0.4115,
      "step": 19244
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6367470026016235,
      "learning_rate": 7.122501883833782e-06,
      "loss": 0.4491,
      "step": 19245
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5843260288238525,
      "learning_rate": 7.1215265613205245e-06,
      "loss": 0.4027,
      "step": 19246
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5835397243499756,
      "learning_rate": 7.120551268661636e-06,
      "loss": 0.4852,
      "step": 19247
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6052440404891968,
      "learning_rate": 7.1195760058672394e-06,
      "loss": 0.4007,
      "step": 19248
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5952016115188599,
      "learning_rate": 7.118600772947446e-06,
      "loss": 0.483,
      "step": 19249
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6152082681655884,
      "learning_rate": 7.1176255699123664e-06,
      "loss": 0.4241,
      "step": 19250
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.540023684501648,
      "learning_rate": 7.116650396772121e-06,
      "loss": 0.4513,
      "step": 19251
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5103609561920166,
      "learning_rate": 7.11567525353682e-06,
      "loss": 0.4005,
      "step": 19252
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.642532229423523,
      "learning_rate": 7.114700140216583e-06,
      "loss": 0.4719,
      "step": 19253
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.518472671508789,
      "learning_rate": 7.1137250568215165e-06,
      "loss": 0.3842,
      "step": 19254
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5021474361419678,
      "learning_rate": 7.112750003361735e-06,
      "loss": 0.4315,
      "step": 19255
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.481745958328247,
      "learning_rate": 7.111774979847354e-06,
      "loss": 0.3969,
      "step": 19256
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6414610147476196,
      "learning_rate": 7.110799986288489e-06,
      "loss": 0.4468,
      "step": 19257
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5248421430587769,
      "learning_rate": 7.109825022695243e-06,
      "loss": 0.4061,
      "step": 19258
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6066827774047852,
      "learning_rate": 7.108850089077736e-06,
      "loss": 0.4422,
      "step": 19259
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5632834434509277,
      "learning_rate": 7.107875185446074e-06,
      "loss": 0.3961,
      "step": 19260
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.494071125984192,
      "learning_rate": 7.1069003118103755e-06,
      "loss": 0.4119,
      "step": 19261
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6341612339019775,
      "learning_rate": 7.105925468180745e-06,
      "loss": 0.4267,
      "step": 19262
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5274180173873901,
      "learning_rate": 7.1049506545672915e-06,
      "loss": 0.4891,
      "step": 19263
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6432979106903076,
      "learning_rate": 7.103975870980133e-06,
      "loss": 0.4207,
      "step": 19264
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5135031938552856,
      "learning_rate": 7.103001117429378e-06,
      "loss": 0.4184,
      "step": 19265
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6528042554855347,
      "learning_rate": 7.102026393925129e-06,
      "loss": 0.3943,
      "step": 19266
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.557978630065918,
      "learning_rate": 7.1010517004775015e-06,
      "loss": 0.4151,
      "step": 19267
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5095880031585693,
      "learning_rate": 7.100077037096603e-06,
      "loss": 0.3919,
      "step": 19268
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6041245460510254,
      "learning_rate": 7.099102403792549e-06,
      "loss": 0.4649,
      "step": 19269
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6689085960388184,
      "learning_rate": 7.098127800575435e-06,
      "loss": 0.4466,
      "step": 19270
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5486469268798828,
      "learning_rate": 7.097153227455379e-06,
      "loss": 0.5099,
      "step": 19271
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.501637578010559,
      "learning_rate": 7.096178684442483e-06,
      "loss": 0.3851,
      "step": 19272
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5429655313491821,
      "learning_rate": 7.095204171546857e-06,
      "loss": 0.4332,
      "step": 19273
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5960661172866821,
      "learning_rate": 7.0942296887786154e-06,
      "loss": 0.4267,
      "step": 19274
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.549437403678894,
      "learning_rate": 7.093255236147853e-06,
      "loss": 0.4749,
      "step": 19275
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5733308792114258,
      "learning_rate": 7.092280813664683e-06,
      "loss": 0.4065,
      "step": 19276
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6218715906143188,
      "learning_rate": 7.09130642133921e-06,
      "loss": 0.4338,
      "step": 19277
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6968281269073486,
      "learning_rate": 7.090332059181544e-06,
      "loss": 0.3856,
      "step": 19278
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.580783486366272,
      "learning_rate": 7.089357727201784e-06,
      "loss": 0.4691,
      "step": 19279
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4832441806793213,
      "learning_rate": 7.0883834254100394e-06,
      "loss": 0.3851,
      "step": 19280
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5913223028182983,
      "learning_rate": 7.087409153816415e-06,
      "loss": 0.4445,
      "step": 19281
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5384278297424316,
      "learning_rate": 7.086434912431017e-06,
      "loss": 0.4011,
      "step": 19282
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4697450399398804,
      "learning_rate": 7.085460701263945e-06,
      "loss": 0.4387,
      "step": 19283
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5563240051269531,
      "learning_rate": 7.084486520325306e-06,
      "loss": 0.3987,
      "step": 19284
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.579620122909546,
      "learning_rate": 7.083512369625203e-06,
      "loss": 0.465,
      "step": 19285
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5325719118118286,
      "learning_rate": 7.082538249173745e-06,
      "loss": 0.3906,
      "step": 19286
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5794726610183716,
      "learning_rate": 7.081564158981029e-06,
      "loss": 0.4635,
      "step": 19287
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5396816730499268,
      "learning_rate": 7.080590099057156e-06,
      "loss": 0.3823,
      "step": 19288
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5656791925430298,
      "learning_rate": 7.079616069412235e-06,
      "loss": 0.4319,
      "step": 19289
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5625063180923462,
      "learning_rate": 7.078642070056367e-06,
      "loss": 0.4371,
      "step": 19290
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6226568222045898,
      "learning_rate": 7.077668100999649e-06,
      "loss": 0.4482,
      "step": 19291
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6106387376785278,
      "learning_rate": 7.076694162252186e-06,
      "loss": 0.3796,
      "step": 19292
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5170502662658691,
      "learning_rate": 7.075720253824079e-06,
      "loss": 0.4041,
      "step": 19293
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7640756368637085,
      "learning_rate": 7.074746375725432e-06,
      "loss": 0.3884,
      "step": 19294
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5481559038162231,
      "learning_rate": 7.073772527966342e-06,
      "loss": 0.4437,
      "step": 19295
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.23097825050354,
      "learning_rate": 7.072798710556907e-06,
      "loss": 0.4249,
      "step": 19296
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.658718466758728,
      "learning_rate": 7.071824923507233e-06,
      "loss": 0.4596,
      "step": 19297
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6109009981155396,
      "learning_rate": 7.070851166827419e-06,
      "loss": 0.3748,
      "step": 19298
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7024779319763184,
      "learning_rate": 7.069877440527557e-06,
      "loss": 0.449,
      "step": 19299
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5087188482284546,
      "learning_rate": 7.068903744617754e-06,
      "loss": 0.3834,
      "step": 19300
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5726673603057861,
      "learning_rate": 7.067930079108105e-06,
      "loss": 0.4362,
      "step": 19301
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6078914403915405,
      "learning_rate": 7.066956444008712e-06,
      "loss": 0.4134,
      "step": 19302
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6221214532852173,
      "learning_rate": 7.06598283932967e-06,
      "loss": 0.4778,
      "step": 19303
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5540376901626587,
      "learning_rate": 7.0650092650810765e-06,
      "loss": 0.3796,
      "step": 19304
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7135272026062012,
      "learning_rate": 7.064035721273032e-06,
      "loss": 0.4364,
      "step": 19305
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5961592197418213,
      "learning_rate": 7.063062207915633e-06,
      "loss": 0.4009,
      "step": 19306
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6779311895370483,
      "learning_rate": 7.062088725018972e-06,
      "loss": 0.4241,
      "step": 19307
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5661453008651733,
      "learning_rate": 7.061115272593151e-06,
      "loss": 0.3776,
      "step": 19308
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4685392379760742,
      "learning_rate": 7.060141850648261e-06,
      "loss": 0.4043,
      "step": 19309
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5623176097869873,
      "learning_rate": 7.059168459194408e-06,
      "loss": 0.4034,
      "step": 19310
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5371073484420776,
      "learning_rate": 7.058195098241675e-06,
      "loss": 0.4961,
      "step": 19311
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4871095418930054,
      "learning_rate": 7.057221767800164e-06,
      "loss": 0.4211,
      "step": 19312
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5908303260803223,
      "learning_rate": 7.056248467879967e-06,
      "loss": 0.4856,
      "step": 19313
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4914968013763428,
      "learning_rate": 7.055275198491186e-06,
      "loss": 0.4193,
      "step": 19314
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5597165822982788,
      "learning_rate": 7.054301959643908e-06,
      "loss": 0.4954,
      "step": 19315
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5627537965774536,
      "learning_rate": 7.053328751348225e-06,
      "loss": 0.4276,
      "step": 19316
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9079614877700806,
      "learning_rate": 7.052355573614238e-06,
      "loss": 0.4696,
      "step": 19317
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5340056419372559,
      "learning_rate": 7.051382426452041e-06,
      "loss": 0.3778,
      "step": 19318
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6260082721710205,
      "learning_rate": 7.050409309871716e-06,
      "loss": 0.5025,
      "step": 19319
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.516493558883667,
      "learning_rate": 7.049436223883368e-06,
      "loss": 0.3964,
      "step": 19320
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5883519649505615,
      "learning_rate": 7.04846316849708e-06,
      "loss": 0.5145,
      "step": 19321
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5372563600540161,
      "learning_rate": 7.047490143722955e-06,
      "loss": 0.379,
      "step": 19322
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5922969579696655,
      "learning_rate": 7.046517149571075e-06,
      "loss": 0.4581,
      "step": 19323
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.461645483970642,
      "learning_rate": 7.045544186051534e-06,
      "loss": 0.3895,
      "step": 19324
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.602062702178955,
      "learning_rate": 7.044571253174426e-06,
      "loss": 0.4303,
      "step": 19325
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5858811140060425,
      "learning_rate": 7.0435983509498415e-06,
      "loss": 0.4066,
      "step": 19326
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6483325958251953,
      "learning_rate": 7.042625479387866e-06,
      "loss": 0.4714,
      "step": 19327
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.531125783920288,
      "learning_rate": 7.041652638498595e-06,
      "loss": 0.4055,
      "step": 19328
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5530846118927002,
      "learning_rate": 7.040679828292115e-06,
      "loss": 0.4364,
      "step": 19329
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5106806755065918,
      "learning_rate": 7.039707048778521e-06,
      "loss": 0.4454,
      "step": 19330
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4880260229110718,
      "learning_rate": 7.038734299967898e-06,
      "loss": 0.446,
      "step": 19331
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4890828132629395,
      "learning_rate": 7.037761581870332e-06,
      "loss": 0.4185,
      "step": 19332
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5683881044387817,
      "learning_rate": 7.036788894495918e-06,
      "loss": 0.4416,
      "step": 19333
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5964947938919067,
      "learning_rate": 7.035816237854744e-06,
      "loss": 0.414,
      "step": 19334
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5278104543685913,
      "learning_rate": 7.034843611956892e-06,
      "loss": 0.477,
      "step": 19335
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5380092859268188,
      "learning_rate": 7.033871016812455e-06,
      "loss": 0.3847,
      "step": 19336
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9510589838027954,
      "learning_rate": 7.032898452431515e-06,
      "loss": 0.5174,
      "step": 19337
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.466255784034729,
      "learning_rate": 7.031925918824167e-06,
      "loss": 0.3835,
      "step": 19338
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4844900369644165,
      "learning_rate": 7.030953416000492e-06,
      "loss": 0.4441,
      "step": 19339
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5504595041275024,
      "learning_rate": 7.0299809439705766e-06,
      "loss": 0.4028,
      "step": 19340
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5814770460128784,
      "learning_rate": 7.0290085027445094e-06,
      "loss": 0.451,
      "step": 19341
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5351183414459229,
      "learning_rate": 7.028036092332378e-06,
      "loss": 0.4037,
      "step": 19342
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5104807615280151,
      "learning_rate": 7.02706371274426e-06,
      "loss": 0.4597,
      "step": 19343
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5711642503738403,
      "learning_rate": 7.026091363990246e-06,
      "loss": 0.3971,
      "step": 19344
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5140459537506104,
      "learning_rate": 7.0251190460804196e-06,
      "loss": 0.4248,
      "step": 19345
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.442218542098999,
      "learning_rate": 7.024146759024869e-06,
      "loss": 0.3848,
      "step": 19346
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6837208271026611,
      "learning_rate": 7.023174502833674e-06,
      "loss": 0.4298,
      "step": 19347
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.674617052078247,
      "learning_rate": 7.022202277516916e-06,
      "loss": 0.4654,
      "step": 19348
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.644362449645996,
      "learning_rate": 7.021230083084684e-06,
      "loss": 0.4307,
      "step": 19349
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4820071458816528,
      "learning_rate": 7.0202579195470646e-06,
      "loss": 0.3759,
      "step": 19350
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.534355640411377,
      "learning_rate": 7.0192857869141296e-06,
      "loss": 0.4091,
      "step": 19351
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5450931787490845,
      "learning_rate": 7.018313685195968e-06,
      "loss": 0.4272,
      "step": 19352
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5878227949142456,
      "learning_rate": 7.017341614402662e-06,
      "loss": 0.4936,
      "step": 19353
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4995471239089966,
      "learning_rate": 7.016369574544295e-06,
      "loss": 0.398,
      "step": 19354
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5662320852279663,
      "learning_rate": 7.015397565630944e-06,
      "loss": 0.4707,
      "step": 19355
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.564435601234436,
      "learning_rate": 7.014425587672693e-06,
      "loss": 0.4024,
      "step": 19356
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5904768705368042,
      "learning_rate": 7.013453640679621e-06,
      "loss": 0.9637,
      "step": 19357
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4907960891723633,
      "learning_rate": 7.012481724661817e-06,
      "loss": 1.0051,
      "step": 19358
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6433560848236084,
      "learning_rate": 7.011509839629351e-06,
      "loss": 0.4638,
      "step": 19359
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5983943939208984,
      "learning_rate": 7.010537985592304e-06,
      "loss": 0.4365,
      "step": 19360
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7253546714782715,
      "learning_rate": 7.009566162560762e-06,
      "loss": 0.4361,
      "step": 19361
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5362850427627563,
      "learning_rate": 7.008594370544801e-06,
      "loss": 0.4262,
      "step": 19362
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.586828589439392,
      "learning_rate": 7.007622609554497e-06,
      "loss": 0.4656,
      "step": 19363
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4710204601287842,
      "learning_rate": 7.006650879599934e-06,
      "loss": 0.447,
      "step": 19364
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.543744444847107,
      "learning_rate": 7.005679180691185e-06,
      "loss": 0.4185,
      "step": 19365
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4873754978179932,
      "learning_rate": 7.004707512838336e-06,
      "loss": 0.4054,
      "step": 19366
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5943738222122192,
      "learning_rate": 7.003735876051459e-06,
      "loss": 0.4628,
      "step": 19367
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.533322811126709,
      "learning_rate": 7.002764270340627e-06,
      "loss": 0.3669,
      "step": 19368
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6448745727539062,
      "learning_rate": 7.0017926957159265e-06,
      "loss": 0.474,
      "step": 19369
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.662801742553711,
      "learning_rate": 7.000821152187431e-06,
      "loss": 0.3972,
      "step": 19370
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7795474529266357,
      "learning_rate": 6.999849639765213e-06,
      "loss": 0.9063,
      "step": 19371
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6018511056900024,
      "learning_rate": 6.998878158459354e-06,
      "loss": 1.079,
      "step": 19372
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5700840950012207,
      "learning_rate": 6.997906708279925e-06,
      "loss": 0.4725,
      "step": 19373
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4535647630691528,
      "learning_rate": 6.996935289237008e-06,
      "loss": 0.3821,
      "step": 19374
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7323188781738281,
      "learning_rate": 6.995963901340672e-06,
      "loss": 0.4721,
      "step": 19375
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.590888500213623,
      "learning_rate": 6.994992544600991e-06,
      "loss": 0.4009,
      "step": 19376
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4910075664520264,
      "learning_rate": 6.994021219028046e-06,
      "loss": 0.441,
      "step": 19377
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5748399496078491,
      "learning_rate": 6.99304992463191e-06,
      "loss": 0.4182,
      "step": 19378
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5661238431930542,
      "learning_rate": 6.99207866142265e-06,
      "loss": 0.4231,
      "step": 19379
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5586128234863281,
      "learning_rate": 6.991107429410346e-06,
      "loss": 0.405,
      "step": 19380
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5716032981872559,
      "learning_rate": 6.990136228605069e-06,
      "loss": 0.4353,
      "step": 19381
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6570404767990112,
      "learning_rate": 6.989165059016896e-06,
      "loss": 0.4281,
      "step": 19382
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.685745358467102,
      "learning_rate": 6.988193920655893e-06,
      "loss": 0.9885,
      "step": 19383
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4426155090332031,
      "learning_rate": 6.987222813532133e-06,
      "loss": 1.0187,
      "step": 19384
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5111727714538574,
      "learning_rate": 6.986251737655692e-06,
      "loss": 0.4114,
      "step": 19385
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6767077445983887,
      "learning_rate": 6.985280693036643e-06,
      "loss": 0.407,
      "step": 19386
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5872957706451416,
      "learning_rate": 6.984309679685049e-06,
      "loss": 0.4361,
      "step": 19387
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5249297618865967,
      "learning_rate": 6.983338697610987e-06,
      "loss": 0.4004,
      "step": 19388
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5784260034561157,
      "learning_rate": 6.982367746824528e-06,
      "loss": 0.423,
      "step": 19389
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5113303661346436,
      "learning_rate": 6.981396827335741e-06,
      "loss": 0.3974,
      "step": 19390
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5044385194778442,
      "learning_rate": 6.980425939154694e-06,
      "loss": 0.4214,
      "step": 19391
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.588812232017517,
      "learning_rate": 6.97945508229146e-06,
      "loss": 0.3932,
      "step": 19392
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6112439632415771,
      "learning_rate": 6.978484256756104e-06,
      "loss": 0.483,
      "step": 19393
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4539076089859009,
      "learning_rate": 6.9775134625587025e-06,
      "loss": 0.3962,
      "step": 19394
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5488442182540894,
      "learning_rate": 6.976542699709319e-06,
      "loss": 0.4545,
      "step": 19395
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4674479961395264,
      "learning_rate": 6.975571968218019e-06,
      "loss": 0.3959,
      "step": 19396
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5521056652069092,
      "learning_rate": 6.974601268094876e-06,
      "loss": 0.4919,
      "step": 19397
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.513490915298462,
      "learning_rate": 6.973630599349959e-06,
      "loss": 0.413,
      "step": 19398
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5784072875976562,
      "learning_rate": 6.972659961993327e-06,
      "loss": 0.4908,
      "step": 19399
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5369547605514526,
      "learning_rate": 6.971689356035053e-06,
      "loss": 0.3931,
      "step": 19400
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5607452392578125,
      "learning_rate": 6.970718781485202e-06,
      "loss": 0.4532,
      "step": 19401
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.553770899772644,
      "learning_rate": 6.9697482383538466e-06,
      "loss": 0.4028,
      "step": 19402
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5827009677886963,
      "learning_rate": 6.968777726651044e-06,
      "loss": 0.4461,
      "step": 19403
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5620237588882446,
      "learning_rate": 6.967807246386862e-06,
      "loss": 0.4096,
      "step": 19404
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5435450077056885,
      "learning_rate": 6.966836797571368e-06,
      "loss": 0.4602,
      "step": 19405
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.569667100906372,
      "learning_rate": 6.965866380214631e-06,
      "loss": 0.3999,
      "step": 19406
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6194261312484741,
      "learning_rate": 6.964895994326707e-06,
      "loss": 0.48,
      "step": 19407
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5303016901016235,
      "learning_rate": 6.963925639917666e-06,
      "loss": 0.395,
      "step": 19408
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.656915307044983,
      "learning_rate": 6.96295531699757e-06,
      "loss": 0.4595,
      "step": 19409
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5568184852600098,
      "learning_rate": 6.961985025576488e-06,
      "loss": 0.425,
      "step": 19410
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5809575319290161,
      "learning_rate": 6.961014765664475e-06,
      "loss": 0.4416,
      "step": 19411
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5964791774749756,
      "learning_rate": 6.9600445372716e-06,
      "loss": 0.4152,
      "step": 19412
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5501925945281982,
      "learning_rate": 6.959074340407923e-06,
      "loss": 0.4356,
      "step": 19413
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5577787160873413,
      "learning_rate": 6.958104175083512e-06,
      "loss": 0.3764,
      "step": 19414
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6978890895843506,
      "learning_rate": 6.95713404130842e-06,
      "loss": 0.4484,
      "step": 19415
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5568394660949707,
      "learning_rate": 6.956163939092717e-06,
      "loss": 0.3812,
      "step": 19416
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5939528942108154,
      "learning_rate": 6.955193868446457e-06,
      "loss": 0.5023,
      "step": 19417
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.547741174697876,
      "learning_rate": 6.9542238293797115e-06,
      "loss": 0.4088,
      "step": 19418
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5610072612762451,
      "learning_rate": 6.953253821902532e-06,
      "loss": 0.4813,
      "step": 19419
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5242044925689697,
      "learning_rate": 6.952283846024981e-06,
      "loss": 0.3833,
      "step": 19420
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4737889766693115,
      "learning_rate": 6.9513139017571214e-06,
      "loss": 0.4026,
      "step": 19421
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5105314254760742,
      "learning_rate": 6.950343989109016e-06,
      "loss": 0.3906,
      "step": 19422
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.608422875404358,
      "learning_rate": 6.949374108090714e-06,
      "loss": 0.4355,
      "step": 19423
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5551038980484009,
      "learning_rate": 6.948404258712282e-06,
      "loss": 0.4182,
      "step": 19424
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5073318481445312,
      "learning_rate": 6.947434440983778e-06,
      "loss": 0.4385,
      "step": 19425
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5230646133422852,
      "learning_rate": 6.946464654915265e-06,
      "loss": 0.3927,
      "step": 19426
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6446679830551147,
      "learning_rate": 6.94549490051679e-06,
      "loss": 0.4417,
      "step": 19427
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.519262671470642,
      "learning_rate": 6.94452517779842e-06,
      "loss": 0.384,
      "step": 19428
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5387016534805298,
      "learning_rate": 6.9435554867702075e-06,
      "loss": 0.4489,
      "step": 19429
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5764856338500977,
      "learning_rate": 6.942585827442215e-06,
      "loss": 0.3648,
      "step": 19430
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5848995447158813,
      "learning_rate": 6.9416161998244965e-06,
      "loss": 0.4209,
      "step": 19431
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6230709552764893,
      "learning_rate": 6.940646603927106e-06,
      "loss": 0.4536,
      "step": 19432
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6648045778274536,
      "learning_rate": 6.939677039760103e-06,
      "loss": 0.4577,
      "step": 19433
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5432448387145996,
      "learning_rate": 6.938707507333547e-06,
      "loss": 0.3941,
      "step": 19434
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5827913284301758,
      "learning_rate": 6.937738006657484e-06,
      "loss": 0.4695,
      "step": 19435
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7697147130966187,
      "learning_rate": 6.936768537741975e-06,
      "loss": 0.3989,
      "step": 19436
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5742404460906982,
      "learning_rate": 6.935799100597076e-06,
      "loss": 0.4564,
      "step": 19437
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6203988790512085,
      "learning_rate": 6.934829695232843e-06,
      "loss": 0.438,
      "step": 19438
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5504624843597412,
      "learning_rate": 6.933860321659325e-06,
      "loss": 0.4343,
      "step": 19439
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4870021343231201,
      "learning_rate": 6.932890979886576e-06,
      "loss": 0.386,
      "step": 19440
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5672560930252075,
      "learning_rate": 6.931921669924655e-06,
      "loss": 0.5211,
      "step": 19441
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5354937314987183,
      "learning_rate": 6.930952391783614e-06,
      "loss": 0.4221,
      "step": 19442
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6164215803146362,
      "learning_rate": 6.9299831454735015e-06,
      "loss": 0.4734,
      "step": 19443
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5258828401565552,
      "learning_rate": 6.929013931004374e-06,
      "loss": 0.4008,
      "step": 19444
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5768201351165771,
      "learning_rate": 6.92804474838628e-06,
      "loss": 0.4594,
      "step": 19445
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.532878041267395,
      "learning_rate": 6.92707559762928e-06,
      "loss": 0.3829,
      "step": 19446
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.567724585533142,
      "learning_rate": 6.926106478743416e-06,
      "loss": 0.4449,
      "step": 19447
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5646783113479614,
      "learning_rate": 6.925137391738743e-06,
      "loss": 0.4114,
      "step": 19448
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6631377935409546,
      "learning_rate": 6.924168336625313e-06,
      "loss": 0.47,
      "step": 19449
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.525880217552185,
      "learning_rate": 6.923199313413178e-06,
      "loss": 0.3931,
      "step": 19450
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5567182302474976,
      "learning_rate": 6.922230322112382e-06,
      "loss": 0.4409,
      "step": 19451
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5664316415786743,
      "learning_rate": 6.921261362732983e-06,
      "loss": 0.3919,
      "step": 19452
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5532195568084717,
      "learning_rate": 6.9202924352850234e-06,
      "loss": 0.4981,
      "step": 19453
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5078675746917725,
      "learning_rate": 6.9193235397785596e-06,
      "loss": 0.3847,
      "step": 19454
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4831008911132812,
      "learning_rate": 6.918354676223635e-06,
      "loss": 0.4088,
      "step": 19455
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5013210773468018,
      "learning_rate": 6.917385844630299e-06,
      "loss": 0.4172,
      "step": 19456
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5555987358093262,
      "learning_rate": 6.916417045008604e-06,
      "loss": 0.4175,
      "step": 19457
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5963698625564575,
      "learning_rate": 6.915448277368595e-06,
      "loss": 0.4133,
      "step": 19458
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7820767164230347,
      "learning_rate": 6.914479541720316e-06,
      "loss": 0.4739,
      "step": 19459
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.546545386314392,
      "learning_rate": 6.91351083807382e-06,
      "loss": 0.3885,
      "step": 19460
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5382905006408691,
      "learning_rate": 6.912542166439151e-06,
      "loss": 0.4401,
      "step": 19461
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4712449312210083,
      "learning_rate": 6.911573526826363e-06,
      "loss": 0.3618,
      "step": 19462
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5458502769470215,
      "learning_rate": 6.910604919245491e-06,
      "loss": 0.4457,
      "step": 19463
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5872392654418945,
      "learning_rate": 6.909636343706585e-06,
      "loss": 0.389,
      "step": 19464
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6046358346939087,
      "learning_rate": 6.908667800219693e-06,
      "loss": 0.46,
      "step": 19465
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5270087718963623,
      "learning_rate": 6.907699288794862e-06,
      "loss": 0.4233,
      "step": 19466
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.619949221611023,
      "learning_rate": 6.906730809442132e-06,
      "loss": 0.4588,
      "step": 19467
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5413848161697388,
      "learning_rate": 6.90576236217155e-06,
      "loss": 0.405,
      "step": 19468
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5445493459701538,
      "learning_rate": 6.90479394699316e-06,
      "loss": 0.4278,
      "step": 19469
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5505492687225342,
      "learning_rate": 6.903825563917008e-06,
      "loss": 0.4289,
      "step": 19470
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4857690334320068,
      "learning_rate": 6.902857212953134e-06,
      "loss": 0.4123,
      "step": 19471
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4942312240600586,
      "learning_rate": 6.9018888941115836e-06,
      "loss": 0.3938,
      "step": 19472
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5830522775650024,
      "learning_rate": 6.900920607402397e-06,
      "loss": 0.5084,
      "step": 19473
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6238595247268677,
      "learning_rate": 6.899952352835626e-06,
      "loss": 0.4328,
      "step": 19474
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5714517831802368,
      "learning_rate": 6.8989841304213025e-06,
      "loss": 0.4152,
      "step": 19475
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.561828374862671,
      "learning_rate": 6.898015940169469e-06,
      "loss": 0.4011,
      "step": 19476
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.511464238166809,
      "learning_rate": 6.897047782090175e-06,
      "loss": 0.4391,
      "step": 19477
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.934325933456421,
      "learning_rate": 6.896079656193459e-06,
      "loss": 0.4251,
      "step": 19478
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6126240491867065,
      "learning_rate": 6.895111562489355e-06,
      "loss": 0.4318,
      "step": 19479
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5759555101394653,
      "learning_rate": 6.894143500987911e-06,
      "loss": 0.4148,
      "step": 19480
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5579649209976196,
      "learning_rate": 6.893175471699165e-06,
      "loss": 0.4264,
      "step": 19481
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4515035152435303,
      "learning_rate": 6.892207474633161e-06,
      "loss": 0.3601,
      "step": 19482
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5690070390701294,
      "learning_rate": 6.891239509799932e-06,
      "loss": 0.4533,
      "step": 19483
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.545556664466858,
      "learning_rate": 6.89027157720952e-06,
      "loss": 0.4298,
      "step": 19484
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4946880340576172,
      "learning_rate": 6.889303676871966e-06,
      "loss": 0.454,
      "step": 19485
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5234789848327637,
      "learning_rate": 6.888335808797309e-06,
      "loss": 0.4046,
      "step": 19486
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6131162643432617,
      "learning_rate": 6.88736797299558e-06,
      "loss": 0.4912,
      "step": 19487
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5173805952072144,
      "learning_rate": 6.886400169476825e-06,
      "loss": 0.3867,
      "step": 19488
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5239065885543823,
      "learning_rate": 6.8854323982510785e-06,
      "loss": 0.4118,
      "step": 19489
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4932271242141724,
      "learning_rate": 6.8844646593283825e-06,
      "loss": 0.3879,
      "step": 19490
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5343592166900635,
      "learning_rate": 6.883496952718767e-06,
      "loss": 0.4382,
      "step": 19491
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5577796697616577,
      "learning_rate": 6.882529278432269e-06,
      "loss": 0.4276,
      "step": 19492
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4698516130447388,
      "learning_rate": 6.881561636478931e-06,
      "loss": 0.4099,
      "step": 19493
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4915857315063477,
      "learning_rate": 6.880594026868787e-06,
      "loss": 0.4012,
      "step": 19494
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6625853776931763,
      "learning_rate": 6.879626449611867e-06,
      "loss": 0.4315,
      "step": 19495
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6416376829147339,
      "learning_rate": 6.878658904718212e-06,
      "loss": 0.4089,
      "step": 19496
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5525249242782593,
      "learning_rate": 6.877691392197854e-06,
      "loss": 0.432,
      "step": 19497
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.611756443977356,
      "learning_rate": 6.876723912060834e-06,
      "loss": 0.412,
      "step": 19498
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.658308506011963,
      "learning_rate": 6.875756464317178e-06,
      "loss": 0.9952,
      "step": 19499
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.498233675956726,
      "learning_rate": 6.874789048976922e-06,
      "loss": 1.0683,
      "step": 19500
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6680676937103271,
      "learning_rate": 6.873821666050103e-06,
      "loss": 0.4147,
      "step": 19501
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4996261596679688,
      "learning_rate": 6.872854315546753e-06,
      "loss": 0.3891,
      "step": 19502
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5126738548278809,
      "learning_rate": 6.8718869974769045e-06,
      "loss": 0.4551,
      "step": 19503
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5603481531143188,
      "learning_rate": 6.870919711850585e-06,
      "loss": 0.3981,
      "step": 19504
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7240920066833496,
      "learning_rate": 6.869952458677838e-06,
      "loss": 0.4772,
      "step": 19505
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5913749933242798,
      "learning_rate": 6.868985237968688e-06,
      "loss": 0.4607,
      "step": 19506
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5435492992401123,
      "learning_rate": 6.868018049733165e-06,
      "loss": 0.4394,
      "step": 19507
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6205042600631714,
      "learning_rate": 6.867050893981304e-06,
      "loss": 0.4315,
      "step": 19508
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5761994123458862,
      "learning_rate": 6.866083770723133e-06,
      "loss": 0.4323,
      "step": 19509
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.526781678199768,
      "learning_rate": 6.865116679968691e-06,
      "loss": 0.3951,
      "step": 19510
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.590451717376709,
      "learning_rate": 6.864149621727998e-06,
      "loss": 0.4437,
      "step": 19511
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.607030987739563,
      "learning_rate": 6.8631825960110866e-06,
      "loss": 0.4097,
      "step": 19512
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7108168601989746,
      "learning_rate": 6.862215602827989e-06,
      "loss": 0.4421,
      "step": 19513
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.551004409790039,
      "learning_rate": 6.861248642188731e-06,
      "loss": 0.4111,
      "step": 19514
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5617971420288086,
      "learning_rate": 6.86028171410335e-06,
      "loss": 0.42,
      "step": 19515
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5618925094604492,
      "learning_rate": 6.859314818581863e-06,
      "loss": 0.4058,
      "step": 19516
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5248929262161255,
      "learning_rate": 6.858347955634304e-06,
      "loss": 0.4741,
      "step": 19517
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5403022766113281,
      "learning_rate": 6.857381125270701e-06,
      "loss": 0.4089,
      "step": 19518
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7600177526474,
      "learning_rate": 6.8564143275010845e-06,
      "loss": 0.9818,
      "step": 19519
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5263081789016724,
      "learning_rate": 6.855447562335473e-06,
      "loss": 1.0127,
      "step": 19520
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6060035228729248,
      "learning_rate": 6.8544808297839e-06,
      "loss": 0.5181,
      "step": 19521
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.608373999595642,
      "learning_rate": 6.853514129856391e-06,
      "loss": 0.407,
      "step": 19522
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6039059162139893,
      "learning_rate": 6.852547462562975e-06,
      "loss": 0.482,
      "step": 19523
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4948077201843262,
      "learning_rate": 6.851580827913673e-06,
      "loss": 0.3993,
      "step": 19524
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6398371458053589,
      "learning_rate": 6.850614225918509e-06,
      "loss": 0.4591,
      "step": 19525
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4790953397750854,
      "learning_rate": 6.8496476565875144e-06,
      "loss": 0.4034,
      "step": 19526
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5663912296295166,
      "learning_rate": 6.848681119930715e-06,
      "loss": 0.4793,
      "step": 19527
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4592374563217163,
      "learning_rate": 6.847714615958125e-06,
      "loss": 0.3968,
      "step": 19528
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7013834714889526,
      "learning_rate": 6.846748144679779e-06,
      "loss": 0.4467,
      "step": 19529
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5569121837615967,
      "learning_rate": 6.8457817061056946e-06,
      "loss": 0.3869,
      "step": 19530
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5701797008514404,
      "learning_rate": 6.844815300245901e-06,
      "loss": 0.4164,
      "step": 19531
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5647623538970947,
      "learning_rate": 6.843848927110417e-06,
      "loss": 0.4191,
      "step": 19532
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.478455662727356,
      "learning_rate": 6.842882586709264e-06,
      "loss": 0.4158,
      "step": 19533
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5053319931030273,
      "learning_rate": 6.841916279052469e-06,
      "loss": 0.408,
      "step": 19534
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6059668064117432,
      "learning_rate": 6.840950004150055e-06,
      "loss": 0.4241,
      "step": 19535
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5688129663467407,
      "learning_rate": 6.839983762012037e-06,
      "loss": 0.3973,
      "step": 19536
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5356625318527222,
      "learning_rate": 6.839017552648441e-06,
      "loss": 0.4699,
      "step": 19537
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5616428852081299,
      "learning_rate": 6.838051376069287e-06,
      "loss": 0.4267,
      "step": 19538
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4996440410614014,
      "learning_rate": 6.8370852322846e-06,
      "loss": 0.4519,
      "step": 19539
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6669405698776245,
      "learning_rate": 6.8361191213043956e-06,
      "loss": 0.4279,
      "step": 19540
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5355607271194458,
      "learning_rate": 6.835153043138692e-06,
      "loss": 0.4317,
      "step": 19541
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5235450267791748,
      "learning_rate": 6.834186997797515e-06,
      "loss": 0.4282,
      "step": 19542
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6441152095794678,
      "learning_rate": 6.833220985290882e-06,
      "loss": 0.4475,
      "step": 19543
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4980124235153198,
      "learning_rate": 6.832255005628812e-06,
      "loss": 0.3967,
      "step": 19544
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6317243576049805,
      "learning_rate": 6.831289058821319e-06,
      "loss": 0.494,
      "step": 19545
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5252290964126587,
      "learning_rate": 6.830323144878428e-06,
      "loss": 0.4033,
      "step": 19546
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8057032823562622,
      "learning_rate": 6.829357263810157e-06,
      "loss": 1.0502,
      "step": 19547
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.505812644958496,
      "learning_rate": 6.828391415626516e-06,
      "loss": 1.0591,
      "step": 19548
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6047868728637695,
      "learning_rate": 6.8274256003375305e-06,
      "loss": 0.4344,
      "step": 19549
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.659758448600769,
      "learning_rate": 6.826459817953213e-06,
      "loss": 0.4355,
      "step": 19550
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5809975862503052,
      "learning_rate": 6.825494068483585e-06,
      "loss": 0.4895,
      "step": 19551
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5359376668930054,
      "learning_rate": 6.824528351938659e-06,
      "loss": 0.4005,
      "step": 19552
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.522576928138733,
      "learning_rate": 6.82356266832845e-06,
      "loss": 0.466,
      "step": 19553
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5278669595718384,
      "learning_rate": 6.8225970176629754e-06,
      "loss": 0.4339,
      "step": 19554
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4796351194381714,
      "learning_rate": 6.821631399952254e-06,
      "loss": 0.4212,
      "step": 19555
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5819799900054932,
      "learning_rate": 6.820665815206293e-06,
      "loss": 0.3875,
      "step": 19556
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6073249578475952,
      "learning_rate": 6.819700263435114e-06,
      "loss": 0.4388,
      "step": 19557
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5496315956115723,
      "learning_rate": 6.8187347446487276e-06,
      "loss": 0.4036,
      "step": 19558
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4958010911941528,
      "learning_rate": 6.817769258857152e-06,
      "loss": 0.4735,
      "step": 19559
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6907473802566528,
      "learning_rate": 6.816803806070395e-06,
      "loss": 0.4059,
      "step": 19560
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5284134149551392,
      "learning_rate": 6.8158383862984725e-06,
      "loss": 0.4524,
      "step": 19561
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5787951946258545,
      "learning_rate": 6.814872999551399e-06,
      "loss": 0.3977,
      "step": 19562
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6298441886901855,
      "learning_rate": 6.813907645839188e-06,
      "loss": 0.416,
      "step": 19563
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6352752447128296,
      "learning_rate": 6.812942325171845e-06,
      "loss": 0.4127,
      "step": 19564
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7436716556549072,
      "learning_rate": 6.811977037559389e-06,
      "loss": 0.9908,
      "step": 19565
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3907400369644165,
      "learning_rate": 6.811011783011827e-06,
      "loss": 0.983,
      "step": 19566
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6601368188858032,
      "learning_rate": 6.810046561539175e-06,
      "loss": 0.4312,
      "step": 19567
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5321033000946045,
      "learning_rate": 6.809081373151441e-06,
      "loss": 0.3918,
      "step": 19568
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5369868278503418,
      "learning_rate": 6.808116217858632e-06,
      "loss": 0.4041,
      "step": 19569
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5338629484176636,
      "learning_rate": 6.807151095670765e-06,
      "loss": 0.403,
      "step": 19570
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6580411195755005,
      "learning_rate": 6.806186006597848e-06,
      "loss": 0.4911,
      "step": 19571
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6718661785125732,
      "learning_rate": 6.8052209506498865e-06,
      "loss": 0.4151,
      "step": 19572
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.507394790649414,
      "learning_rate": 6.804255927836893e-06,
      "loss": 0.4193,
      "step": 19573
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5489201545715332,
      "learning_rate": 6.803290938168875e-06,
      "loss": 0.401,
      "step": 19574
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5765039920806885,
      "learning_rate": 6.802325981655845e-06,
      "loss": 0.4609,
      "step": 19575
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.500534176826477,
      "learning_rate": 6.801361058307807e-06,
      "loss": 0.3953,
      "step": 19576
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5463292598724365,
      "learning_rate": 6.800396168134766e-06,
      "loss": 0.4356,
      "step": 19577
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5709360837936401,
      "learning_rate": 6.799431311146736e-06,
      "loss": 0.4066,
      "step": 19578
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5625725984573364,
      "learning_rate": 6.798466487353723e-06,
      "loss": 0.4533,
      "step": 19579
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6410995721817017,
      "learning_rate": 6.797501696765731e-06,
      "loss": 0.4024,
      "step": 19580
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5212187767028809,
      "learning_rate": 6.796536939392764e-06,
      "loss": 0.4381,
      "step": 19581
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5006998777389526,
      "learning_rate": 6.795572215244834e-06,
      "loss": 0.4054,
      "step": 19582
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5288540124893188,
      "learning_rate": 6.794607524331948e-06,
      "loss": 0.4395,
      "step": 19583
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5952942371368408,
      "learning_rate": 6.793642866664102e-06,
      "loss": 0.4101,
      "step": 19584
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.540607213973999,
      "learning_rate": 6.7926782422513094e-06,
      "loss": 0.472,
      "step": 19585
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4519000053405762,
      "learning_rate": 6.791713651103569e-06,
      "loss": 0.3904,
      "step": 19586
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6352763175964355,
      "learning_rate": 6.790749093230893e-06,
      "loss": 0.485,
      "step": 19587
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.521756649017334,
      "learning_rate": 6.789784568643278e-06,
      "loss": 0.4044,
      "step": 19588
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5089586973190308,
      "learning_rate": 6.788820077350731e-06,
      "loss": 0.4044,
      "step": 19589
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5821644067764282,
      "learning_rate": 6.787855619363255e-06,
      "loss": 0.3819,
      "step": 19590
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6830874681472778,
      "learning_rate": 6.786891194690855e-06,
      "loss": 0.4389,
      "step": 19591
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5414854288101196,
      "learning_rate": 6.785926803343526e-06,
      "loss": 0.3847,
      "step": 19592
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5641220808029175,
      "learning_rate": 6.784962445331279e-06,
      "loss": 0.4811,
      "step": 19593
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4773069620132446,
      "learning_rate": 6.78399812066411e-06,
      "loss": 0.3814,
      "step": 19594
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6579053401947021,
      "learning_rate": 6.7830338293520285e-06,
      "loss": 0.4967,
      "step": 19595
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4515410661697388,
      "learning_rate": 6.782069571405025e-06,
      "loss": 0.3802,
      "step": 19596
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6576870679855347,
      "learning_rate": 6.781105346833107e-06,
      "loss": 0.4229,
      "step": 19597
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.694942831993103,
      "learning_rate": 6.780141155646275e-06,
      "loss": 0.4682,
      "step": 19598
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.50063157081604,
      "learning_rate": 6.77917699785453e-06,
      "loss": 0.4066,
      "step": 19599
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.542906403541565,
      "learning_rate": 6.778212873467866e-06,
      "loss": 0.4108,
      "step": 19600
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.550418496131897,
      "learning_rate": 6.777248782496287e-06,
      "loss": 0.4911,
      "step": 19601
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5248658657073975,
      "learning_rate": 6.776284724949791e-06,
      "loss": 0.4081,
      "step": 19602
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6204919815063477,
      "learning_rate": 6.775320700838382e-06,
      "loss": 0.4942,
      "step": 19603
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5151076316833496,
      "learning_rate": 6.77435671017205e-06,
      "loss": 0.4111,
      "step": 19604
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6309186220169067,
      "learning_rate": 6.7733927529607964e-06,
      "loss": 0.4088,
      "step": 19605
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5650259256362915,
      "learning_rate": 6.772428829214621e-06,
      "loss": 0.4042,
      "step": 19606
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5173994302749634,
      "learning_rate": 6.771464938943522e-06,
      "loss": 0.4265,
      "step": 19607
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.52793550491333,
      "learning_rate": 6.77050108215749e-06,
      "loss": 0.4122,
      "step": 19608
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6326956748962402,
      "learning_rate": 6.769537258866528e-06,
      "loss": 0.4729,
      "step": 19609
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5682748556137085,
      "learning_rate": 6.768573469080628e-06,
      "loss": 0.4225,
      "step": 19610
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5714833736419678,
      "learning_rate": 6.767609712809793e-06,
      "loss": 0.4149,
      "step": 19611
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5380325317382812,
      "learning_rate": 6.7666459900640136e-06,
      "loss": 0.4021,
      "step": 19612
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6452420949935913,
      "learning_rate": 6.7656823008532815e-06,
      "loss": 0.4638,
      "step": 19613
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5246477127075195,
      "learning_rate": 6.764718645187598e-06,
      "loss": 0.3972,
      "step": 19614
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5393415689468384,
      "learning_rate": 6.763755023076959e-06,
      "loss": 0.4175,
      "step": 19615
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5362987518310547,
      "learning_rate": 6.7627914345313505e-06,
      "loss": 0.4366,
      "step": 19616
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6219538450241089,
      "learning_rate": 6.761827879560774e-06,
      "loss": 0.4716,
      "step": 19617
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4809695482254028,
      "learning_rate": 6.760864358175217e-06,
      "loss": 0.3812,
      "step": 19618
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5359795093536377,
      "learning_rate": 6.759900870384684e-06,
      "loss": 0.4302,
      "step": 19619
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5531296730041504,
      "learning_rate": 6.758937416199153e-06,
      "loss": 0.3887,
      "step": 19620
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5497100353240967,
      "learning_rate": 6.7579739956286265e-06,
      "loss": 0.4505,
      "step": 19621
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5206847190856934,
      "learning_rate": 6.757010608683091e-06,
      "loss": 0.3991,
      "step": 19622
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.519476056098938,
      "learning_rate": 6.7560472553725465e-06,
      "loss": 0.4376,
      "step": 19623
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5381920337677002,
      "learning_rate": 6.755083935706977e-06,
      "loss": 0.416,
      "step": 19624
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7052785158157349,
      "learning_rate": 6.754120649696373e-06,
      "loss": 0.4243,
      "step": 19625
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6131772994995117,
      "learning_rate": 6.753157397350731e-06,
      "loss": 0.4159,
      "step": 19626
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.66544771194458,
      "learning_rate": 6.752194178680041e-06,
      "loss": 0.4465,
      "step": 19627
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5415518283843994,
      "learning_rate": 6.7512309936942866e-06,
      "loss": 0.3982,
      "step": 19628
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5447883605957031,
      "learning_rate": 6.7502678424034626e-06,
      "loss": 0.4592,
      "step": 19629
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.477881669998169,
      "learning_rate": 6.749304724817557e-06,
      "loss": 0.4054,
      "step": 19630
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5410890579223633,
      "learning_rate": 6.748341640946563e-06,
      "loss": 0.4514,
      "step": 19631
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5648612976074219,
      "learning_rate": 6.747378590800464e-06,
      "loss": 0.3892,
      "step": 19632
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5757129192352295,
      "learning_rate": 6.746415574389249e-06,
      "loss": 0.451,
      "step": 19633
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.59148108959198,
      "learning_rate": 6.7454525917229084e-06,
      "loss": 0.3728,
      "step": 19634
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.509579062461853,
      "learning_rate": 6.744489642811431e-06,
      "loss": 0.4917,
      "step": 19635
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5244885683059692,
      "learning_rate": 6.743526727664797e-06,
      "loss": 0.4127,
      "step": 19636
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5514822006225586,
      "learning_rate": 6.742563846293002e-06,
      "loss": 0.4622,
      "step": 19637
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8107284307479858,
      "learning_rate": 6.741600998706026e-06,
      "loss": 0.3974,
      "step": 19638
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5819263458251953,
      "learning_rate": 6.740638184913861e-06,
      "loss": 0.4669,
      "step": 19639
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5899043083190918,
      "learning_rate": 6.73967540492649e-06,
      "loss": 0.3853,
      "step": 19640
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5360047817230225,
      "learning_rate": 6.738712658753895e-06,
      "loss": 0.4428,
      "step": 19641
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6729276180267334,
      "learning_rate": 6.737749946406067e-06,
      "loss": 0.3999,
      "step": 19642
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.562770128250122,
      "learning_rate": 6.736787267892991e-06,
      "loss": 0.4313,
      "step": 19643
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5008831024169922,
      "learning_rate": 6.735824623224645e-06,
      "loss": 0.4287,
      "step": 19644
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6640230417251587,
      "learning_rate": 6.73486201241102e-06,
      "loss": 0.4385,
      "step": 19645
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4764052629470825,
      "learning_rate": 6.733899435462094e-06,
      "loss": 0.4076,
      "step": 19646
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6439824104309082,
      "learning_rate": 6.732936892387859e-06,
      "loss": 0.4973,
      "step": 19647
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5283836126327515,
      "learning_rate": 6.73197438319829e-06,
      "loss": 0.3826,
      "step": 19648
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6346162557601929,
      "learning_rate": 6.7310119079033705e-06,
      "loss": 0.4554,
      "step": 19649
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4934091567993164,
      "learning_rate": 6.730049466513086e-06,
      "loss": 0.3801,
      "step": 19650
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5653210878372192,
      "learning_rate": 6.7290870590374205e-06,
      "loss": 0.4577,
      "step": 19651
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5375018119812012,
      "learning_rate": 6.728124685486348e-06,
      "loss": 0.4079,
      "step": 19652
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5849027633666992,
      "learning_rate": 6.727162345869858e-06,
      "loss": 0.4457,
      "step": 19653
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5737987756729126,
      "learning_rate": 6.726200040197925e-06,
      "loss": 0.3881,
      "step": 19654
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6090362071990967,
      "learning_rate": 6.725237768480538e-06,
      "loss": 0.4607,
      "step": 19655
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4878826141357422,
      "learning_rate": 6.724275530727667e-06,
      "loss": 0.3905,
      "step": 19656
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6423300504684448,
      "learning_rate": 6.723313326949297e-06,
      "loss": 0.4299,
      "step": 19657
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6366362571716309,
      "learning_rate": 6.722351157155408e-06,
      "loss": 0.4469,
      "step": 19658
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.546055555343628,
      "learning_rate": 6.721389021355982e-06,
      "loss": 0.4619,
      "step": 19659
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4788657426834106,
      "learning_rate": 6.720426919560992e-06,
      "loss": 0.4047,
      "step": 19660
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1114139556884766,
      "learning_rate": 6.7194648517804175e-06,
      "loss": 0.9887,
      "step": 19661
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.505873680114746,
      "learning_rate": 6.7185028180242395e-06,
      "loss": 1.0658,
      "step": 19662
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6327173709869385,
      "learning_rate": 6.717540818302438e-06,
      "loss": 0.4413,
      "step": 19663
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6217284202575684,
      "learning_rate": 6.716578852624982e-06,
      "loss": 0.4203,
      "step": 19664
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.564439296722412,
      "learning_rate": 6.715616921001857e-06,
      "loss": 0.4463,
      "step": 19665
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5661059617996216,
      "learning_rate": 6.714655023443033e-06,
      "loss": 0.4104,
      "step": 19666
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.90151846408844,
      "learning_rate": 6.713693159958496e-06,
      "loss": 0.4602,
      "step": 19667
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.483407974243164,
      "learning_rate": 6.712731330558212e-06,
      "loss": 0.3841,
      "step": 19668
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.022690773010254,
      "learning_rate": 6.711769535252159e-06,
      "loss": 0.439,
      "step": 19669
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5233914852142334,
      "learning_rate": 6.710807774050316e-06,
      "loss": 0.3849,
      "step": 19670
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4856419563293457,
      "learning_rate": 6.709846046962661e-06,
      "loss": 0.4345,
      "step": 19671
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.550220251083374,
      "learning_rate": 6.7088843539991556e-06,
      "loss": 0.3889,
      "step": 19672
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7191320657730103,
      "learning_rate": 6.7079226951697865e-06,
      "loss": 0.4403,
      "step": 19673
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.555379867553711,
      "learning_rate": 6.70696107048452e-06,
      "loss": 0.4296,
      "step": 19674
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6004581451416016,
      "learning_rate": 6.705999479953338e-06,
      "loss": 0.4201,
      "step": 19675
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4578700065612793,
      "learning_rate": 6.705037923586206e-06,
      "loss": 0.3819,
      "step": 19676
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.510218620300293,
      "learning_rate": 6.704076401393098e-06,
      "loss": 0.4375,
      "step": 19677
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.525481104850769,
      "learning_rate": 6.703114913383991e-06,
      "loss": 0.4555,
      "step": 19678
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5026426315307617,
      "learning_rate": 6.702153459568856e-06,
      "loss": 0.429,
      "step": 19679
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5451302528381348,
      "learning_rate": 6.701192039957658e-06,
      "loss": 0.4046,
      "step": 19680
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5694154500961304,
      "learning_rate": 6.700230654560378e-06,
      "loss": 0.4758,
      "step": 19681
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5905417203903198,
      "learning_rate": 6.6992693033869785e-06,
      "loss": 0.4042,
      "step": 19682
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5150107145309448,
      "learning_rate": 6.698307986447439e-06,
      "loss": 0.4455,
      "step": 19683
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6379324197769165,
      "learning_rate": 6.697346703751724e-06,
      "loss": 0.4292,
      "step": 19684
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4894623756408691,
      "learning_rate": 6.696385455309802e-06,
      "loss": 0.4328,
      "step": 19685
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5444368124008179,
      "learning_rate": 6.695424241131647e-06,
      "loss": 0.4149,
      "step": 19686
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5491403341293335,
      "learning_rate": 6.694463061227231e-06,
      "loss": 0.4303,
      "step": 19687
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.506917953491211,
      "learning_rate": 6.693501915606513e-06,
      "loss": 0.3915,
      "step": 19688
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5172163248062134,
      "learning_rate": 6.69254080427947e-06,
      "loss": 0.443,
      "step": 19689
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.582478404045105,
      "learning_rate": 6.691579727256067e-06,
      "loss": 0.4306,
      "step": 19690
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.498035192489624,
      "learning_rate": 6.690618684546275e-06,
      "loss": 0.3986,
      "step": 19691
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5409183502197266,
      "learning_rate": 6.689657676160058e-06,
      "loss": 0.4072,
      "step": 19692
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5604803562164307,
      "learning_rate": 6.688696702107383e-06,
      "loss": 0.4505,
      "step": 19693
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4922242164611816,
      "learning_rate": 6.687735762398218e-06,
      "loss": 0.425,
      "step": 19694
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4851471185684204,
      "learning_rate": 6.686774857042534e-06,
      "loss": 0.4772,
      "step": 19695
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5938684940338135,
      "learning_rate": 6.68581398605029e-06,
      "loss": 0.4014,
      "step": 19696
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6129525899887085,
      "learning_rate": 6.684853149431453e-06,
      "loss": 0.4766,
      "step": 19697
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5256716012954712,
      "learning_rate": 6.6838923471959915e-06,
      "loss": 0.3822,
      "step": 19698
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5493866205215454,
      "learning_rate": 6.68293157935387e-06,
      "loss": 0.441,
      "step": 19699
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4959431886672974,
      "learning_rate": 6.681970845915051e-06,
      "loss": 0.3816,
      "step": 19700
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5697965621948242,
      "learning_rate": 6.681010146889501e-06,
      "loss": 0.4384,
      "step": 19701
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5464311838150024,
      "learning_rate": 6.680049482287181e-06,
      "loss": 0.4003,
      "step": 19702
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.634827733039856,
      "learning_rate": 6.679088852118061e-06,
      "loss": 0.4997,
      "step": 19703
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4596168994903564,
      "learning_rate": 6.678128256392097e-06,
      "loss": 0.3612,
      "step": 19704
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5400784015655518,
      "learning_rate": 6.677167695119253e-06,
      "loss": 0.3932,
      "step": 19705
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6171972751617432,
      "learning_rate": 6.676207168309497e-06,
      "loss": 0.4444,
      "step": 19706
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4953052997589111,
      "learning_rate": 6.675246675972789e-06,
      "loss": 0.4465,
      "step": 19707
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5429562330245972,
      "learning_rate": 6.674286218119084e-06,
      "loss": 0.4308,
      "step": 19708
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.2750871181488037,
      "learning_rate": 6.6733257947583514e-06,
      "loss": 0.4501,
      "step": 19709
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4782930612564087,
      "learning_rate": 6.67236540590055e-06,
      "loss": 0.3805,
      "step": 19710
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6459299325942993,
      "learning_rate": 6.671405051555643e-06,
      "loss": 0.4533,
      "step": 19711
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.552034854888916,
      "learning_rate": 6.670444731733587e-06,
      "loss": 0.4169,
      "step": 19712
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6167024374008179,
      "learning_rate": 6.66948444644434e-06,
      "loss": 0.485,
      "step": 19713
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6292593479156494,
      "learning_rate": 6.6685241956978675e-06,
      "loss": 0.4055,
      "step": 19714
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6125099658966064,
      "learning_rate": 6.667563979504129e-06,
      "loss": 0.4391,
      "step": 19715
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4964840412139893,
      "learning_rate": 6.666603797873077e-06,
      "loss": 0.3629,
      "step": 19716
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5740299224853516,
      "learning_rate": 6.6656436508146745e-06,
      "loss": 0.4314,
      "step": 19717
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4709863662719727,
      "learning_rate": 6.6646835383388785e-06,
      "loss": 0.3848,
      "step": 19718
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6119219064712524,
      "learning_rate": 6.663723460455652e-06,
      "loss": 0.4532,
      "step": 19719
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6075875759124756,
      "learning_rate": 6.662763417174947e-06,
      "loss": 0.4175,
      "step": 19720
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7110775709152222,
      "learning_rate": 6.661803408506717e-06,
      "loss": 0.9499,
      "step": 19721
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4677544832229614,
      "learning_rate": 6.660843434460929e-06,
      "loss": 0.9685,
      "step": 19722
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7602301836013794,
      "learning_rate": 6.659883495047534e-06,
      "loss": 0.9638,
      "step": 19723
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.577973484992981,
      "learning_rate": 6.658923590276485e-06,
      "loss": 1.0414,
      "step": 19724
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.991410255432129,
      "learning_rate": 6.657963720157743e-06,
      "loss": 0.4307,
      "step": 19725
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5964839458465576,
      "learning_rate": 6.657003884701259e-06,
      "loss": 0.4202,
      "step": 19726
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6022080183029175,
      "learning_rate": 6.656044083916994e-06,
      "loss": 0.4284,
      "step": 19727
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5368270874023438,
      "learning_rate": 6.655084317814897e-06,
      "loss": 0.3997,
      "step": 19728
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.572509765625,
      "learning_rate": 6.6541245864049245e-06,
      "loss": 0.4754,
      "step": 19729
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5602809190750122,
      "learning_rate": 6.65316488969703e-06,
      "loss": 0.4496,
      "step": 19730
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5416768789291382,
      "learning_rate": 6.652205227701171e-06,
      "loss": 0.4403,
      "step": 19731
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5306731462478638,
      "learning_rate": 6.651245600427292e-06,
      "loss": 0.3986,
      "step": 19732
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5747570991516113,
      "learning_rate": 6.650286007885354e-06,
      "loss": 0.4307,
      "step": 19733
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5586230754852295,
      "learning_rate": 6.649326450085305e-06,
      "loss": 0.3947,
      "step": 19734
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6675459146499634,
      "learning_rate": 6.648366927037104e-06,
      "loss": 0.4807,
      "step": 19735
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4701534509658813,
      "learning_rate": 6.647407438750691e-06,
      "loss": 0.3855,
      "step": 19736
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.50967538356781,
      "learning_rate": 6.646447985236026e-06,
      "loss": 0.445,
      "step": 19737
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.534182071685791,
      "learning_rate": 6.645488566503057e-06,
      "loss": 0.3735,
      "step": 19738
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6290359497070312,
      "learning_rate": 6.6445291825617385e-06,
      "loss": 0.4314,
      "step": 19739
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.469187617301941,
      "learning_rate": 6.643569833422017e-06,
      "loss": 0.382,
      "step": 19740
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5029652118682861,
      "learning_rate": 6.642610519093841e-06,
      "loss": 0.4482,
      "step": 19741
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6186202764511108,
      "learning_rate": 6.641651239587166e-06,
      "loss": 0.4116,
      "step": 19742
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5128957033157349,
      "learning_rate": 6.640691994911939e-06,
      "loss": 0.4452,
      "step": 19743
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4749999046325684,
      "learning_rate": 6.639732785078102e-06,
      "loss": 0.3965,
      "step": 19744
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5742460489273071,
      "learning_rate": 6.6387736100956136e-06,
      "loss": 0.4515,
      "step": 19745
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5751267671585083,
      "learning_rate": 6.637814469974415e-06,
      "loss": 0.4367,
      "step": 19746
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5224311351776123,
      "learning_rate": 6.6368553647244616e-06,
      "loss": 0.4282,
      "step": 19747
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6504987478256226,
      "learning_rate": 6.635896294355693e-06,
      "loss": 0.4204,
      "step": 19748
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7056220769882202,
      "learning_rate": 6.634937258878057e-06,
      "loss": 0.4213,
      "step": 19749
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.487712025642395,
      "learning_rate": 6.633978258301506e-06,
      "loss": 0.381,
      "step": 19750
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6878410577774048,
      "learning_rate": 6.633019292635983e-06,
      "loss": 0.433,
      "step": 19751
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4926308393478394,
      "learning_rate": 6.632060361891431e-06,
      "loss": 0.3998,
      "step": 19752
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6765908002853394,
      "learning_rate": 6.631101466077801e-06,
      "loss": 0.4739,
      "step": 19753
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.481477975845337,
      "learning_rate": 6.630142605205034e-06,
      "loss": 0.3997,
      "step": 19754
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4905567169189453,
      "learning_rate": 6.629183779283079e-06,
      "loss": 0.4436,
      "step": 19755
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.511680245399475,
      "learning_rate": 6.62822498832188e-06,
      "loss": 0.3909,
      "step": 19756
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5301308631896973,
      "learning_rate": 6.627266232331375e-06,
      "loss": 0.45,
      "step": 19757
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6406230926513672,
      "learning_rate": 6.6263075113215145e-06,
      "loss": 0.4061,
      "step": 19758
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5581148862838745,
      "learning_rate": 6.62534882530224e-06,
      "loss": 0.4327,
      "step": 19759
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.47678542137146,
      "learning_rate": 6.6243901742834975e-06,
      "loss": 0.3856,
      "step": 19760
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6327760219573975,
      "learning_rate": 6.623431558275224e-06,
      "loss": 0.451,
      "step": 19761
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.673232078552246,
      "learning_rate": 6.622472977287362e-06,
      "loss": 0.4175,
      "step": 19762
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6011372804641724,
      "learning_rate": 6.62151443132986e-06,
      "loss": 0.4458,
      "step": 19763
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4955371618270874,
      "learning_rate": 6.620555920412657e-06,
      "loss": 0.3842,
      "step": 19764
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5946533679962158,
      "learning_rate": 6.619597444545689e-06,
      "loss": 0.4489,
      "step": 19765
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5213813781738281,
      "learning_rate": 6.618639003738902e-06,
      "loss": 0.3857,
      "step": 19766
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5349324941635132,
      "learning_rate": 6.617680598002235e-06,
      "loss": 0.4284,
      "step": 19767
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.510233998298645,
      "learning_rate": 6.6167222273456334e-06,
      "loss": 0.3885,
      "step": 19768
    },
    {
      "epoch": 0.62,
      "grad_norm": 3.051269292831421,
      "learning_rate": 6.615763891779029e-06,
      "loss": 0.4505,
      "step": 19769
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5538705587387085,
      "learning_rate": 6.614805591312363e-06,
      "loss": 0.3949,
      "step": 19770
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.633517861366272,
      "learning_rate": 6.613847325955578e-06,
      "loss": 0.4899,
      "step": 19771
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5493602752685547,
      "learning_rate": 6.612889095718614e-06,
      "loss": 0.4136,
      "step": 19772
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5929723978042603,
      "learning_rate": 6.611930900611404e-06,
      "loss": 0.4841,
      "step": 19773
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4517381191253662,
      "learning_rate": 6.610972740643886e-06,
      "loss": 0.3699,
      "step": 19774
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6731204986572266,
      "learning_rate": 6.6100146158260015e-06,
      "loss": 0.4352,
      "step": 19775
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4832026958465576,
      "learning_rate": 6.60905652616769e-06,
      "loss": 0.3883,
      "step": 19776
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4946578741073608,
      "learning_rate": 6.608098471678879e-06,
      "loss": 0.4144,
      "step": 19777
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5645424127578735,
      "learning_rate": 6.6071404523695135e-06,
      "loss": 0.4107,
      "step": 19778
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5323967933654785,
      "learning_rate": 6.606182468249525e-06,
      "loss": 0.4168,
      "step": 19779
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5571109056472778,
      "learning_rate": 6.605224519328854e-06,
      "loss": 0.4352,
      "step": 19780
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5791449546813965,
      "learning_rate": 6.604266605617433e-06,
      "loss": 0.4297,
      "step": 19781
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5285574197769165,
      "learning_rate": 6.603308727125195e-06,
      "loss": 0.3796,
      "step": 19782
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5094594955444336,
      "learning_rate": 6.602350883862078e-06,
      "loss": 0.4525,
      "step": 19783
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.561560869216919,
      "learning_rate": 6.601393075838019e-06,
      "loss": 0.4047,
      "step": 19784
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5437564849853516,
      "learning_rate": 6.600435303062944e-06,
      "loss": 0.4478,
      "step": 19785
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4921609163284302,
      "learning_rate": 6.599477565546793e-06,
      "loss": 0.4031,
      "step": 19786
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5411335229873657,
      "learning_rate": 6.598519863299495e-06,
      "loss": 0.472,
      "step": 19787
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5273191928863525,
      "learning_rate": 6.59756219633099e-06,
      "loss": 0.392,
      "step": 19788
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.533478021621704,
      "learning_rate": 6.596604564651204e-06,
      "loss": 0.4239,
      "step": 19789
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6121982336044312,
      "learning_rate": 6.595646968270067e-06,
      "loss": 0.453,
      "step": 19790
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4760297536849976,
      "learning_rate": 6.594689407197519e-06,
      "loss": 0.4356,
      "step": 19791
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4955919981002808,
      "learning_rate": 6.593731881443488e-06,
      "loss": 0.3917,
      "step": 19792
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6033570766448975,
      "learning_rate": 6.5927743910179e-06,
      "loss": 0.4872,
      "step": 19793
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5984903573989868,
      "learning_rate": 6.591816935930693e-06,
      "loss": 0.3756,
      "step": 19794
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4558131694793701,
      "learning_rate": 6.590859516191792e-06,
      "loss": 0.4307,
      "step": 19795
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6420689821243286,
      "learning_rate": 6.589902131811132e-06,
      "loss": 0.4278,
      "step": 19796
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.672890067100525,
      "learning_rate": 6.58894478279864e-06,
      "loss": 0.4598,
      "step": 19797
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.531949520111084,
      "learning_rate": 6.587987469164242e-06,
      "loss": 0.3738,
      "step": 19798
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4847335815429688,
      "learning_rate": 6.587030190917872e-06,
      "loss": 0.4405,
      "step": 19799
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6362630128860474,
      "learning_rate": 6.586072948069459e-06,
      "loss": 0.4719,
      "step": 19800
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5760602951049805,
      "learning_rate": 6.585115740628924e-06,
      "loss": 0.4226,
      "step": 19801
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5230168104171753,
      "learning_rate": 6.584158568606203e-06,
      "loss": 0.4159,
      "step": 19802
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5839663743972778,
      "learning_rate": 6.583201432011217e-06,
      "loss": 0.4776,
      "step": 19803
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5121763944625854,
      "learning_rate": 6.5822443308539e-06,
      "loss": 0.3893,
      "step": 19804
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5437477827072144,
      "learning_rate": 6.581287265144172e-06,
      "loss": 0.4537,
      "step": 19805
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.683060884475708,
      "learning_rate": 6.580330234891958e-06,
      "loss": 0.3923,
      "step": 19806
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7164698839187622,
      "learning_rate": 6.579373240107193e-06,
      "loss": 0.937,
      "step": 19807
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.445088505744934,
      "learning_rate": 6.578416280799798e-06,
      "loss": 0.9699,
      "step": 19808
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6209014654159546,
      "learning_rate": 6.577459356979694e-06,
      "loss": 0.4368,
      "step": 19809
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5548380613327026,
      "learning_rate": 6.576502468656811e-06,
      "loss": 0.3873,
      "step": 19810
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.54633629322052,
      "learning_rate": 6.57554561584107e-06,
      "loss": 0.4127,
      "step": 19811
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.604371428489685,
      "learning_rate": 6.5745887985424026e-06,
      "loss": 0.4018,
      "step": 19812
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.523976445198059,
      "learning_rate": 6.573632016770721e-06,
      "loss": 0.4517,
      "step": 19813
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.491217851638794,
      "learning_rate": 6.572675270535958e-06,
      "loss": 0.3907,
      "step": 19814
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6028647422790527,
      "learning_rate": 6.571718559848029e-06,
      "loss": 0.4531,
      "step": 19815
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4890251159667969,
      "learning_rate": 6.570761884716865e-06,
      "loss": 0.3681,
      "step": 19816
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.594659447669983,
      "learning_rate": 6.569805245152383e-06,
      "loss": 0.4428,
      "step": 19817
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4918304681777954,
      "learning_rate": 6.5688486411645025e-06,
      "loss": 0.3982,
      "step": 19818
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.572306752204895,
      "learning_rate": 6.5678920727631516e-06,
      "loss": 0.4357,
      "step": 19819
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4582420587539673,
      "learning_rate": 6.5669355399582476e-06,
      "loss": 0.3854,
      "step": 19820
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.637270450592041,
      "learning_rate": 6.56597904275971e-06,
      "loss": 0.4725,
      "step": 19821
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5543876886367798,
      "learning_rate": 6.565022581177462e-06,
      "loss": 0.3981,
      "step": 19822
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6809895038604736,
      "learning_rate": 6.564066155221423e-06,
      "loss": 0.4319,
      "step": 19823
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.587530493736267,
      "learning_rate": 6.563109764901514e-06,
      "loss": 0.4132,
      "step": 19824
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.582868218421936,
      "learning_rate": 6.562153410227651e-06,
      "loss": 0.4453,
      "step": 19825
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5436931848526,
      "learning_rate": 6.561197091209752e-06,
      "loss": 0.3862,
      "step": 19826
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6806315183639526,
      "learning_rate": 6.560240807857741e-06,
      "loss": 0.4863,
      "step": 19827
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.451805591583252,
      "learning_rate": 6.559284560181534e-06,
      "loss": 0.3705,
      "step": 19828
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6326324939727783,
      "learning_rate": 6.558328348191045e-06,
      "loss": 0.4656,
      "step": 19829
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4879732131958008,
      "learning_rate": 6.5573721718961965e-06,
      "loss": 0.3967,
      "step": 19830
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6350164413452148,
      "learning_rate": 6.556416031306903e-06,
      "loss": 0.4563,
      "step": 19831
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4857282638549805,
      "learning_rate": 6.5554599264330836e-06,
      "loss": 0.3725,
      "step": 19832
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.516058087348938,
      "learning_rate": 6.554503857284652e-06,
      "loss": 0.4547,
      "step": 19833
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5346883535385132,
      "learning_rate": 6.553547823871522e-06,
      "loss": 0.3906,
      "step": 19834
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5969576835632324,
      "learning_rate": 6.552591826203616e-06,
      "loss": 0.5212,
      "step": 19835
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5076274871826172,
      "learning_rate": 6.551635864290846e-06,
      "loss": 0.4061,
      "step": 19836
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8459819555282593,
      "learning_rate": 6.550679938143123e-06,
      "loss": 0.9543,
      "step": 19837
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5601532459259033,
      "learning_rate": 6.549724047770367e-06,
      "loss": 1.0922,
      "step": 19838
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5978405475616455,
      "learning_rate": 6.5487681931824865e-06,
      "loss": 0.4846,
      "step": 19839
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5192917585372925,
      "learning_rate": 6.547812374389404e-06,
      "loss": 0.3604,
      "step": 19840
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6681933403015137,
      "learning_rate": 6.546856591401025e-06,
      "loss": 0.5037,
      "step": 19841
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5167229175567627,
      "learning_rate": 6.545900844227263e-06,
      "loss": 0.3881,
      "step": 19842
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4924776554107666,
      "learning_rate": 6.544945132878036e-06,
      "loss": 0.4297,
      "step": 19843
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.537529468536377,
      "learning_rate": 6.543989457363253e-06,
      "loss": 0.3954,
      "step": 19844
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5557605028152466,
      "learning_rate": 6.543033817692822e-06,
      "loss": 0.4226,
      "step": 19845
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5774339437484741,
      "learning_rate": 6.542078213876661e-06,
      "loss": 0.4212,
      "step": 19846
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5034072399139404,
      "learning_rate": 6.541122645924676e-06,
      "loss": 0.4391,
      "step": 19847
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.564970850944519,
      "learning_rate": 6.540167113846786e-06,
      "loss": 0.4145,
      "step": 19848
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4774622917175293,
      "learning_rate": 6.53921161765289e-06,
      "loss": 0.3957,
      "step": 19849
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.448894739151001,
      "learning_rate": 6.5382561573529025e-06,
      "loss": 0.3952,
      "step": 19850
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5460221767425537,
      "learning_rate": 6.537300732956737e-06,
      "loss": 0.4161,
      "step": 19851
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5345124006271362,
      "learning_rate": 6.536345344474303e-06,
      "loss": 0.4148,
      "step": 19852
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5390641689300537,
      "learning_rate": 6.535389991915504e-06,
      "loss": 0.4547,
      "step": 19853
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.526814341545105,
      "learning_rate": 6.534434675290249e-06,
      "loss": 0.4037,
      "step": 19854
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5049965381622314,
      "learning_rate": 6.533479394608448e-06,
      "loss": 0.4463,
      "step": 19855
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5001200437545776,
      "learning_rate": 6.532524149880014e-06,
      "loss": 0.3931,
      "step": 19856
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5297534465789795,
      "learning_rate": 6.531568941114844e-06,
      "loss": 0.4535,
      "step": 19857
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5887068510055542,
      "learning_rate": 6.530613768322853e-06,
      "loss": 0.4136,
      "step": 19858
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5205605030059814,
      "learning_rate": 6.529658631513942e-06,
      "loss": 0.4501,
      "step": 19859
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.482208490371704,
      "learning_rate": 6.528703530698025e-06,
      "loss": 0.3674,
      "step": 19860
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6133520603179932,
      "learning_rate": 6.527748465885001e-06,
      "loss": 0.4557,
      "step": 19861
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6409335136413574,
      "learning_rate": 6.526793437084775e-06,
      "loss": 0.424,
      "step": 19862
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4848661422729492,
      "learning_rate": 6.525838444307259e-06,
      "loss": 0.4626,
      "step": 19863
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4881868362426758,
      "learning_rate": 6.524883487562354e-06,
      "loss": 0.3976,
      "step": 19864
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7560678720474243,
      "learning_rate": 6.523928566859962e-06,
      "loss": 0.4353,
      "step": 19865
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.54489004611969,
      "learning_rate": 6.52297368220999e-06,
      "loss": 0.394,
      "step": 19866
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6615177392959595,
      "learning_rate": 6.522018833622339e-06,
      "loss": 0.4406,
      "step": 19867
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5486141443252563,
      "learning_rate": 6.5210640211069175e-06,
      "loss": 0.4085,
      "step": 19868
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5832459926605225,
      "learning_rate": 6.520109244673625e-06,
      "loss": 0.428,
      "step": 19869
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5005760192871094,
      "learning_rate": 6.519154504332361e-06,
      "loss": 0.3932,
      "step": 19870
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4685174226760864,
      "learning_rate": 6.518199800093034e-06,
      "loss": 0.4616,
      "step": 19871
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5627764463424683,
      "learning_rate": 6.517245131965544e-06,
      "loss": 0.3903,
      "step": 19872
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4678895473480225,
      "learning_rate": 6.5162904999597876e-06,
      "loss": 0.4184,
      "step": 19873
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5825117826461792,
      "learning_rate": 6.515335904085671e-06,
      "loss": 0.4055,
      "step": 19874
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.696453332901001,
      "learning_rate": 6.5143813443530915e-06,
      "loss": 0.5,
      "step": 19875
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7031488418579102,
      "learning_rate": 6.513426820771955e-06,
      "loss": 0.391,
      "step": 19876
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5503172874450684,
      "learning_rate": 6.512472333352156e-06,
      "loss": 0.4251,
      "step": 19877
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5654456615447998,
      "learning_rate": 6.511517882103593e-06,
      "loss": 0.3986,
      "step": 19878
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5770249366760254,
      "learning_rate": 6.5105634670361705e-06,
      "loss": 0.431,
      "step": 19879
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5781726837158203,
      "learning_rate": 6.509609088159787e-06,
      "loss": 0.3841,
      "step": 19880
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5002657175064087,
      "learning_rate": 6.508654745484335e-06,
      "loss": 0.4144,
      "step": 19881
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.53670334815979,
      "learning_rate": 6.507700439019717e-06,
      "loss": 0.4142,
      "step": 19882
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5258327722549438,
      "learning_rate": 6.506746168775829e-06,
      "loss": 0.4018,
      "step": 19883
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4421935081481934,
      "learning_rate": 6.505791934762573e-06,
      "loss": 0.3719,
      "step": 19884
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6122522354125977,
      "learning_rate": 6.50483773698984e-06,
      "loss": 0.4494,
      "step": 19885
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5078198909759521,
      "learning_rate": 6.5038835754675265e-06,
      "loss": 0.3947,
      "step": 19886
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5187108516693115,
      "learning_rate": 6.502929450205533e-06,
      "loss": 0.395,
      "step": 19887
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.56746506690979,
      "learning_rate": 6.501975361213756e-06,
      "loss": 0.4115,
      "step": 19888
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5924220085144043,
      "learning_rate": 6.5010213085020854e-06,
      "loss": 0.4491,
      "step": 19889
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.8293824195861816,
      "learning_rate": 6.500067292080417e-06,
      "loss": 0.3932,
      "step": 19890
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.697461485862732,
      "learning_rate": 6.499113311958649e-06,
      "loss": 0.4452,
      "step": 19891
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5561680793762207,
      "learning_rate": 6.498159368146677e-06,
      "loss": 0.4137,
      "step": 19892
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.547062635421753,
      "learning_rate": 6.497205460654388e-06,
      "loss": 0.4287,
      "step": 19893
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5715444087982178,
      "learning_rate": 6.496251589491682e-06,
      "loss": 0.3901,
      "step": 19894
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5263466835021973,
      "learning_rate": 6.495297754668446e-06,
      "loss": 0.4478,
      "step": 19895
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.473811388015747,
      "learning_rate": 6.494343956194582e-06,
      "loss": 0.3933,
      "step": 19896
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5643727779388428,
      "learning_rate": 6.493390194079976e-06,
      "loss": 0.4246,
      "step": 19897
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5374174118041992,
      "learning_rate": 6.4924364683345175e-06,
      "loss": 0.3777,
      "step": 19898
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5664310455322266,
      "learning_rate": 6.491482778968103e-06,
      "loss": 0.4268,
      "step": 19899
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.525613784790039,
      "learning_rate": 6.490529125990626e-06,
      "loss": 0.4022,
      "step": 19900
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5716389417648315,
      "learning_rate": 6.489575509411969e-06,
      "loss": 0.4366,
      "step": 19901
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5645705461502075,
      "learning_rate": 6.488621929242031e-06,
      "loss": 0.4026,
      "step": 19902
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5144050121307373,
      "learning_rate": 6.487668385490695e-06,
      "loss": 0.4401,
      "step": 19903
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5373152494430542,
      "learning_rate": 6.486714878167859e-06,
      "loss": 0.4004,
      "step": 19904
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.629082441329956,
      "learning_rate": 6.485761407283404e-06,
      "loss": 0.4451,
      "step": 19905
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5255064964294434,
      "learning_rate": 6.4848079728472225e-06,
      "loss": 0.3899,
      "step": 19906
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4789656400680542,
      "learning_rate": 6.483854574869206e-06,
      "loss": 0.4212,
      "step": 19907
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.477665901184082,
      "learning_rate": 6.4829012133592405e-06,
      "loss": 0.3933,
      "step": 19908
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5923868417739868,
      "learning_rate": 6.481947888327211e-06,
      "loss": 0.5235,
      "step": 19909
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5773866176605225,
      "learning_rate": 6.480994599783008e-06,
      "loss": 0.4184,
      "step": 19910
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6062531471252441,
      "learning_rate": 6.480041347736518e-06,
      "loss": 0.4527,
      "step": 19911
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5512503385543823,
      "learning_rate": 6.479088132197631e-06,
      "loss": 0.3965,
      "step": 19912
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5706324577331543,
      "learning_rate": 6.478134953176227e-06,
      "loss": 0.4641,
      "step": 19913
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4458351135253906,
      "learning_rate": 6.477181810682196e-06,
      "loss": 0.3821,
      "step": 19914
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5907918214797974,
      "learning_rate": 6.476228704725423e-06,
      "loss": 0.4492,
      "step": 19915
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6369621753692627,
      "learning_rate": 6.475275635315796e-06,
      "loss": 0.3922,
      "step": 19916
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.589362621307373,
      "learning_rate": 6.474322602463192e-06,
      "loss": 0.4224,
      "step": 19917
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4766169786453247,
      "learning_rate": 6.473369606177503e-06,
      "loss": 0.4217,
      "step": 19918
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5773369073867798,
      "learning_rate": 6.472416646468609e-06,
      "loss": 0.4352,
      "step": 19919
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5364545583724976,
      "learning_rate": 6.471463723346399e-06,
      "loss": 0.3815,
      "step": 19920
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5426400899887085,
      "learning_rate": 6.4705108368207496e-06,
      "loss": 0.4291,
      "step": 19921
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4794522523880005,
      "learning_rate": 6.469557986901546e-06,
      "loss": 0.3782,
      "step": 19922
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.608970046043396,
      "learning_rate": 6.468605173598673e-06,
      "loss": 0.4313,
      "step": 19923
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5868300199508667,
      "learning_rate": 6.467652396922013e-06,
      "loss": 0.4104,
      "step": 19924
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5058951377868652,
      "learning_rate": 6.4666996568814435e-06,
      "loss": 0.4152,
      "step": 19925
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.476924180984497,
      "learning_rate": 6.465746953486849e-06,
      "loss": 0.3791,
      "step": 19926
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5790581703186035,
      "learning_rate": 6.464794286748108e-06,
      "loss": 0.4404,
      "step": 19927
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.551383137702942,
      "learning_rate": 6.463841656675112e-06,
      "loss": 0.3957,
      "step": 19928
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.526650309562683,
      "learning_rate": 6.4628890632777245e-06,
      "loss": 0.4176,
      "step": 19929
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5157551765441895,
      "learning_rate": 6.461936506565836e-06,
      "loss": 0.3907,
      "step": 19930
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.633979320526123,
      "learning_rate": 6.460983986549321e-06,
      "loss": 0.4643,
      "step": 19931
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.486032485961914,
      "learning_rate": 6.460031503238066e-06,
      "loss": 0.3875,
      "step": 19932
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.578624963760376,
      "learning_rate": 6.4590790566419425e-06,
      "loss": 0.4348,
      "step": 19933
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7753055095672607,
      "learning_rate": 6.458126646770829e-06,
      "loss": 0.428,
      "step": 19934
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.525496244430542,
      "learning_rate": 6.457174273634609e-06,
      "loss": 0.4235,
      "step": 19935
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4773643016815186,
      "learning_rate": 6.456221937243159e-06,
      "loss": 0.3975,
      "step": 19936
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5202234983444214,
      "learning_rate": 6.455269637606349e-06,
      "loss": 0.4247,
      "step": 19937
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5515737533569336,
      "learning_rate": 6.454317374734065e-06,
      "loss": 0.3748,
      "step": 19938
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5215452909469604,
      "learning_rate": 6.453365148636176e-06,
      "loss": 0.4219,
      "step": 19939
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5525201559066772,
      "learning_rate": 6.4524129593225675e-06,
      "loss": 0.4051,
      "step": 19940
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4986011981964111,
      "learning_rate": 6.4514608068031044e-06,
      "loss": 0.4029,
      "step": 19941
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.451469898223877,
      "learning_rate": 6.450508691087668e-06,
      "loss": 0.3896,
      "step": 19942
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5020878314971924,
      "learning_rate": 6.4495566121861344e-06,
      "loss": 0.4745,
      "step": 19943
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.61445152759552,
      "learning_rate": 6.448604570108377e-06,
      "loss": 0.4563,
      "step": 19944
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5336942672729492,
      "learning_rate": 6.447652564864265e-06,
      "loss": 0.4662,
      "step": 19945
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7269307374954224,
      "learning_rate": 6.446700596463679e-06,
      "loss": 0.405,
      "step": 19946
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6667699813842773,
      "learning_rate": 6.4457486649164866e-06,
      "loss": 0.4474,
      "step": 19947
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9696906805038452,
      "learning_rate": 6.444796770232571e-06,
      "loss": 0.4263,
      "step": 19948
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4936184883117676,
      "learning_rate": 6.443844912421793e-06,
      "loss": 0.4188,
      "step": 19949
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4965426921844482,
      "learning_rate": 6.442893091494029e-06,
      "loss": 0.423,
      "step": 19950
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4670255184173584,
      "learning_rate": 6.441941307459154e-06,
      "loss": 0.4275,
      "step": 19951
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4578176736831665,
      "learning_rate": 6.440989560327039e-06,
      "loss": 0.3766,
      "step": 19952
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6576018333435059,
      "learning_rate": 6.44003785010755e-06,
      "loss": 0.4718,
      "step": 19953
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6002963781356812,
      "learning_rate": 6.439086176810563e-06,
      "loss": 0.4032,
      "step": 19954
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5086387395858765,
      "learning_rate": 6.438134540445944e-06,
      "loss": 0.4429,
      "step": 19955
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5766890048980713,
      "learning_rate": 6.4371829410235685e-06,
      "loss": 0.4263,
      "step": 19956
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5625226497650146,
      "learning_rate": 6.436231378553303e-06,
      "loss": 0.4588,
      "step": 19957
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5878502130508423,
      "learning_rate": 6.4352798530450136e-06,
      "loss": 0.4084,
      "step": 19958
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5916837453842163,
      "learning_rate": 6.434328364508575e-06,
      "loss": 0.439,
      "step": 19959
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5503389835357666,
      "learning_rate": 6.433376912953855e-06,
      "loss": 0.3986,
      "step": 19960
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6872515678405762,
      "learning_rate": 6.4324254983907155e-06,
      "loss": 0.4797,
      "step": 19961
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4877017736434937,
      "learning_rate": 6.431474120829031e-06,
      "loss": 0.4022,
      "step": 19962
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5784804821014404,
      "learning_rate": 6.430522780278664e-06,
      "loss": 0.4646,
      "step": 19963
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5202877521514893,
      "learning_rate": 6.42957147674949e-06,
      "loss": 0.3884,
      "step": 19964
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5605567693710327,
      "learning_rate": 6.428620210251362e-06,
      "loss": 0.4163,
      "step": 19965
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5856515169143677,
      "learning_rate": 6.427668980794156e-06,
      "loss": 0.4163,
      "step": 19966
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4792777299880981,
      "learning_rate": 6.426717788387733e-06,
      "loss": 0.3794,
      "step": 19967
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4992401599884033,
      "learning_rate": 6.425766633041965e-06,
      "loss": 0.3947,
      "step": 19968
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5615642070770264,
      "learning_rate": 6.4248155147667105e-06,
      "loss": 0.4255,
      "step": 19969
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5324976444244385,
      "learning_rate": 6.423864433571833e-06,
      "loss": 0.4359,
      "step": 19970
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.529900074005127,
      "learning_rate": 6.422913389467202e-06,
      "loss": 0.4691,
      "step": 19971
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5330617427825928,
      "learning_rate": 6.421962382462683e-06,
      "loss": 0.4204,
      "step": 19972
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5754185914993286,
      "learning_rate": 6.42101141256813e-06,
      "loss": 0.4035,
      "step": 19973
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4552379846572876,
      "learning_rate": 6.420060479793415e-06,
      "loss": 0.3898,
      "step": 19974
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5767165422439575,
      "learning_rate": 6.419109584148395e-06,
      "loss": 0.4377,
      "step": 19975
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6212677955627441,
      "learning_rate": 6.418158725642938e-06,
      "loss": 0.3986,
      "step": 19976
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6327931880950928,
      "learning_rate": 6.417207904286902e-06,
      "loss": 0.4772,
      "step": 19977
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4737842082977295,
      "learning_rate": 6.416257120090146e-06,
      "loss": 0.4049,
      "step": 19978
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6273962259292603,
      "learning_rate": 6.415306373062537e-06,
      "loss": 0.427,
      "step": 19979
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.487618327140808,
      "learning_rate": 6.414355663213934e-06,
      "loss": 0.4113,
      "step": 19980
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5955109596252441,
      "learning_rate": 6.4134049905541925e-06,
      "loss": 0.4629,
      "step": 19981
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5486763715744019,
      "learning_rate": 6.412454355093181e-06,
      "loss": 0.3921,
      "step": 19982
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.59205162525177,
      "learning_rate": 6.4115037568407505e-06,
      "loss": 0.446,
      "step": 19983
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6436951160430908,
      "learning_rate": 6.410553195806768e-06,
      "loss": 0.422,
      "step": 19984
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6690548658370972,
      "learning_rate": 6.4096026720010876e-06,
      "loss": 0.4695,
      "step": 19985
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5707530975341797,
      "learning_rate": 6.408652185433567e-06,
      "loss": 0.4044,
      "step": 19986
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5985149145126343,
      "learning_rate": 6.4077017361140666e-06,
      "loss": 0.4755,
      "step": 19987
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.564929723739624,
      "learning_rate": 6.4067513240524485e-06,
      "loss": 0.3841,
      "step": 19988
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.530436396598816,
      "learning_rate": 6.405800949258558e-06,
      "loss": 0.4505,
      "step": 19989
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4866360425949097,
      "learning_rate": 6.404850611742263e-06,
      "loss": 0.397,
      "step": 19990
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6110557317733765,
      "learning_rate": 6.4039003115134135e-06,
      "loss": 0.4371,
      "step": 19991
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5640416145324707,
      "learning_rate": 6.402950048581873e-06,
      "loss": 0.3761,
      "step": 19992
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5353587865829468,
      "learning_rate": 6.40199982295749e-06,
      "loss": 0.431,
      "step": 19993
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8061294555664062,
      "learning_rate": 6.401049634650119e-06,
      "loss": 0.4146,
      "step": 19994
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7155492305755615,
      "learning_rate": 6.400099483669622e-06,
      "loss": 0.4595,
      "step": 19995
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5110650062561035,
      "learning_rate": 6.399149370025847e-06,
      "loss": 0.388,
      "step": 19996
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5213799476623535,
      "learning_rate": 6.398199293728655e-06,
      "loss": 0.4542,
      "step": 19997
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8754165172576904,
      "learning_rate": 6.397249254787894e-06,
      "loss": 0.414,
      "step": 19998
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5328885316848755,
      "learning_rate": 6.396299253213417e-06,
      "loss": 0.421,
      "step": 19999
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.802393913269043,
      "learning_rate": 6.395349289015082e-06,
      "loss": 0.386,
      "step": 20000
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.537107229232788,
      "learning_rate": 6.394399362202741e-06,
      "loss": 0.4375,
      "step": 20001
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5561963319778442,
      "learning_rate": 6.393449472786239e-06,
      "loss": 0.3988,
      "step": 20002
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7046945095062256,
      "learning_rate": 6.392499620775438e-06,
      "loss": 0.419,
      "step": 20003
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5295534133911133,
      "learning_rate": 6.391549806180179e-06,
      "loss": 0.4007,
      "step": 20004
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5431222915649414,
      "learning_rate": 6.390600029010327e-06,
      "loss": 0.4465,
      "step": 20005
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5035994052886963,
      "learning_rate": 6.38965028927572e-06,
      "loss": 0.414,
      "step": 20006
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.557044267654419,
      "learning_rate": 6.388700586986212e-06,
      "loss": 0.4527,
      "step": 20007
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5856049060821533,
      "learning_rate": 6.387750922151652e-06,
      "loss": 0.4074,
      "step": 20008
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.862329125404358,
      "learning_rate": 6.3868012947818955e-06,
      "loss": 0.4604,
      "step": 20009
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.468347430229187,
      "learning_rate": 6.385851704886785e-06,
      "loss": 0.3747,
      "step": 20010
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5033551454544067,
      "learning_rate": 6.384902152476171e-06,
      "loss": 0.4426,
      "step": 20011
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.506553053855896,
      "learning_rate": 6.383952637559903e-06,
      "loss": 0.3903,
      "step": 20012
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5230765342712402,
      "learning_rate": 6.383003160147831e-06,
      "loss": 0.4438,
      "step": 20013
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5706974267959595,
      "learning_rate": 6.382053720249797e-06,
      "loss": 0.4087,
      "step": 20014
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.676421880722046,
      "learning_rate": 6.381104317875653e-06,
      "loss": 0.4264,
      "step": 20015
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.518011212348938,
      "learning_rate": 6.3801549530352415e-06,
      "loss": 0.3863,
      "step": 20016
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5274102687835693,
      "learning_rate": 6.3792056257384165e-06,
      "loss": 0.4286,
      "step": 20017
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6149365901947021,
      "learning_rate": 6.378256335995016e-06,
      "loss": 0.4104,
      "step": 20018
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.507062315940857,
      "learning_rate": 6.377307083814887e-06,
      "loss": 0.4298,
      "step": 20019
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5236876010894775,
      "learning_rate": 6.376357869207879e-06,
      "loss": 0.3974,
      "step": 20020
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5359265804290771,
      "learning_rate": 6.3754086921838355e-06,
      "loss": 0.4472,
      "step": 20021
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5818381309509277,
      "learning_rate": 6.374459552752597e-06,
      "loss": 0.3847,
      "step": 20022
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5447049140930176,
      "learning_rate": 6.373510450924011e-06,
      "loss": 0.4359,
      "step": 20023
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.638717770576477,
      "learning_rate": 6.37256138670792e-06,
      "loss": 0.3943,
      "step": 20024
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5223791599273682,
      "learning_rate": 6.371612360114171e-06,
      "loss": 0.4234,
      "step": 20025
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8485873937606812,
      "learning_rate": 6.370663371152602e-06,
      "loss": 0.423,
      "step": 20026
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.791918396949768,
      "learning_rate": 6.369714419833056e-06,
      "loss": 0.456,
      "step": 20027
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4902597665786743,
      "learning_rate": 6.368765506165379e-06,
      "loss": 0.3997,
      "step": 20028
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4721622467041016,
      "learning_rate": 6.367816630159411e-06,
      "loss": 0.4424,
      "step": 20029
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.501056432723999,
      "learning_rate": 6.366867791824991e-06,
      "loss": 0.3912,
      "step": 20030
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5077048540115356,
      "learning_rate": 6.365918991171962e-06,
      "loss": 0.4155,
      "step": 20031
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4920541048049927,
      "learning_rate": 6.364970228210161e-06,
      "loss": 0.3797,
      "step": 20032
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5089572668075562,
      "learning_rate": 6.364021502949438e-06,
      "loss": 0.4297,
      "step": 20033
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5505528450012207,
      "learning_rate": 6.363072815399623e-06,
      "loss": 0.443,
      "step": 20034
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5739703178405762,
      "learning_rate": 6.362124165570557e-06,
      "loss": 0.4271,
      "step": 20035
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6581192016601562,
      "learning_rate": 6.361175553472082e-06,
      "loss": 0.4177,
      "step": 20036
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.548146367073059,
      "learning_rate": 6.360226979114038e-06,
      "loss": 0.4422,
      "step": 20037
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5220001935958862,
      "learning_rate": 6.359278442506256e-06,
      "loss": 0.3959,
      "step": 20038
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.585677146911621,
      "learning_rate": 6.358329943658581e-06,
      "loss": 0.4554,
      "step": 20039
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5481436252593994,
      "learning_rate": 6.357381482580844e-06,
      "loss": 0.4114,
      "step": 20040
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6609501838684082,
      "learning_rate": 6.356433059282894e-06,
      "loss": 0.4388,
      "step": 20041
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5374773740768433,
      "learning_rate": 6.355484673774553e-06,
      "loss": 0.3976,
      "step": 20042
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5448282957077026,
      "learning_rate": 6.3545363260656634e-06,
      "loss": 0.4342,
      "step": 20043
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5032881498336792,
      "learning_rate": 6.3535880161660635e-06,
      "loss": 0.3916,
      "step": 20044
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.608760952949524,
      "learning_rate": 6.3526397440855894e-06,
      "loss": 0.4631,
      "step": 20045
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5746835470199585,
      "learning_rate": 6.351691509834072e-06,
      "loss": 0.4096,
      "step": 20046
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5727956295013428,
      "learning_rate": 6.350743313421344e-06,
      "loss": 0.4932,
      "step": 20047
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.447007417678833,
      "learning_rate": 6.349795154857245e-06,
      "loss": 0.3846,
      "step": 20048
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.512376308441162,
      "learning_rate": 6.3488470341516115e-06,
      "loss": 0.4562,
      "step": 20049
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5030874013900757,
      "learning_rate": 6.347898951314268e-06,
      "loss": 0.3816,
      "step": 20050
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5631674528121948,
      "learning_rate": 6.346950906355054e-06,
      "loss": 0.4884,
      "step": 20051
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5084033012390137,
      "learning_rate": 6.346002899283799e-06,
      "loss": 0.427,
      "step": 20052
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6486570835113525,
      "learning_rate": 6.345054930110339e-06,
      "loss": 0.4623,
      "step": 20053
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4771136045455933,
      "learning_rate": 6.344106998844504e-06,
      "loss": 0.376,
      "step": 20054
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5389972925186157,
      "learning_rate": 6.343159105496123e-06,
      "loss": 0.4683,
      "step": 20055
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5890003442764282,
      "learning_rate": 6.342211250075031e-06,
      "loss": 0.3953,
      "step": 20056
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.625327706336975,
      "learning_rate": 6.3412634325910605e-06,
      "loss": 0.4806,
      "step": 20057
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5204577445983887,
      "learning_rate": 6.340315653054035e-06,
      "loss": 0.395,
      "step": 20058
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5635018348693848,
      "learning_rate": 6.339367911473788e-06,
      "loss": 0.4536,
      "step": 20059
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.507857322692871,
      "learning_rate": 6.3384202078601496e-06,
      "loss": 0.4084,
      "step": 20060
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5194172859191895,
      "learning_rate": 6.337472542222953e-06,
      "loss": 0.486,
      "step": 20061
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.455000638961792,
      "learning_rate": 6.33652491457202e-06,
      "loss": 0.392,
      "step": 20062
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5410884618759155,
      "learning_rate": 6.335577324917179e-06,
      "loss": 0.4262,
      "step": 20063
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4898059368133545,
      "learning_rate": 6.334629773268264e-06,
      "loss": 0.3992,
      "step": 20064
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5678956508636475,
      "learning_rate": 6.333682259635101e-06,
      "loss": 0.444,
      "step": 20065
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5318646430969238,
      "learning_rate": 6.332734784027512e-06,
      "loss": 0.3853,
      "step": 20066
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7191044092178345,
      "learning_rate": 6.331787346455329e-06,
      "loss": 0.416,
      "step": 20067
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4790401458740234,
      "learning_rate": 6.3308399469283755e-06,
      "loss": 0.3684,
      "step": 20068
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.456160545349121,
      "learning_rate": 6.329892585456483e-06,
      "loss": 0.4075,
      "step": 20069
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.555729627609253,
      "learning_rate": 6.3289452620494715e-06,
      "loss": 0.427,
      "step": 20070
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5985381603240967,
      "learning_rate": 6.327997976717166e-06,
      "loss": 0.4213,
      "step": 20071
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5712169408798218,
      "learning_rate": 6.3270507294693954e-06,
      "loss": 0.4074,
      "step": 20072
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5265982151031494,
      "learning_rate": 6.326103520315983e-06,
      "loss": 0.435,
      "step": 20073
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4759312868118286,
      "learning_rate": 6.325156349266749e-06,
      "loss": 0.3963,
      "step": 20074
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5983644723892212,
      "learning_rate": 6.32420921633152e-06,
      "loss": 0.5001,
      "step": 20075
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6071693897247314,
      "learning_rate": 6.32326212152012e-06,
      "loss": 0.4128,
      "step": 20076
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.580947995185852,
      "learning_rate": 6.322315064842375e-06,
      "loss": 0.4468,
      "step": 20077
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6136521100997925,
      "learning_rate": 6.3213680463081e-06,
      "loss": 0.4299,
      "step": 20078
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6706010103225708,
      "learning_rate": 6.32042106592712e-06,
      "loss": 0.5063,
      "step": 20079
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.507478952407837,
      "learning_rate": 6.319474123709259e-06,
      "loss": 0.3997,
      "step": 20080
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8195127248764038,
      "learning_rate": 6.318527219664338e-06,
      "loss": 0.9693,
      "step": 20081
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5118719339370728,
      "learning_rate": 6.317580353802176e-06,
      "loss": 0.9738,
      "step": 20082
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5873348712921143,
      "learning_rate": 6.316633526132592e-06,
      "loss": 0.483,
      "step": 20083
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5422290563583374,
      "learning_rate": 6.315686736665409e-06,
      "loss": 0.4146,
      "step": 20084
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5483067035675049,
      "learning_rate": 6.3147399854104505e-06,
      "loss": 0.4485,
      "step": 20085
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7064456939697266,
      "learning_rate": 6.3137932723775266e-06,
      "loss": 0.4133,
      "step": 20086
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6328506469726562,
      "learning_rate": 6.312846597576462e-06,
      "loss": 0.4769,
      "step": 20087
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4269899129867554,
      "learning_rate": 6.311899961017073e-06,
      "loss": 0.3796,
      "step": 20088
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5204464197158813,
      "learning_rate": 6.310953362709182e-06,
      "loss": 0.4571,
      "step": 20089
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5734785795211792,
      "learning_rate": 6.310006802662603e-06,
      "loss": 0.386,
      "step": 20090
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.523084044456482,
      "learning_rate": 6.309060280887151e-06,
      "loss": 0.4359,
      "step": 20091
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.49177086353302,
      "learning_rate": 6.308113797392649e-06,
      "loss": 0.4016,
      "step": 20092
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5252882242202759,
      "learning_rate": 6.307167352188912e-06,
      "loss": 0.4246,
      "step": 20093
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5059185028076172,
      "learning_rate": 6.306220945285751e-06,
      "loss": 0.3879,
      "step": 20094
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4994769096374512,
      "learning_rate": 6.305274576692987e-06,
      "loss": 0.4166,
      "step": 20095
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5660724639892578,
      "learning_rate": 6.304328246420432e-06,
      "loss": 0.3984,
      "step": 20096
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5380655527114868,
      "learning_rate": 6.303381954477906e-06,
      "loss": 0.4327,
      "step": 20097
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5468138456344604,
      "learning_rate": 6.30243570087522e-06,
      "loss": 0.408,
      "step": 20098
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5638606548309326,
      "learning_rate": 6.301489485622185e-06,
      "loss": 0.4205,
      "step": 20099
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4831074476242065,
      "learning_rate": 6.30054330872862e-06,
      "loss": 0.3872,
      "step": 20100
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5631437301635742,
      "learning_rate": 6.29959717020434e-06,
      "loss": 0.4516,
      "step": 20101
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.476525902748108,
      "learning_rate": 6.298651070059149e-06,
      "loss": 0.4191,
      "step": 20102
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4995478391647339,
      "learning_rate": 6.29770500830287e-06,
      "loss": 0.4268,
      "step": 20103
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5543547868728638,
      "learning_rate": 6.296758984945306e-06,
      "loss": 0.4087,
      "step": 20104
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6566874980926514,
      "learning_rate": 6.295812999996278e-06,
      "loss": 0.478,
      "step": 20105
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5370935201644897,
      "learning_rate": 6.29486705346559e-06,
      "loss": 0.3925,
      "step": 20106
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.633715033531189,
      "learning_rate": 6.2939211453630535e-06,
      "loss": 0.4373,
      "step": 20107
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5187987089157104,
      "learning_rate": 6.292975275698483e-06,
      "loss": 0.4093,
      "step": 20108
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6484521627426147,
      "learning_rate": 6.292029444481689e-06,
      "loss": 0.5315,
      "step": 20109
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5009222030639648,
      "learning_rate": 6.291083651722474e-06,
      "loss": 0.3999,
      "step": 20110
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.54616379737854,
      "learning_rate": 6.290137897430655e-06,
      "loss": 0.4225,
      "step": 20111
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6476327180862427,
      "learning_rate": 6.289192181616037e-06,
      "loss": 0.4002,
      "step": 20112
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5477962493896484,
      "learning_rate": 6.288246504288434e-06,
      "loss": 0.4243,
      "step": 20113
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.600324034690857,
      "learning_rate": 6.287300865457648e-06,
      "loss": 0.3945,
      "step": 20114
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.542083740234375,
      "learning_rate": 6.286355265133486e-06,
      "loss": 0.4536,
      "step": 20115
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.505388617515564,
      "learning_rate": 6.285409703325761e-06,
      "loss": 0.3886,
      "step": 20116
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6069337129592896,
      "learning_rate": 6.2844641800442784e-06,
      "loss": 0.5184,
      "step": 20117
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.502429723739624,
      "learning_rate": 6.2835186952988415e-06,
      "loss": 0.4115,
      "step": 20118
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6025569438934326,
      "learning_rate": 6.28257324909926e-06,
      "loss": 0.4653,
      "step": 20119
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4865800142288208,
      "learning_rate": 6.281627841455336e-06,
      "loss": 0.3922,
      "step": 20120
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.594651699066162,
      "learning_rate": 6.280682472376884e-06,
      "loss": 0.4743,
      "step": 20121
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.52164626121521,
      "learning_rate": 6.279737141873694e-06,
      "loss": 0.3909,
      "step": 20122
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5077271461486816,
      "learning_rate": 6.278791849955583e-06,
      "loss": 0.4502,
      "step": 20123
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4967432022094727,
      "learning_rate": 6.2778465966323485e-06,
      "loss": 0.3956,
      "step": 20124
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4972171783447266,
      "learning_rate": 6.2769013819138e-06,
      "loss": 0.4022,
      "step": 20125
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.545235276222229,
      "learning_rate": 6.275956205809736e-06,
      "loss": 0.4021,
      "step": 20126
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6037828922271729,
      "learning_rate": 6.275011068329962e-06,
      "loss": 0.4452,
      "step": 20127
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4639040231704712,
      "learning_rate": 6.274065969484279e-06,
      "loss": 0.3819,
      "step": 20128
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6257810592651367,
      "learning_rate": 6.273120909282493e-06,
      "loss": 0.4674,
      "step": 20129
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5152819156646729,
      "learning_rate": 6.272175887734397e-06,
      "loss": 0.3995,
      "step": 20130
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5013396739959717,
      "learning_rate": 6.271230904849801e-06,
      "loss": 0.4279,
      "step": 20131
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5145933628082275,
      "learning_rate": 6.2702859606385026e-06,
      "loss": 0.3866,
      "step": 20132
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5444258451461792,
      "learning_rate": 6.2693410551103055e-06,
      "loss": 0.4726,
      "step": 20133
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6065596342086792,
      "learning_rate": 6.2683961882750055e-06,
      "loss": 0.3931,
      "step": 20134
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5413610935211182,
      "learning_rate": 6.267451360142403e-06,
      "loss": 0.4781,
      "step": 20135
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.602367877960205,
      "learning_rate": 6.2665065707223e-06,
      "loss": 0.4246,
      "step": 20136
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4579743146896362,
      "learning_rate": 6.265561820024495e-06,
      "loss": 0.4644,
      "step": 20137
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5095524787902832,
      "learning_rate": 6.2646171080587815e-06,
      "loss": 0.3856,
      "step": 20138
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5093727111816406,
      "learning_rate": 6.263672434834966e-06,
      "loss": 0.4549,
      "step": 20139
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5115211009979248,
      "learning_rate": 6.262727800362838e-06,
      "loss": 0.4003,
      "step": 20140
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.667714238166809,
      "learning_rate": 6.261783204652203e-06,
      "loss": 0.4524,
      "step": 20141
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5104386806488037,
      "learning_rate": 6.260838647712852e-06,
      "loss": 0.3979,
      "step": 20142
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5062158107757568,
      "learning_rate": 6.259894129554582e-06,
      "loss": 0.4388,
      "step": 20143
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.575614094734192,
      "learning_rate": 6.258949650187194e-06,
      "loss": 0.4045,
      "step": 20144
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4839199781417847,
      "learning_rate": 6.25800520962048e-06,
      "loss": 0.4234,
      "step": 20145
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4680449962615967,
      "learning_rate": 6.257060807864232e-06,
      "loss": 0.375,
      "step": 20146
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5779826641082764,
      "learning_rate": 6.256116444928253e-06,
      "loss": 0.4341,
      "step": 20147
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5088685750961304,
      "learning_rate": 6.255172120822328e-06,
      "loss": 0.4194,
      "step": 20148
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6359524726867676,
      "learning_rate": 6.254227835556264e-06,
      "loss": 0.4546,
      "step": 20149
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4958215951919556,
      "learning_rate": 6.253283589139842e-06,
      "loss": 0.3695,
      "step": 20150
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5150842666625977,
      "learning_rate": 6.252339381582859e-06,
      "loss": 0.4533,
      "step": 20151
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.435341715812683,
      "learning_rate": 6.251395212895113e-06,
      "loss": 0.3918,
      "step": 20152
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6477550268173218,
      "learning_rate": 6.2504510830863925e-06,
      "loss": 0.465,
      "step": 20153
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4726214408874512,
      "learning_rate": 6.2495069921664886e-06,
      "loss": 0.3765,
      "step": 20154
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5650743246078491,
      "learning_rate": 6.2485629401451954e-06,
      "loss": 0.4378,
      "step": 20155
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6023869514465332,
      "learning_rate": 6.2476189270323015e-06,
      "loss": 0.4277,
      "step": 20156
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6337552070617676,
      "learning_rate": 6.246674952837606e-06,
      "loss": 0.4548,
      "step": 20157
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5911012887954712,
      "learning_rate": 6.245731017570886e-06,
      "loss": 0.4153,
      "step": 20158
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5149627923965454,
      "learning_rate": 6.244787121241942e-06,
      "loss": 0.4334,
      "step": 20159
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5919004678726196,
      "learning_rate": 6.2438432638605585e-06,
      "loss": 0.3852,
      "step": 20160
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7243999242782593,
      "learning_rate": 6.24289944543653e-06,
      "loss": 0.4775,
      "step": 20161
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5336956977844238,
      "learning_rate": 6.2419556659796395e-06,
      "loss": 0.3926,
      "step": 20162
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.640551209449768,
      "learning_rate": 6.2410119254996774e-06,
      "loss": 0.4736,
      "step": 20163
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5743976831436157,
      "learning_rate": 6.240068224006435e-06,
      "loss": 0.4317,
      "step": 20164
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6260533332824707,
      "learning_rate": 6.239124561509698e-06,
      "loss": 0.4567,
      "step": 20165
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5465887784957886,
      "learning_rate": 6.23818093801925e-06,
      "loss": 0.3719,
      "step": 20166
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5623661279678345,
      "learning_rate": 6.237237353544882e-06,
      "loss": 0.4463,
      "step": 20167
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6407442092895508,
      "learning_rate": 6.236293808096378e-06,
      "loss": 0.406,
      "step": 20168
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.525288701057434,
      "learning_rate": 6.2353503016835315e-06,
      "loss": 0.4632,
      "step": 20169
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6382038593292236,
      "learning_rate": 6.23440683431612e-06,
      "loss": 0.3964,
      "step": 20170
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6089164018630981,
      "learning_rate": 6.233463406003928e-06,
      "loss": 0.4668,
      "step": 20171
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.793260097503662,
      "learning_rate": 6.232520016756746e-06,
      "loss": 0.3837,
      "step": 20172
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5352858304977417,
      "learning_rate": 6.231576666584358e-06,
      "loss": 0.419,
      "step": 20173
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6267575025558472,
      "learning_rate": 6.230633355496542e-06,
      "loss": 0.4308,
      "step": 20174
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5626590251922607,
      "learning_rate": 6.229690083503086e-06,
      "loss": 0.3932,
      "step": 20175
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5321943759918213,
      "learning_rate": 6.2287468506137724e-06,
      "loss": 0.397,
      "step": 20176
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5687963962554932,
      "learning_rate": 6.2278036568383894e-06,
      "loss": 0.4525,
      "step": 20177
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5614991188049316,
      "learning_rate": 6.226860502186711e-06,
      "loss": 0.4191,
      "step": 20178
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.531010389328003,
      "learning_rate": 6.225917386668521e-06,
      "loss": 0.4283,
      "step": 20179
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4912232160568237,
      "learning_rate": 6.224974310293605e-06,
      "loss": 0.3905,
      "step": 20180
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5821856260299683,
      "learning_rate": 6.224031273071745e-06,
      "loss": 0.453,
      "step": 20181
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.50992751121521,
      "learning_rate": 6.223088275012712e-06,
      "loss": 0.3826,
      "step": 20182
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5841567516326904,
      "learning_rate": 6.222145316126298e-06,
      "loss": 0.4442,
      "step": 20183
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5591000318527222,
      "learning_rate": 6.221202396422274e-06,
      "loss": 0.3909,
      "step": 20184
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5320676565170288,
      "learning_rate": 6.220259515910429e-06,
      "loss": 0.4053,
      "step": 20185
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.46953284740448,
      "learning_rate": 6.219316674600535e-06,
      "loss": 0.3938,
      "step": 20186
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.657645344734192,
      "learning_rate": 6.21837387250237e-06,
      "loss": 0.5232,
      "step": 20187
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.524473786354065,
      "learning_rate": 6.217431109625718e-06,
      "loss": 0.4058,
      "step": 20188
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5726072788238525,
      "learning_rate": 6.2164883859803545e-06,
      "loss": 0.4263,
      "step": 20189
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5014288425445557,
      "learning_rate": 6.215545701576055e-06,
      "loss": 0.3657,
      "step": 20190
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5206843614578247,
      "learning_rate": 6.214603056422599e-06,
      "loss": 0.4574,
      "step": 20191
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5539073944091797,
      "learning_rate": 6.21366045052976e-06,
      "loss": 0.4198,
      "step": 20192
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5686249732971191,
      "learning_rate": 6.212717883907321e-06,
      "loss": 0.4098,
      "step": 20193
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.666253924369812,
      "learning_rate": 6.211775356565052e-06,
      "loss": 0.4012,
      "step": 20194
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7900450229644775,
      "learning_rate": 6.210832868512729e-06,
      "loss": 0.426,
      "step": 20195
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5728763341903687,
      "learning_rate": 6.209890419760127e-06,
      "loss": 0.4131,
      "step": 20196
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4728251695632935,
      "learning_rate": 6.208948010317028e-06,
      "loss": 0.4137,
      "step": 20197
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5282326936721802,
      "learning_rate": 6.208005640193195e-06,
      "loss": 0.3932,
      "step": 20198
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5654566287994385,
      "learning_rate": 6.207063309398409e-06,
      "loss": 0.4172,
      "step": 20199
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5448740720748901,
      "learning_rate": 6.206121017942441e-06,
      "loss": 0.4082,
      "step": 20200
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4960581064224243,
      "learning_rate": 6.205178765835067e-06,
      "loss": 0.385,
      "step": 20201
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5404086112976074,
      "learning_rate": 6.204236553086054e-06,
      "loss": 0.3981,
      "step": 20202
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.682281732559204,
      "learning_rate": 6.20329437970518e-06,
      "loss": 0.4678,
      "step": 20203
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5910046100616455,
      "learning_rate": 6.202352245702212e-06,
      "loss": 0.4231,
      "step": 20204
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5601880550384521,
      "learning_rate": 6.201410151086928e-06,
      "loss": 0.4187,
      "step": 20205
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4784127473831177,
      "learning_rate": 6.200468095869092e-06,
      "loss": 0.4322,
      "step": 20206
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6011239290237427,
      "learning_rate": 6.199526080058478e-06,
      "loss": 0.431,
      "step": 20207
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5033068656921387,
      "learning_rate": 6.1985841036648555e-06,
      "loss": 0.3962,
      "step": 20208
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5950632095336914,
      "learning_rate": 6.197642166697997e-06,
      "loss": 0.5292,
      "step": 20209
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5798128843307495,
      "learning_rate": 6.196700269167667e-06,
      "loss": 0.4264,
      "step": 20210
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6136614084243774,
      "learning_rate": 6.195758411083638e-06,
      "loss": 0.4747,
      "step": 20211
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5792747735977173,
      "learning_rate": 6.1948165924556745e-06,
      "loss": 0.4149,
      "step": 20212
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5444726943969727,
      "learning_rate": 6.193874813293555e-06,
      "loss": 0.4293,
      "step": 20213
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.465626835823059,
      "learning_rate": 6.192933073607035e-06,
      "loss": 0.4034,
      "step": 20214
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4863828420639038,
      "learning_rate": 6.191991373405886e-06,
      "loss": 0.3892,
      "step": 20215
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4909007549285889,
      "learning_rate": 6.191049712699877e-06,
      "loss": 0.3988,
      "step": 20216
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7373558282852173,
      "learning_rate": 6.190108091498775e-06,
      "loss": 0.4486,
      "step": 20217
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.477994441986084,
      "learning_rate": 6.189166509812341e-06,
      "loss": 0.3708,
      "step": 20218
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6969869136810303,
      "learning_rate": 6.188224967650347e-06,
      "loss": 0.5527,
      "step": 20219
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5341393947601318,
      "learning_rate": 6.187283465022552e-06,
      "loss": 0.3865,
      "step": 20220
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5372573137283325,
      "learning_rate": 6.18634200193873e-06,
      "loss": 0.4455,
      "step": 20221
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5809979438781738,
      "learning_rate": 6.185400578408636e-06,
      "loss": 0.4016,
      "step": 20222
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6657633781433105,
      "learning_rate": 6.184459194442036e-06,
      "loss": 0.4222,
      "step": 20223
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5622209310531616,
      "learning_rate": 6.183517850048699e-06,
      "loss": 0.3923,
      "step": 20224
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5344526767730713,
      "learning_rate": 6.182576545238385e-06,
      "loss": 0.4669,
      "step": 20225
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4679415225982666,
      "learning_rate": 6.181635280020854e-06,
      "loss": 0.3801,
      "step": 20226
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.619311809539795,
      "learning_rate": 6.180694054405873e-06,
      "loss": 0.459,
      "step": 20227
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5505974292755127,
      "learning_rate": 6.179752868403199e-06,
      "loss": 0.4099,
      "step": 20228
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.533016324043274,
      "learning_rate": 6.178811722022602e-06,
      "loss": 0.4388,
      "step": 20229
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4374725818634033,
      "learning_rate": 6.177870615273835e-06,
      "loss": 0.3726,
      "step": 20230
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5260275602340698,
      "learning_rate": 6.176929548166659e-06,
      "loss": 0.4626,
      "step": 20231
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4708327054977417,
      "learning_rate": 6.175988520710841e-06,
      "loss": 0.3804,
      "step": 20232
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5686733722686768,
      "learning_rate": 6.175047532916137e-06,
      "loss": 0.4696,
      "step": 20233
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5554113388061523,
      "learning_rate": 6.174106584792305e-06,
      "loss": 0.4171,
      "step": 20234
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4981456995010376,
      "learning_rate": 6.173165676349103e-06,
      "loss": 0.422,
      "step": 20235
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6059274673461914,
      "learning_rate": 6.172224807596294e-06,
      "loss": 0.3971,
      "step": 20236
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5597050189971924,
      "learning_rate": 6.171283978543635e-06,
      "loss": 0.4376,
      "step": 20237
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6727327108383179,
      "learning_rate": 6.170343189200887e-06,
      "loss": 0.413,
      "step": 20238
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5990434885025024,
      "learning_rate": 6.169402439577801e-06,
      "loss": 0.4967,
      "step": 20239
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.602255940437317,
      "learning_rate": 6.168461729684136e-06,
      "loss": 0.4278,
      "step": 20240
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5933855772018433,
      "learning_rate": 6.167521059529651e-06,
      "loss": 0.4323,
      "step": 20241
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5393105745315552,
      "learning_rate": 6.166580429124103e-06,
      "loss": 0.3961,
      "step": 20242
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.560457706451416,
      "learning_rate": 6.165639838477242e-06,
      "loss": 0.483,
      "step": 20243
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5994060039520264,
      "learning_rate": 6.1646992875988295e-06,
      "loss": 0.4159,
      "step": 20244
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.564299464225769,
      "learning_rate": 6.163758776498616e-06,
      "loss": 0.4712,
      "step": 20245
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.571444034576416,
      "learning_rate": 6.1628183051863645e-06,
      "loss": 0.3946,
      "step": 20246
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6289732456207275,
      "learning_rate": 6.161877873671819e-06,
      "loss": 0.532,
      "step": 20247
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4778811931610107,
      "learning_rate": 6.160937481964735e-06,
      "loss": 0.3901,
      "step": 20248
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5898185968399048,
      "learning_rate": 6.159997130074872e-06,
      "loss": 0.4943,
      "step": 20249
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.453952670097351,
      "learning_rate": 6.15905681801198e-06,
      "loss": 0.3779,
      "step": 20250
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5015549659729004,
      "learning_rate": 6.158116545785808e-06,
      "loss": 0.3789,
      "step": 20251
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.497483491897583,
      "learning_rate": 6.1571763134061125e-06,
      "loss": 0.3946,
      "step": 20252
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5327593088150024,
      "learning_rate": 6.156236120882641e-06,
      "loss": 0.456,
      "step": 20253
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5099066495895386,
      "learning_rate": 6.1552959682251515e-06,
      "loss": 0.3767,
      "step": 20254
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6037999391555786,
      "learning_rate": 6.154355855443388e-06,
      "loss": 0.4304,
      "step": 20255
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4809294939041138,
      "learning_rate": 6.153415782547103e-06,
      "loss": 0.4242,
      "step": 20256
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5525027513504028,
      "learning_rate": 6.15247574954605e-06,
      "loss": 0.4415,
      "step": 20257
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5305671691894531,
      "learning_rate": 6.151535756449977e-06,
      "loss": 0.3913,
      "step": 20258
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5566976070404053,
      "learning_rate": 6.150595803268628e-06,
      "loss": 0.4247,
      "step": 20259
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4663466215133667,
      "learning_rate": 6.149655890011759e-06,
      "loss": 0.3853,
      "step": 20260
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.490134835243225,
      "learning_rate": 6.148716016689111e-06,
      "loss": 0.4022,
      "step": 20261
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.529327392578125,
      "learning_rate": 6.147776183310443e-06,
      "loss": 0.4111,
      "step": 20262
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5285168886184692,
      "learning_rate": 6.146836389885493e-06,
      "loss": 0.5081,
      "step": 20263
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5309581756591797,
      "learning_rate": 6.145896636424009e-06,
      "loss": 0.421,
      "step": 20264
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6324858665466309,
      "learning_rate": 6.144956922935743e-06,
      "loss": 0.4524,
      "step": 20265
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5257530212402344,
      "learning_rate": 6.144017249430439e-06,
      "loss": 0.3951,
      "step": 20266
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5592763423919678,
      "learning_rate": 6.1430776159178384e-06,
      "loss": 0.4589,
      "step": 20267
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5101683139801025,
      "learning_rate": 6.142138022407693e-06,
      "loss": 0.4463,
      "step": 20268
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5490950345993042,
      "learning_rate": 6.1411984689097436e-06,
      "loss": 0.4426,
      "step": 20269
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.661624789237976,
      "learning_rate": 6.14025895543374e-06,
      "loss": 0.4073,
      "step": 20270
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7549654245376587,
      "learning_rate": 6.1393194819894205e-06,
      "loss": 0.438,
      "step": 20271
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.528399109840393,
      "learning_rate": 6.13838004858653e-06,
      "loss": 0.3967,
      "step": 20272
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5372263193130493,
      "learning_rate": 6.137440655234815e-06,
      "loss": 0.4415,
      "step": 20273
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6141287088394165,
      "learning_rate": 6.13650130194402e-06,
      "loss": 0.453,
      "step": 20274
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4807575941085815,
      "learning_rate": 6.135561988723882e-06,
      "loss": 0.4107,
      "step": 20275
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.539732813835144,
      "learning_rate": 6.1346227155841444e-06,
      "loss": 0.4501,
      "step": 20276
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6346747875213623,
      "learning_rate": 6.133683482534552e-06,
      "loss": 0.4765,
      "step": 20277
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4914597272872925,
      "learning_rate": 6.1327442895848455e-06,
      "loss": 0.3852,
      "step": 20278
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5427637100219727,
      "learning_rate": 6.131805136744762e-06,
      "loss": 0.4105,
      "step": 20279
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4702121019363403,
      "learning_rate": 6.130866024024044e-06,
      "loss": 0.3853,
      "step": 20280
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.55718195438385,
      "learning_rate": 6.1299269514324345e-06,
      "loss": 0.4637,
      "step": 20281
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.575212001800537,
      "learning_rate": 6.128987918979672e-06,
      "loss": 0.4094,
      "step": 20282
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5615533590316772,
      "learning_rate": 6.128048926675494e-06,
      "loss": 0.4333,
      "step": 20283
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.852573275566101,
      "learning_rate": 6.127109974529639e-06,
      "loss": 0.4317,
      "step": 20284
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5040584802627563,
      "learning_rate": 6.126171062551846e-06,
      "loss": 0.4131,
      "step": 20285
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5787086486816406,
      "learning_rate": 6.125232190751859e-06,
      "loss": 0.42,
      "step": 20286
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5428987741470337,
      "learning_rate": 6.124293359139404e-06,
      "loss": 0.4501,
      "step": 20287
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6404063701629639,
      "learning_rate": 6.1233545677242265e-06,
      "loss": 0.4474,
      "step": 20288
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6288143396377563,
      "learning_rate": 6.1224158165160606e-06,
      "loss": 0.4771,
      "step": 20289
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6321383714675903,
      "learning_rate": 6.121477105524647e-06,
      "loss": 0.4151,
      "step": 20290
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5026503801345825,
      "learning_rate": 6.120538434759715e-06,
      "loss": 0.4457,
      "step": 20291
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5406205654144287,
      "learning_rate": 6.1195998042310015e-06,
      "loss": 0.3978,
      "step": 20292
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7226589918136597,
      "learning_rate": 6.118661213948245e-06,
      "loss": 0.4326,
      "step": 20293
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5134031772613525,
      "learning_rate": 6.117722663921179e-06,
      "loss": 0.3918,
      "step": 20294
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.711044430732727,
      "learning_rate": 6.116784154159535e-06,
      "loss": 0.4315,
      "step": 20295
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5173250436782837,
      "learning_rate": 6.115845684673051e-06,
      "loss": 0.3629,
      "step": 20296
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.660573124885559,
      "learning_rate": 6.114907255471455e-06,
      "loss": 0.4889,
      "step": 20297
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5752030611038208,
      "learning_rate": 6.113968866564488e-06,
      "loss": 0.3936,
      "step": 20298
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6485743522644043,
      "learning_rate": 6.113030517961876e-06,
      "loss": 0.4628,
      "step": 20299
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5871272087097168,
      "learning_rate": 6.11209220967335e-06,
      "loss": 0.4301,
      "step": 20300
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.488173246383667,
      "learning_rate": 6.111153941708649e-06,
      "loss": 0.4537,
      "step": 20301
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4754663705825806,
      "learning_rate": 6.1102157140775006e-06,
      "loss": 0.3891,
      "step": 20302
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5705679655075073,
      "learning_rate": 6.109277526789632e-06,
      "loss": 0.4918,
      "step": 20303
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5332032442092896,
      "learning_rate": 6.1083393798547785e-06,
      "loss": 0.3854,
      "step": 20304
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.836512804031372,
      "learning_rate": 6.107401273282667e-06,
      "loss": 0.5232,
      "step": 20305
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5420305728912354,
      "learning_rate": 6.106463207083033e-06,
      "loss": 0.3982,
      "step": 20306
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7026846408843994,
      "learning_rate": 6.105525181265599e-06,
      "loss": 0.986,
      "step": 20307
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5177514553070068,
      "learning_rate": 6.104587195840095e-06,
      "loss": 1.0321,
      "step": 20308
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6555514335632324,
      "learning_rate": 6.103649250816253e-06,
      "loss": 0.4487,
      "step": 20309
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.613494634628296,
      "learning_rate": 6.1027113462038e-06,
      "loss": 0.3785,
      "step": 20310
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.600325107574463,
      "learning_rate": 6.101773482012459e-06,
      "loss": 0.4345,
      "step": 20311
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7459908723831177,
      "learning_rate": 6.100835658251962e-06,
      "loss": 0.3905,
      "step": 20312
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5859748125076294,
      "learning_rate": 6.099897874932034e-06,
      "loss": 0.4789,
      "step": 20313
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.576809048652649,
      "learning_rate": 6.098960132062406e-06,
      "loss": 0.4399,
      "step": 20314
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.769624948501587,
      "learning_rate": 6.0980224296527946e-06,
      "loss": 1.004,
      "step": 20315
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4458469152450562,
      "learning_rate": 6.097084767712931e-06,
      "loss": 1.0349,
      "step": 20316
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5190178155899048,
      "learning_rate": 6.096147146252538e-06,
      "loss": 0.4071,
      "step": 20317
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5164811611175537,
      "learning_rate": 6.095209565281346e-06,
      "loss": 0.3717,
      "step": 20318
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6987693309783936,
      "learning_rate": 6.094272024809073e-06,
      "loss": 0.4655,
      "step": 20319
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4913091659545898,
      "learning_rate": 6.093334524845442e-06,
      "loss": 0.3743,
      "step": 20320
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.621786117553711,
      "learning_rate": 6.092397065400182e-06,
      "loss": 0.429,
      "step": 20321
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5554388761520386,
      "learning_rate": 6.091459646483013e-06,
      "loss": 0.3888,
      "step": 20322
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4908171892166138,
      "learning_rate": 6.090522268103656e-06,
      "loss": 0.3988,
      "step": 20323
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5761064291000366,
      "learning_rate": 6.089584930271834e-06,
      "loss": 0.3968,
      "step": 20324
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.511757493019104,
      "learning_rate": 6.08864763299727e-06,
      "loss": 0.436,
      "step": 20325
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5652430057525635,
      "learning_rate": 6.087710376289688e-06,
      "loss": 0.3775,
      "step": 20326
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5657868385314941,
      "learning_rate": 6.086773160158803e-06,
      "loss": 0.4559,
      "step": 20327
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.541344165802002,
      "learning_rate": 6.085835984614335e-06,
      "loss": 0.4156,
      "step": 20328
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6257669925689697,
      "learning_rate": 6.08489884966601e-06,
      "loss": 0.4695,
      "step": 20329
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4645122289657593,
      "learning_rate": 6.083961755323546e-06,
      "loss": 0.3803,
      "step": 20330
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5243905782699585,
      "learning_rate": 6.083024701596657e-06,
      "loss": 0.4254,
      "step": 20331
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5412625074386597,
      "learning_rate": 6.082087688495068e-06,
      "loss": 0.3996,
      "step": 20332
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5802077054977417,
      "learning_rate": 6.081150716028492e-06,
      "loss": 0.472,
      "step": 20333
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6734046936035156,
      "learning_rate": 6.080213784206653e-06,
      "loss": 0.4132,
      "step": 20334
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5107125043869019,
      "learning_rate": 6.079276893039263e-06,
      "loss": 0.428,
      "step": 20335
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5846740007400513,
      "learning_rate": 6.0783400425360415e-06,
      "loss": 0.4258,
      "step": 20336
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4970424175262451,
      "learning_rate": 6.0774032327067045e-06,
      "loss": 0.4084,
      "step": 20337
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6922500133514404,
      "learning_rate": 6.0764664635609715e-06,
      "loss": 0.4209,
      "step": 20338
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.630077838897705,
      "learning_rate": 6.075529735108552e-06,
      "loss": 0.4458,
      "step": 20339
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4923393726348877,
      "learning_rate": 6.074593047359165e-06,
      "loss": 0.3862,
      "step": 20340
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5610787868499756,
      "learning_rate": 6.073656400322524e-06,
      "loss": 0.469,
      "step": 20341
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5447701215744019,
      "learning_rate": 6.072719794008348e-06,
      "loss": 0.3855,
      "step": 20342
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4799299240112305,
      "learning_rate": 6.071783228426347e-06,
      "loss": 0.454,
      "step": 20343
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.55275297164917,
      "learning_rate": 6.0708467035862324e-06,
      "loss": 0.3834,
      "step": 20344
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6527713537216187,
      "learning_rate": 6.069910219497723e-06,
      "loss": 0.4553,
      "step": 20345
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4366087913513184,
      "learning_rate": 6.068973776170531e-06,
      "loss": 0.3856,
      "step": 20346
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5932313203811646,
      "learning_rate": 6.068037373614364e-06,
      "loss": 0.4153,
      "step": 20347
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6977227926254272,
      "learning_rate": 6.067101011838938e-06,
      "loss": 0.4075,
      "step": 20348
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5999029874801636,
      "learning_rate": 6.066164690853962e-06,
      "loss": 0.4735,
      "step": 20349
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5438810586929321,
      "learning_rate": 6.065228410669153e-06,
      "loss": 0.3963,
      "step": 20350
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5520617961883545,
      "learning_rate": 6.064292171294213e-06,
      "loss": 0.4651,
      "step": 20351
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.505936861038208,
      "learning_rate": 6.063355972738858e-06,
      "loss": 0.398,
      "step": 20352
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6105087995529175,
      "learning_rate": 6.0624198150127954e-06,
      "loss": 0.4518,
      "step": 20353
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8253517150878906,
      "learning_rate": 6.06148369812574e-06,
      "loss": 0.3996,
      "step": 20354
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3711248636245728,
      "learning_rate": 6.0605476220873935e-06,
      "loss": 0.3605,
      "step": 20355
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4514448642730713,
      "learning_rate": 6.059611586907466e-06,
      "loss": 0.3893,
      "step": 20356
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.551579475402832,
      "learning_rate": 6.05867559259567e-06,
      "loss": 0.4774,
      "step": 20357
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5320045948028564,
      "learning_rate": 6.057739639161713e-06,
      "loss": 0.4269,
      "step": 20358
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6642343997955322,
      "learning_rate": 6.056803726615294e-06,
      "loss": 0.4853,
      "step": 20359
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5391730070114136,
      "learning_rate": 6.055867854966129e-06,
      "loss": 0.3902,
      "step": 20360
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.539319396018982,
      "learning_rate": 6.054932024223919e-06,
      "loss": 0.4268,
      "step": 20361
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5916721820831299,
      "learning_rate": 6.053996234398376e-06,
      "loss": 0.3909,
      "step": 20362
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.559935450553894,
      "learning_rate": 6.053060485499202e-06,
      "loss": 0.4377,
      "step": 20363
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5349208116531372,
      "learning_rate": 6.0521247775360995e-06,
      "loss": 0.38,
      "step": 20364
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6174463033676147,
      "learning_rate": 6.051189110518776e-06,
      "loss": 0.4878,
      "step": 20365
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.503187656402588,
      "learning_rate": 6.05025348445694e-06,
      "loss": 0.384,
      "step": 20366
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.512189507484436,
      "learning_rate": 6.049317899360286e-06,
      "loss": 0.4345,
      "step": 20367
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5130455493927002,
      "learning_rate": 6.048382355238525e-06,
      "loss": 0.3643,
      "step": 20368
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.582017421722412,
      "learning_rate": 6.047446852101358e-06,
      "loss": 0.4636,
      "step": 20369
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4745687246322632,
      "learning_rate": 6.046511389958489e-06,
      "loss": 0.3622,
      "step": 20370
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6413832902908325,
      "learning_rate": 6.045575968819618e-06,
      "loss": 0.4657,
      "step": 20371
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5449730157852173,
      "learning_rate": 6.044640588694446e-06,
      "loss": 0.4347,
      "step": 20372
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5634684562683105,
      "learning_rate": 6.0437052495926776e-06,
      "loss": 0.4214,
      "step": 20373
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5785722732543945,
      "learning_rate": 6.042769951524014e-06,
      "loss": 0.4033,
      "step": 20374
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.534016728401184,
      "learning_rate": 6.04183469449815e-06,
      "loss": 0.4011,
      "step": 20375
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4931819438934326,
      "learning_rate": 6.040899478524791e-06,
      "loss": 0.4034,
      "step": 20376
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6167632341384888,
      "learning_rate": 6.039964303613634e-06,
      "loss": 0.4145,
      "step": 20377
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.017423152923584,
      "learning_rate": 6.039029169774384e-06,
      "loss": 0.4072,
      "step": 20378
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5797361135482788,
      "learning_rate": 6.038094077016734e-06,
      "loss": 0.4324,
      "step": 20379
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5037024021148682,
      "learning_rate": 6.037159025350381e-06,
      "loss": 0.4009,
      "step": 20380
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5777173042297363,
      "learning_rate": 6.036224014785027e-06,
      "loss": 0.4607,
      "step": 20381
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6966978311538696,
      "learning_rate": 6.035289045330371e-06,
      "loss": 0.3897,
      "step": 20382
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5501444339752197,
      "learning_rate": 6.034354116996104e-06,
      "loss": 0.4207,
      "step": 20383
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5060995817184448,
      "learning_rate": 6.033419229791928e-06,
      "loss": 0.3782,
      "step": 20384
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5722613334655762,
      "learning_rate": 6.032484383727535e-06,
      "loss": 0.4707,
      "step": 20385
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5447027683258057,
      "learning_rate": 6.031549578812628e-06,
      "loss": 0.4033,
      "step": 20386
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5698763132095337,
      "learning_rate": 6.030614815056895e-06,
      "loss": 0.4362,
      "step": 20387
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5547000169754028,
      "learning_rate": 6.0296800924700315e-06,
      "loss": 0.425,
      "step": 20388
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5715866088867188,
      "learning_rate": 6.028745411061737e-06,
      "loss": 0.4993,
      "step": 20389
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5284725427627563,
      "learning_rate": 6.027810770841704e-06,
      "loss": 0.4101,
      "step": 20390
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5939288139343262,
      "learning_rate": 6.0268761718196246e-06,
      "loss": 0.4238,
      "step": 20391
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.593927025794983,
      "learning_rate": 6.02594161400519e-06,
      "loss": 0.4449,
      "step": 20392
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.508257269859314,
      "learning_rate": 6.025007097408098e-06,
      "loss": 0.4175,
      "step": 20393
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6039187908172607,
      "learning_rate": 6.02407262203804e-06,
      "loss": 0.4192,
      "step": 20394
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.622768521308899,
      "learning_rate": 6.023138187904704e-06,
      "loss": 0.4552,
      "step": 20395
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.45675528049469,
      "learning_rate": 6.022203795017784e-06,
      "loss": 0.4076,
      "step": 20396
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6062613725662231,
      "learning_rate": 6.0212694433869726e-06,
      "loss": 0.462,
      "step": 20397
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5952025651931763,
      "learning_rate": 6.020335133021962e-06,
      "loss": 0.3912,
      "step": 20398
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5430766344070435,
      "learning_rate": 6.0194008639324365e-06,
      "loss": 0.4649,
      "step": 20399
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5575799942016602,
      "learning_rate": 6.018466636128089e-06,
      "loss": 0.4047,
      "step": 20400
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5009410381317139,
      "learning_rate": 6.017532449618611e-06,
      "loss": 0.4043,
      "step": 20401
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4900294542312622,
      "learning_rate": 6.016598304413691e-06,
      "loss": 0.403,
      "step": 20402
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5705324411392212,
      "learning_rate": 6.015664200523013e-06,
      "loss": 0.4335,
      "step": 20403
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5551215410232544,
      "learning_rate": 6.01473013795627e-06,
      "loss": 0.4124,
      "step": 20404
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6729556322097778,
      "learning_rate": 6.013796116723147e-06,
      "loss": 0.5101,
      "step": 20405
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5400974750518799,
      "learning_rate": 6.012862136833336e-06,
      "loss": 0.3955,
      "step": 20406
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7061786651611328,
      "learning_rate": 6.011928198296519e-06,
      "loss": 0.4362,
      "step": 20407
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7043975591659546,
      "learning_rate": 6.010994301122381e-06,
      "loss": 0.3749,
      "step": 20408
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.535313367843628,
      "learning_rate": 6.010060445320614e-06,
      "loss": 0.4608,
      "step": 20409
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5350614786148071,
      "learning_rate": 6.009126630900901e-06,
      "loss": 0.387,
      "step": 20410
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7189003229141235,
      "learning_rate": 6.0081928578729235e-06,
      "loss": 0.5125,
      "step": 20411
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5387139320373535,
      "learning_rate": 6.007259126246371e-06,
      "loss": 0.3872,
      "step": 20412
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5635311603546143,
      "learning_rate": 6.006325436030924e-06,
      "loss": 0.4314,
      "step": 20413
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.535468339920044,
      "learning_rate": 6.005391787236272e-06,
      "loss": 0.409,
      "step": 20414
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5196735858917236,
      "learning_rate": 6.004458179872095e-06,
      "loss": 0.4515,
      "step": 20415
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5217504501342773,
      "learning_rate": 6.003524613948071e-06,
      "loss": 0.3743,
      "step": 20416
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6503620147705078,
      "learning_rate": 6.0025910894738905e-06,
      "loss": 0.489,
      "step": 20417
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5151387453079224,
      "learning_rate": 6.001657606459234e-06,
      "loss": 0.4104,
      "step": 20418
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0790107250213623,
      "learning_rate": 6.000724164913781e-06,
      "loss": 0.401,
      "step": 20419
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.498199701309204,
      "learning_rate": 5.999790764847212e-06,
      "loss": 0.3862,
      "step": 20420
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5429866313934326,
      "learning_rate": 5.998857406269208e-06,
      "loss": 0.4488,
      "step": 20421
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5586204528808594,
      "learning_rate": 5.997924089189455e-06,
      "loss": 0.4237,
      "step": 20422
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5128241777420044,
      "learning_rate": 5.9969908136176284e-06,
      "loss": 0.4373,
      "step": 20423
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.605186104774475,
      "learning_rate": 5.996057579563406e-06,
      "loss": 0.3928,
      "step": 20424
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.561795711517334,
      "learning_rate": 5.995124387036468e-06,
      "loss": 0.4467,
      "step": 20425
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.591870903968811,
      "learning_rate": 5.9941912360465e-06,
      "loss": 0.4077,
      "step": 20426
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.501984715461731,
      "learning_rate": 5.993258126603169e-06,
      "loss": 0.4788,
      "step": 20427
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5069799423217773,
      "learning_rate": 5.99232505871616e-06,
      "loss": 0.3808,
      "step": 20428
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6135196685791016,
      "learning_rate": 5.991392032395147e-06,
      "loss": 0.511,
      "step": 20429
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6826274394989014,
      "learning_rate": 5.990459047649815e-06,
      "loss": 0.3933,
      "step": 20430
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.569283127784729,
      "learning_rate": 5.989526104489827e-06,
      "loss": 0.4618,
      "step": 20431
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5394889116287231,
      "learning_rate": 5.9885932029248705e-06,
      "loss": 0.3964,
      "step": 20432
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7110885381698608,
      "learning_rate": 5.987660342964613e-06,
      "loss": 0.9339,
      "step": 20433
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5575965642929077,
      "learning_rate": 5.986727524618738e-06,
      "loss": 1.0684,
      "step": 20434
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5454939603805542,
      "learning_rate": 5.985794747896916e-06,
      "loss": 0.4269,
      "step": 20435
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6940603256225586,
      "learning_rate": 5.984862012808819e-06,
      "loss": 0.404,
      "step": 20436
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5772385597229004,
      "learning_rate": 5.983929319364123e-06,
      "loss": 0.449,
      "step": 20437
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5992077589035034,
      "learning_rate": 5.982996667572507e-06,
      "loss": 0.4033,
      "step": 20438
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5279560089111328,
      "learning_rate": 5.982064057443633e-06,
      "loss": 0.4491,
      "step": 20439
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4974721670150757,
      "learning_rate": 5.981131488987183e-06,
      "loss": 0.3999,
      "step": 20440
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5283186435699463,
      "learning_rate": 5.980198962212824e-06,
      "loss": 0.4286,
      "step": 20441
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4920543432235718,
      "learning_rate": 5.979266477130232e-06,
      "loss": 0.446,
      "step": 20442
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5403629541397095,
      "learning_rate": 5.978334033749076e-06,
      "loss": 0.4009,
      "step": 20443
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4802486896514893,
      "learning_rate": 5.977401632079025e-06,
      "loss": 0.3754,
      "step": 20444
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5322999954223633,
      "learning_rate": 5.9764692721297525e-06,
      "loss": 0.4394,
      "step": 20445
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5676826238632202,
      "learning_rate": 5.97553695391093e-06,
      "loss": 0.3951,
      "step": 20446
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6056115627288818,
      "learning_rate": 5.974604677432221e-06,
      "loss": 0.4541,
      "step": 20447
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5178110599517822,
      "learning_rate": 5.9736724427033e-06,
      "loss": 0.4022,
      "step": 20448
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5402448177337646,
      "learning_rate": 5.972740249733832e-06,
      "loss": 0.3706,
      "step": 20449
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6446621417999268,
      "learning_rate": 5.971808098533492e-06,
      "loss": 0.4166,
      "step": 20450
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7218899726867676,
      "learning_rate": 5.970875989111941e-06,
      "loss": 0.4937,
      "step": 20451
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5916860103607178,
      "learning_rate": 5.969943921478848e-06,
      "loss": 0.3897,
      "step": 20452
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5763219594955444,
      "learning_rate": 5.969011895643881e-06,
      "loss": 0.4497,
      "step": 20453
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5010390281677246,
      "learning_rate": 5.968079911616711e-06,
      "loss": 0.4051,
      "step": 20454
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4750899076461792,
      "learning_rate": 5.967147969406994e-06,
      "loss": 0.4415,
      "step": 20455
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.539772629737854,
      "learning_rate": 5.966216069024404e-06,
      "loss": 0.395,
      "step": 20456
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.538901686668396,
      "learning_rate": 5.965284210478601e-06,
      "loss": 0.4577,
      "step": 20457
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5964878797531128,
      "learning_rate": 5.964352393779258e-06,
      "loss": 0.3951,
      "step": 20458
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6750614643096924,
      "learning_rate": 5.963420618936031e-06,
      "loss": 0.4372,
      "step": 20459
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.542827844619751,
      "learning_rate": 5.962488885958584e-06,
      "loss": 0.3957,
      "step": 20460
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6160778999328613,
      "learning_rate": 5.961557194856587e-06,
      "loss": 0.4373,
      "step": 20461
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5832548141479492,
      "learning_rate": 5.960625545639702e-06,
      "loss": 0.4264,
      "step": 20462
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7529089450836182,
      "learning_rate": 5.959693938317584e-06,
      "loss": 0.4642,
      "step": 20463
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8681092262268066,
      "learning_rate": 5.958762372899903e-06,
      "loss": 0.4321,
      "step": 20464
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5446373224258423,
      "learning_rate": 5.957830849396318e-06,
      "loss": 0.4762,
      "step": 20465
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5374557971954346,
      "learning_rate": 5.956899367816495e-06,
      "loss": 0.3911,
      "step": 20466
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5113650560379028,
      "learning_rate": 5.955967928170086e-06,
      "loss": 0.4581,
      "step": 20467
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5294889211654663,
      "learning_rate": 5.955036530466758e-06,
      "loss": 0.3833,
      "step": 20468
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8127447366714478,
      "learning_rate": 5.954105174716166e-06,
      "loss": 1.0682,
      "step": 20469
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4767171144485474,
      "learning_rate": 5.9531738609279784e-06,
      "loss": 1.0039,
      "step": 20470
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5578426122665405,
      "learning_rate": 5.952242589111847e-06,
      "loss": 0.41,
      "step": 20471
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7438791990280151,
      "learning_rate": 5.95131135927743e-06,
      "loss": 0.4134,
      "step": 20472
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5211811065673828,
      "learning_rate": 5.9503801714343915e-06,
      "loss": 0.4315,
      "step": 20473
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5048069953918457,
      "learning_rate": 5.949449025592387e-06,
      "loss": 0.3988,
      "step": 20474
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.635770320892334,
      "learning_rate": 5.94851792176107e-06,
      "loss": 0.4929,
      "step": 20475
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.20350980758667,
      "learning_rate": 5.947586859950103e-06,
      "loss": 0.4134,
      "step": 20476
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.55221426486969,
      "learning_rate": 5.946655840169137e-06,
      "loss": 0.4298,
      "step": 20477
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.545335054397583,
      "learning_rate": 5.945724862427837e-06,
      "loss": 0.412,
      "step": 20478
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7057747840881348,
      "learning_rate": 5.944793926735853e-06,
      "loss": 0.4312,
      "step": 20479
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5668116807937622,
      "learning_rate": 5.943863033102836e-06,
      "loss": 0.3978,
      "step": 20480
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5216330289840698,
      "learning_rate": 5.942932181538448e-06,
      "loss": 0.4486,
      "step": 20481
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.593698263168335,
      "learning_rate": 5.9420013720523385e-06,
      "loss": 0.422,
      "step": 20482
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5785986185073853,
      "learning_rate": 5.941070604654169e-06,
      "loss": 0.4552,
      "step": 20483
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5098026990890503,
      "learning_rate": 5.940139879353585e-06,
      "loss": 0.3841,
      "step": 20484
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5475455522537231,
      "learning_rate": 5.9392091961602406e-06,
      "loss": 0.4292,
      "step": 20485
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6297622919082642,
      "learning_rate": 5.938278555083792e-06,
      "loss": 0.4103,
      "step": 20486
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5775055885314941,
      "learning_rate": 5.937347956133892e-06,
      "loss": 0.4876,
      "step": 20487
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6053745746612549,
      "learning_rate": 5.936417399320187e-06,
      "loss": 0.3884,
      "step": 20488
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5610910654067993,
      "learning_rate": 5.935486884652333e-06,
      "loss": 0.4326,
      "step": 20489
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.667823076248169,
      "learning_rate": 5.934556412139977e-06,
      "loss": 0.373,
      "step": 20490
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5656843185424805,
      "learning_rate": 5.933625981792777e-06,
      "loss": 0.4784,
      "step": 20491
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5601506233215332,
      "learning_rate": 5.932695593620374e-06,
      "loss": 0.4011,
      "step": 20492
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5639454126358032,
      "learning_rate": 5.931765247632421e-06,
      "loss": 0.4604,
      "step": 20493
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.49712336063385,
      "learning_rate": 5.9308349438385705e-06,
      "loss": 0.3784,
      "step": 20494
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5179063081741333,
      "learning_rate": 5.929904682248471e-06,
      "loss": 0.4215,
      "step": 20495
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.523277759552002,
      "learning_rate": 5.928974462871762e-06,
      "loss": 0.3763,
      "step": 20496
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.605885624885559,
      "learning_rate": 5.928044285718102e-06,
      "loss": 0.4254,
      "step": 20497
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5866116285324097,
      "learning_rate": 5.927114150797131e-06,
      "loss": 0.4045,
      "step": 20498
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.513566017150879,
      "learning_rate": 5.926184058118503e-06,
      "loss": 0.4056,
      "step": 20499
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.571904182434082,
      "learning_rate": 5.925254007691859e-06,
      "loss": 0.413,
      "step": 20500
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5172289609909058,
      "learning_rate": 5.924323999526844e-06,
      "loss": 0.451,
      "step": 20501
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5582383871078491,
      "learning_rate": 5.9233940336331094e-06,
      "loss": 0.432,
      "step": 20502
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.772573471069336,
      "learning_rate": 5.922464110020301e-06,
      "loss": 0.9721,
      "step": 20503
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6697791814804077,
      "learning_rate": 5.9215342286980534e-06,
      "loss": 1.0465,
      "step": 20504
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8171252012252808,
      "learning_rate": 5.9206043896760215e-06,
      "loss": 0.4679,
      "step": 20505
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5326505899429321,
      "learning_rate": 5.919674592963842e-06,
      "loss": 0.3856,
      "step": 20506
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.557957410812378,
      "learning_rate": 5.918744838571169e-06,
      "loss": 0.4391,
      "step": 20507
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.558471918106079,
      "learning_rate": 5.9178151265076314e-06,
      "loss": 0.3722,
      "step": 20508
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6817035675048828,
      "learning_rate": 5.91688545678288e-06,
      "loss": 0.4744,
      "step": 20509
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5649452209472656,
      "learning_rate": 5.915955829406554e-06,
      "loss": 0.3966,
      "step": 20510
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5565571784973145,
      "learning_rate": 5.915026244388302e-06,
      "loss": 0.4547,
      "step": 20511
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5437034368515015,
      "learning_rate": 5.914096701737757e-06,
      "loss": 0.3846,
      "step": 20512
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5152767896652222,
      "learning_rate": 5.913167201464562e-06,
      "loss": 0.417,
      "step": 20513
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4949721097946167,
      "learning_rate": 5.912237743578359e-06,
      "loss": 0.3963,
      "step": 20514
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6548867225646973,
      "learning_rate": 5.9113083280887896e-06,
      "loss": 0.4491,
      "step": 20515
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4993058443069458,
      "learning_rate": 5.910378955005487e-06,
      "loss": 0.3773,
      "step": 20516
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.598894715309143,
      "learning_rate": 5.9094496243380976e-06,
      "loss": 0.4665,
      "step": 20517
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4908723831176758,
      "learning_rate": 5.908520336096253e-06,
      "loss": 0.3814,
      "step": 20518
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.532096028327942,
      "learning_rate": 5.907591090289599e-06,
      "loss": 0.406,
      "step": 20519
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5244755744934082,
      "learning_rate": 5.906661886927769e-06,
      "loss": 0.3735,
      "step": 20520
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6592153310775757,
      "learning_rate": 5.905732726020398e-06,
      "loss": 0.4606,
      "step": 20521
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.515555739402771,
      "learning_rate": 5.9048036075771274e-06,
      "loss": 0.399,
      "step": 20522
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6156657934188843,
      "learning_rate": 5.903874531607594e-06,
      "loss": 0.4207,
      "step": 20523
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5708223581314087,
      "learning_rate": 5.9029454981214285e-06,
      "loss": 0.4118,
      "step": 20524
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5743550062179565,
      "learning_rate": 5.902016507128271e-06,
      "loss": 0.4823,
      "step": 20525
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5559015274047852,
      "learning_rate": 5.901087558637753e-06,
      "loss": 0.3576,
      "step": 20526
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.560653567314148,
      "learning_rate": 5.900158652659515e-06,
      "loss": 0.4201,
      "step": 20527
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5445802211761475,
      "learning_rate": 5.899229789203188e-06,
      "loss": 0.4231,
      "step": 20528
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7192779779434204,
      "learning_rate": 5.898300968278403e-06,
      "loss": 0.4569,
      "step": 20529
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.474199891090393,
      "learning_rate": 5.897372189894797e-06,
      "loss": 0.3879,
      "step": 20530
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6135891675949097,
      "learning_rate": 5.896443454062004e-06,
      "loss": 0.4454,
      "step": 20531
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.527987003326416,
      "learning_rate": 5.89551476078965e-06,
      "loss": 0.4059,
      "step": 20532
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6240589618682861,
      "learning_rate": 5.894586110087373e-06,
      "loss": 0.4281,
      "step": 20533
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5045539140701294,
      "learning_rate": 5.893657501964803e-06,
      "loss": 0.3847,
      "step": 20534
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.598130226135254,
      "learning_rate": 5.892728936431573e-06,
      "loss": 0.4201,
      "step": 20535
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.522632122039795,
      "learning_rate": 5.891800413497311e-06,
      "loss": 0.3909,
      "step": 20536
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6286842823028564,
      "learning_rate": 5.890871933171645e-06,
      "loss": 0.443,
      "step": 20537
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5660312175750732,
      "learning_rate": 5.88994349546421e-06,
      "loss": 0.4189,
      "step": 20538
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6125316619873047,
      "learning_rate": 5.889015100384636e-06,
      "loss": 0.4588,
      "step": 20539
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.692960500717163,
      "learning_rate": 5.888086747942545e-06,
      "loss": 0.3881,
      "step": 20540
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5693573951721191,
      "learning_rate": 5.887158438147571e-06,
      "loss": 0.4353,
      "step": 20541
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6324490308761597,
      "learning_rate": 5.886230171009339e-06,
      "loss": 0.4363,
      "step": 20542
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4759546518325806,
      "learning_rate": 5.885301946537485e-06,
      "loss": 0.4735,
      "step": 20543
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5614408254623413,
      "learning_rate": 5.884373764741624e-06,
      "loss": 0.376,
      "step": 20544
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.557773232460022,
      "learning_rate": 5.8834456256313895e-06,
      "loss": 0.509,
      "step": 20545
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5377731323242188,
      "learning_rate": 5.882517529216405e-06,
      "loss": 0.3853,
      "step": 20546
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6287983655929565,
      "learning_rate": 5.881589475506301e-06,
      "loss": 0.4698,
      "step": 20547
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.572299838066101,
      "learning_rate": 5.8806614645106985e-06,
      "loss": 0.4145,
      "step": 20548
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6031687259674072,
      "learning_rate": 5.879733496239222e-06,
      "loss": 0.4236,
      "step": 20549
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5794810056686401,
      "learning_rate": 5.878805570701499e-06,
      "loss": 0.3788,
      "step": 20550
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4989813566207886,
      "learning_rate": 5.877877687907154e-06,
      "loss": 0.4223,
      "step": 20551
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.632477045059204,
      "learning_rate": 5.876949847865807e-06,
      "loss": 0.4257,
      "step": 20552
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5296474695205688,
      "learning_rate": 5.876022050587082e-06,
      "loss": 0.4174,
      "step": 20553
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8366727828979492,
      "learning_rate": 5.875094296080603e-06,
      "loss": 0.384,
      "step": 20554
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.631612777709961,
      "learning_rate": 5.874166584355997e-06,
      "loss": 0.467,
      "step": 20555
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.530091643333435,
      "learning_rate": 5.873238915422878e-06,
      "loss": 0.3797,
      "step": 20556
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5257418155670166,
      "learning_rate": 5.872311289290869e-06,
      "loss": 0.4207,
      "step": 20557
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.497031569480896,
      "learning_rate": 5.871383705969593e-06,
      "loss": 0.3955,
      "step": 20558
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.549663782119751,
      "learning_rate": 5.870456165468674e-06,
      "loss": 0.4524,
      "step": 20559
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.516981840133667,
      "learning_rate": 5.869528667797722e-06,
      "loss": 0.3748,
      "step": 20560
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.634524941444397,
      "learning_rate": 5.868601212966366e-06,
      "loss": 0.4528,
      "step": 20561
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.542893648147583,
      "learning_rate": 5.86767380098422e-06,
      "loss": 0.3974,
      "step": 20562
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4810099601745605,
      "learning_rate": 5.866746431860908e-06,
      "loss": 0.4492,
      "step": 20563
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.607479453086853,
      "learning_rate": 5.865819105606044e-06,
      "loss": 0.3837,
      "step": 20564
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4578996896743774,
      "learning_rate": 5.864891822229242e-06,
      "loss": 0.3902,
      "step": 20565
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5109574794769287,
      "learning_rate": 5.863964581740129e-06,
      "loss": 0.4068,
      "step": 20566
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5724570751190186,
      "learning_rate": 5.863037384148318e-06,
      "loss": 0.4065,
      "step": 20567
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4869022369384766,
      "learning_rate": 5.86211022946342e-06,
      "loss": 0.3888,
      "step": 20568
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5243297815322876,
      "learning_rate": 5.8611831176950595e-06,
      "loss": 0.4448,
      "step": 20569
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5349591970443726,
      "learning_rate": 5.8602560488528456e-06,
      "loss": 0.4008,
      "step": 20570
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7027562856674194,
      "learning_rate": 5.8593290229464e-06,
      "loss": 0.4185,
      "step": 20571
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4851409196853638,
      "learning_rate": 5.858402039985331e-06,
      "loss": 0.3822,
      "step": 20572
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.546788215637207,
      "learning_rate": 5.8574750999792554e-06,
      "loss": 0.4439,
      "step": 20573
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.595746397972107,
      "learning_rate": 5.856548202937787e-06,
      "loss": 0.424,
      "step": 20574
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6086896657943726,
      "learning_rate": 5.8556213488705434e-06,
      "loss": 0.4899,
      "step": 20575
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8246724605560303,
      "learning_rate": 5.854694537787129e-06,
      "loss": 0.4441,
      "step": 20576
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.569167137145996,
      "learning_rate": 5.853767769697163e-06,
      "loss": 0.4481,
      "step": 20577
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5336395502090454,
      "learning_rate": 5.852841044610254e-06,
      "loss": 0.3897,
      "step": 20578
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.581508994102478,
      "learning_rate": 5.851914362536018e-06,
      "loss": 0.4499,
      "step": 20579
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4863934516906738,
      "learning_rate": 5.850987723484061e-06,
      "loss": 0.406,
      "step": 20580
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7166239023208618,
      "learning_rate": 5.850061127463993e-06,
      "loss": 0.9976,
      "step": 20581
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4257794618606567,
      "learning_rate": 5.849134574485431e-06,
      "loss": 1.0866,
      "step": 20582
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6455754041671753,
      "learning_rate": 5.848208064557983e-06,
      "loss": 0.4271,
      "step": 20583
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5386486053466797,
      "learning_rate": 5.847281597691253e-06,
      "loss": 0.3769,
      "step": 20584
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5744349956512451,
      "learning_rate": 5.8463551738948534e-06,
      "loss": 0.4456,
      "step": 20585
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5089792013168335,
      "learning_rate": 5.845428793178393e-06,
      "loss": 0.3605,
      "step": 20586
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5674593448638916,
      "learning_rate": 5.844502455551482e-06,
      "loss": 0.4983,
      "step": 20587
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4727293252944946,
      "learning_rate": 5.843576161023722e-06,
      "loss": 0.3833,
      "step": 20588
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4429916143417358,
      "learning_rate": 5.842649909604725e-06,
      "loss": 0.3999,
      "step": 20589
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.58795964717865,
      "learning_rate": 5.841723701304094e-06,
      "loss": 0.4207,
      "step": 20590
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5124396085739136,
      "learning_rate": 5.840797536131443e-06,
      "loss": 0.4379,
      "step": 20591
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4889286756515503,
      "learning_rate": 5.83987141409637e-06,
      "loss": 0.3887,
      "step": 20592
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5924193859100342,
      "learning_rate": 5.83894533520848e-06,
      "loss": 0.4895,
      "step": 20593
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.600858211517334,
      "learning_rate": 5.838019299477384e-06,
      "loss": 0.4254,
      "step": 20594
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5795321464538574,
      "learning_rate": 5.837093306912684e-06,
      "loss": 0.4463,
      "step": 20595
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5332502126693726,
      "learning_rate": 5.83616735752398e-06,
      "loss": 0.3989,
      "step": 20596
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6026241779327393,
      "learning_rate": 5.835241451320879e-06,
      "loss": 0.454,
      "step": 20597
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5345956087112427,
      "learning_rate": 5.834315588312987e-06,
      "loss": 0.4007,
      "step": 20598
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5580519437789917,
      "learning_rate": 5.833389768509904e-06,
      "loss": 0.4551,
      "step": 20599
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.483330488204956,
      "learning_rate": 5.832463991921228e-06,
      "loss": 0.3771,
      "step": 20600
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.522274136543274,
      "learning_rate": 5.831538258556564e-06,
      "loss": 0.4069,
      "step": 20601
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5635219812393188,
      "learning_rate": 5.830612568425516e-06,
      "loss": 0.4007,
      "step": 20602
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4873838424682617,
      "learning_rate": 5.829686921537685e-06,
      "loss": 0.4138,
      "step": 20603
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5254569053649902,
      "learning_rate": 5.828761317902669e-06,
      "loss": 0.3834,
      "step": 20604
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.683140516281128,
      "learning_rate": 5.827835757530065e-06,
      "loss": 0.452,
      "step": 20605
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5227231979370117,
      "learning_rate": 5.826910240429475e-06,
      "loss": 0.3906,
      "step": 20606
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6787691116333008,
      "learning_rate": 5.825984766610504e-06,
      "loss": 0.4507,
      "step": 20607
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5254558324813843,
      "learning_rate": 5.825059336082741e-06,
      "loss": 0.3777,
      "step": 20608
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8750998973846436,
      "learning_rate": 5.8241339488557936e-06,
      "loss": 0.4458,
      "step": 20609
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4957492351531982,
      "learning_rate": 5.82320860493925e-06,
      "loss": 0.3862,
      "step": 20610
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5064220428466797,
      "learning_rate": 5.822283304342718e-06,
      "loss": 0.4338,
      "step": 20611
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6019116640090942,
      "learning_rate": 5.821358047075781e-06,
      "loss": 0.4315,
      "step": 20612
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5448589324951172,
      "learning_rate": 5.8204328331480455e-06,
      "loss": 0.4231,
      "step": 20613
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4948840141296387,
      "learning_rate": 5.819507662569108e-06,
      "loss": 0.3898,
      "step": 20614
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5789766311645508,
      "learning_rate": 5.818582535348563e-06,
      "loss": 0.442,
      "step": 20615
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5375244617462158,
      "learning_rate": 5.817657451495997e-06,
      "loss": 0.3807,
      "step": 20616
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6641660928726196,
      "learning_rate": 5.816732411021012e-06,
      "loss": 0.4447,
      "step": 20617
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5291725397109985,
      "learning_rate": 5.815807413933201e-06,
      "loss": 0.4131,
      "step": 20618
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4852792024612427,
      "learning_rate": 5.814882460242167e-06,
      "loss": 0.4518,
      "step": 20619
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5188987255096436,
      "learning_rate": 5.813957549957485e-06,
      "loss": 0.3822,
      "step": 20620
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8373950719833374,
      "learning_rate": 5.813032683088758e-06,
      "loss": 0.4508,
      "step": 20621
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6966969966888428,
      "learning_rate": 5.812107859645577e-06,
      "loss": 0.39,
      "step": 20622
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5630608797073364,
      "learning_rate": 5.811183079637539e-06,
      "loss": 0.4231,
      "step": 20623
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5535143613815308,
      "learning_rate": 5.810258343074229e-06,
      "loss": 0.4095,
      "step": 20624
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.540373682975769,
      "learning_rate": 5.809333649965237e-06,
      "loss": 0.4431,
      "step": 20625
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4833564758300781,
      "learning_rate": 5.808409000320156e-06,
      "loss": 0.3847,
      "step": 20626
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.483127236366272,
      "learning_rate": 5.807484394148581e-06,
      "loss": 0.4116,
      "step": 20627
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4925700426101685,
      "learning_rate": 5.806559831460092e-06,
      "loss": 0.3926,
      "step": 20628
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5362335443496704,
      "learning_rate": 5.805635312264288e-06,
      "loss": 0.4297,
      "step": 20629
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4884833097457886,
      "learning_rate": 5.804710836570748e-06,
      "loss": 0.3843,
      "step": 20630
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.590916395187378,
      "learning_rate": 5.803786404389069e-06,
      "loss": 0.4594,
      "step": 20631
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5115011930465698,
      "learning_rate": 5.802862015728833e-06,
      "loss": 0.3795,
      "step": 20632
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5699189901351929,
      "learning_rate": 5.801937670599628e-06,
      "loss": 0.464,
      "step": 20633
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5199155807495117,
      "learning_rate": 5.801013369011046e-06,
      "loss": 0.3659,
      "step": 20634
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5126549005508423,
      "learning_rate": 5.800089110972671e-06,
      "loss": 0.447,
      "step": 20635
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5071637630462646,
      "learning_rate": 5.799164896494084e-06,
      "loss": 0.4126,
      "step": 20636
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6139265298843384,
      "learning_rate": 5.798240725584874e-06,
      "loss": 0.4184,
      "step": 20637
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6016286611557007,
      "learning_rate": 5.797316598254626e-06,
      "loss": 0.3959,
      "step": 20638
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5625804662704468,
      "learning_rate": 5.79639251451293e-06,
      "loss": 0.4225,
      "step": 20639
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5679115056991577,
      "learning_rate": 5.795468474369365e-06,
      "loss": 0.415,
      "step": 20640
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5105457305908203,
      "learning_rate": 5.794544477833512e-06,
      "loss": 0.4136,
      "step": 20641
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4981708526611328,
      "learning_rate": 5.793620524914958e-06,
      "loss": 0.3793,
      "step": 20642
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5024216175079346,
      "learning_rate": 5.792696615623288e-06,
      "loss": 0.4394,
      "step": 20643
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5188111066818237,
      "learning_rate": 5.7917727499680785e-06,
      "loss": 0.3896,
      "step": 20644
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5980610847473145,
      "learning_rate": 5.790848927958921e-06,
      "loss": 0.4297,
      "step": 20645
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4843764305114746,
      "learning_rate": 5.7899251496053845e-06,
      "loss": 0.367,
      "step": 20646
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5678346157073975,
      "learning_rate": 5.789001414917061e-06,
      "loss": 0.457,
      "step": 20647
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5274068117141724,
      "learning_rate": 5.788077723903523e-06,
      "loss": 0.3819,
      "step": 20648
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.704876184463501,
      "learning_rate": 5.787154076574353e-06,
      "loss": 0.4809,
      "step": 20649
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.514265775680542,
      "learning_rate": 5.786230472939136e-06,
      "loss": 0.3669,
      "step": 20650
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5414845943450928,
      "learning_rate": 5.785306913007449e-06,
      "loss": 0.4257,
      "step": 20651
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5671970844268799,
      "learning_rate": 5.784383396788864e-06,
      "loss": 0.3907,
      "step": 20652
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5777195692062378,
      "learning_rate": 5.783459924292964e-06,
      "loss": 0.48,
      "step": 20653
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5704050064086914,
      "learning_rate": 5.782536495529327e-06,
      "loss": 0.3811,
      "step": 20654
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9842888116836548,
      "learning_rate": 5.781613110507535e-06,
      "loss": 0.4237,
      "step": 20655
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6112807989120483,
      "learning_rate": 5.78068976923716e-06,
      "loss": 0.4422,
      "step": 20656
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5761044025421143,
      "learning_rate": 5.779766471727774e-06,
      "loss": 0.4842,
      "step": 20657
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.495947003364563,
      "learning_rate": 5.778843217988961e-06,
      "loss": 0.3895,
      "step": 20658
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.686856985092163,
      "learning_rate": 5.777920008030297e-06,
      "loss": 0.4148,
      "step": 20659
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5129262208938599,
      "learning_rate": 5.7769968418613535e-06,
      "loss": 0.3894,
      "step": 20660
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.572992205619812,
      "learning_rate": 5.7760737194917015e-06,
      "loss": 0.5282,
      "step": 20661
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.523677110671997,
      "learning_rate": 5.77515064093092e-06,
      "loss": 0.4088,
      "step": 20662
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5284496545791626,
      "learning_rate": 5.774227606188585e-06,
      "loss": 0.4751,
      "step": 20663
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5541871786117554,
      "learning_rate": 5.773304615274265e-06,
      "loss": 0.4406,
      "step": 20664
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5024852752685547,
      "learning_rate": 5.772381668197537e-06,
      "loss": 0.4195,
      "step": 20665
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5898195505142212,
      "learning_rate": 5.771458764967968e-06,
      "loss": 0.4001,
      "step": 20666
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5436317920684814,
      "learning_rate": 5.770535905595138e-06,
      "loss": 0.4973,
      "step": 20667
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5504926443099976,
      "learning_rate": 5.76961309008861e-06,
      "loss": 0.3767,
      "step": 20668
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5105023384094238,
      "learning_rate": 5.768690318457959e-06,
      "loss": 0.483,
      "step": 20669
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6000068187713623,
      "learning_rate": 5.767767590712758e-06,
      "loss": 0.4039,
      "step": 20670
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5159047842025757,
      "learning_rate": 5.7668449068625765e-06,
      "loss": 0.4369,
      "step": 20671
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.512128472328186,
      "learning_rate": 5.765922266916978e-06,
      "loss": 0.3939,
      "step": 20672
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5972983837127686,
      "learning_rate": 5.764999670885536e-06,
      "loss": 0.4904,
      "step": 20673
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6390278339385986,
      "learning_rate": 5.76407711877782e-06,
      "loss": 0.425,
      "step": 20674
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5936999320983887,
      "learning_rate": 5.7631546106034005e-06,
      "loss": 0.4278,
      "step": 20675
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5469311475753784,
      "learning_rate": 5.762232146371844e-06,
      "loss": 0.4317,
      "step": 20676
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4696390628814697,
      "learning_rate": 5.761309726092713e-06,
      "loss": 0.4553,
      "step": 20677
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.624210238456726,
      "learning_rate": 5.760387349775578e-06,
      "loss": 0.4556,
      "step": 20678
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6327283382415771,
      "learning_rate": 5.759465017430008e-06,
      "loss": 0.4811,
      "step": 20679
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5967743396759033,
      "learning_rate": 5.758542729065565e-06,
      "loss": 0.3985,
      "step": 20680
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6689752340316772,
      "learning_rate": 5.757620484691819e-06,
      "loss": 0.4305,
      "step": 20681
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5122449398040771,
      "learning_rate": 5.75669828431833e-06,
      "loss": 0.3868,
      "step": 20682
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4877985715866089,
      "learning_rate": 5.755776127954667e-06,
      "loss": 0.4329,
      "step": 20683
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4986401796340942,
      "learning_rate": 5.754854015610389e-06,
      "loss": 0.4065,
      "step": 20684
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.620194435119629,
      "learning_rate": 5.753931947295064e-06,
      "loss": 0.4411,
      "step": 20685
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4874268770217896,
      "learning_rate": 5.753009923018258e-06,
      "loss": 0.3783,
      "step": 20686
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5566480159759521,
      "learning_rate": 5.7520879427895305e-06,
      "loss": 0.4512,
      "step": 20687
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4845261573791504,
      "learning_rate": 5.751166006618441e-06,
      "loss": 0.3935,
      "step": 20688
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5192393064498901,
      "learning_rate": 5.7502441145145515e-06,
      "loss": 0.4445,
      "step": 20689
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4972140789031982,
      "learning_rate": 5.749322266487428e-06,
      "loss": 0.4152,
      "step": 20690
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5544360876083374,
      "learning_rate": 5.748400462546633e-06,
      "loss": 0.4132,
      "step": 20691
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4726436138153076,
      "learning_rate": 5.747478702701726e-06,
      "loss": 0.3867,
      "step": 20692
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.548790693283081,
      "learning_rate": 5.746556986962258e-06,
      "loss": 0.4298,
      "step": 20693
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5783330202102661,
      "learning_rate": 5.7456353153377976e-06,
      "loss": 0.4297,
      "step": 20694
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.3474278450012207,
      "learning_rate": 5.744713687837905e-06,
      "loss": 0.4068,
      "step": 20695
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.663428544998169,
      "learning_rate": 5.743792104472132e-06,
      "loss": 0.3858,
      "step": 20696
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.526615023612976,
      "learning_rate": 5.742870565250046e-06,
      "loss": 0.4458,
      "step": 20697
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.563517689704895,
      "learning_rate": 5.741949070181195e-06,
      "loss": 0.4249,
      "step": 20698
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5756062269210815,
      "learning_rate": 5.741027619275146e-06,
      "loss": 0.4454,
      "step": 20699
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5456290245056152,
      "learning_rate": 5.740106212541445e-06,
      "loss": 0.4067,
      "step": 20700
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5780701637268066,
      "learning_rate": 5.739184849989661e-06,
      "loss": 0.4536,
      "step": 20701
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5464223623275757,
      "learning_rate": 5.7382635316293376e-06,
      "loss": 0.4014,
      "step": 20702
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.552441954612732,
      "learning_rate": 5.73734225747004e-06,
      "loss": 0.4631,
      "step": 20703
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5361636877059937,
      "learning_rate": 5.736421027521317e-06,
      "loss": 0.3524,
      "step": 20704
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6403615474700928,
      "learning_rate": 5.7354998417927256e-06,
      "loss": 0.4745,
      "step": 20705
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4911236763000488,
      "learning_rate": 5.7345787002938225e-06,
      "loss": 0.3799,
      "step": 20706
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5902811288833618,
      "learning_rate": 5.733657603034162e-06,
      "loss": 0.4731,
      "step": 20707
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6420098543167114,
      "learning_rate": 5.7327365500232875e-06,
      "loss": 0.4097,
      "step": 20708
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.546221375465393,
      "learning_rate": 5.73181554127076e-06,
      "loss": 0.4602,
      "step": 20709
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5945194959640503,
      "learning_rate": 5.73089457678613e-06,
      "loss": 0.4034,
      "step": 20710
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.208289861679077,
      "learning_rate": 5.729973656578954e-06,
      "loss": 0.4561,
      "step": 20711
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5419321060180664,
      "learning_rate": 5.729052780658779e-06,
      "loss": 0.3993,
      "step": 20712
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5362448692321777,
      "learning_rate": 5.7281319490351536e-06,
      "loss": 0.4268,
      "step": 20713
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5582836866378784,
      "learning_rate": 5.727211161717629e-06,
      "loss": 0.4099,
      "step": 20714
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5209516286849976,
      "learning_rate": 5.7262904187157635e-06,
      "loss": 0.4526,
      "step": 20715
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4787017107009888,
      "learning_rate": 5.7253697200390945e-06,
      "loss": 0.3967,
      "step": 20716
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5845686197280884,
      "learning_rate": 5.724449065697182e-06,
      "loss": 0.4136,
      "step": 20717
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5172005891799927,
      "learning_rate": 5.723528455699565e-06,
      "loss": 0.3885,
      "step": 20718
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6831470727920532,
      "learning_rate": 5.7226078900557954e-06,
      "loss": 1.0435,
      "step": 20719
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4418226480484009,
      "learning_rate": 5.721687368775428e-06,
      "loss": 0.9805,
      "step": 20720
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5803406238555908,
      "learning_rate": 5.720766891867999e-06,
      "loss": 0.4272,
      "step": 20721
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6081669330596924,
      "learning_rate": 5.7198464593430635e-06,
      "loss": 0.4192,
      "step": 20722
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5923036336898804,
      "learning_rate": 5.718926071210162e-06,
      "loss": 0.4578,
      "step": 20723
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4962526559829712,
      "learning_rate": 5.718005727478846e-06,
      "loss": 0.3842,
      "step": 20724
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.524107813835144,
      "learning_rate": 5.717085428158654e-06,
      "loss": 0.438,
      "step": 20725
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5297776460647583,
      "learning_rate": 5.716165173259135e-06,
      "loss": 0.4186,
      "step": 20726
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5980764627456665,
      "learning_rate": 5.715244962789836e-06,
      "loss": 0.4472,
      "step": 20727
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6231887340545654,
      "learning_rate": 5.714324796760301e-06,
      "loss": 0.4127,
      "step": 20728
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5163204669952393,
      "learning_rate": 5.713404675180065e-06,
      "loss": 0.4106,
      "step": 20729
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5962828397750854,
      "learning_rate": 5.712484598058678e-06,
      "loss": 0.3946,
      "step": 20730
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5426088571548462,
      "learning_rate": 5.711564565405682e-06,
      "loss": 0.4561,
      "step": 20731
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5857117176055908,
      "learning_rate": 5.710644577230622e-06,
      "loss": 0.3989,
      "step": 20732
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5740963220596313,
      "learning_rate": 5.709724633543037e-06,
      "loss": 0.4925,
      "step": 20733
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6168164014816284,
      "learning_rate": 5.708804734352465e-06,
      "loss": 0.3959,
      "step": 20734
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.536567211151123,
      "learning_rate": 5.707884879668449e-06,
      "loss": 0.4419,
      "step": 20735
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5745975971221924,
      "learning_rate": 5.706965069500534e-06,
      "loss": 0.3785,
      "step": 20736
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6132274866104126,
      "learning_rate": 5.706045303858257e-06,
      "loss": 0.4257,
      "step": 20737
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6102447509765625,
      "learning_rate": 5.705125582751152e-06,
      "loss": 0.3966,
      "step": 20738
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5924283266067505,
      "learning_rate": 5.7042059061887626e-06,
      "loss": 0.4839,
      "step": 20739
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5149407386779785,
      "learning_rate": 5.7032862741806315e-06,
      "loss": 0.4086,
      "step": 20740
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5567591190338135,
      "learning_rate": 5.70236668673629e-06,
      "loss": 0.3887,
      "step": 20741
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5101977586746216,
      "learning_rate": 5.7014471438652795e-06,
      "loss": 0.3888,
      "step": 20742
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5357420444488525,
      "learning_rate": 5.700527645577133e-06,
      "loss": 0.4523,
      "step": 20743
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.568561315536499,
      "learning_rate": 5.699608191881394e-06,
      "loss": 0.3828,
      "step": 20744
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8871040344238281,
      "learning_rate": 5.698688782787591e-06,
      "loss": 0.9297,
      "step": 20745
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5118051767349243,
      "learning_rate": 5.697769418305264e-06,
      "loss": 0.9996,
      "step": 20746
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.63999605178833,
      "learning_rate": 5.69685009844395e-06,
      "loss": 0.4889,
      "step": 20747
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5606791973114014,
      "learning_rate": 5.695930823213183e-06,
      "loss": 0.3939,
      "step": 20748
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6521995067596436,
      "learning_rate": 5.695011592622491e-06,
      "loss": 0.4499,
      "step": 20749
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5259031057357788,
      "learning_rate": 5.694092406681413e-06,
      "loss": 0.4004,
      "step": 20750
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5596297979354858,
      "learning_rate": 5.6931732653994835e-06,
      "loss": 0.4637,
      "step": 20751
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5678496360778809,
      "learning_rate": 5.692254168786237e-06,
      "loss": 0.403,
      "step": 20752
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5112483501434326,
      "learning_rate": 5.691335116851202e-06,
      "loss": 0.4135,
      "step": 20753
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5967127084732056,
      "learning_rate": 5.69041610960391e-06,
      "loss": 0.3682,
      "step": 20754
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.56781804561615,
      "learning_rate": 5.689497147053893e-06,
      "loss": 0.4348,
      "step": 20755
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4822876453399658,
      "learning_rate": 5.688578229210688e-06,
      "loss": 0.3736,
      "step": 20756
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5934211015701294,
      "learning_rate": 5.687659356083817e-06,
      "loss": 0.4479,
      "step": 20757
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5282139778137207,
      "learning_rate": 5.686740527682819e-06,
      "loss": 0.4028,
      "step": 20758
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9366061687469482,
      "learning_rate": 5.685821744017213e-06,
      "loss": 0.4371,
      "step": 20759
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.506184458732605,
      "learning_rate": 5.684903005096541e-06,
      "loss": 0.414,
      "step": 20760
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5082676410675049,
      "learning_rate": 5.683984310930319e-06,
      "loss": 0.3974,
      "step": 20761
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4963669776916504,
      "learning_rate": 5.683065661528082e-06,
      "loss": 0.4015,
      "step": 20762
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5428348779678345,
      "learning_rate": 5.682147056899361e-06,
      "loss": 0.3974,
      "step": 20763
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5903313159942627,
      "learning_rate": 5.681228497053679e-06,
      "loss": 0.395,
      "step": 20764
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.579408049583435,
      "learning_rate": 5.68030998200056e-06,
      "loss": 0.4734,
      "step": 20765
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5438076257705688,
      "learning_rate": 5.679391511749535e-06,
      "loss": 0.3748,
      "step": 20766
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4822025299072266,
      "learning_rate": 5.6784730863101275e-06,
      "loss": 0.4388,
      "step": 20767
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5301519632339478,
      "learning_rate": 5.67755470569187e-06,
      "loss": 0.4117,
      "step": 20768
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.630184292793274,
      "learning_rate": 5.6766363699042815e-06,
      "loss": 0.4826,
      "step": 20769
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4596518278121948,
      "learning_rate": 5.675718078956882e-06,
      "loss": 0.3781,
      "step": 20770
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5317091941833496,
      "learning_rate": 5.674799832859204e-06,
      "loss": 0.4086,
      "step": 20771
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.541034460067749,
      "learning_rate": 5.673881631620769e-06,
      "loss": 0.3956,
      "step": 20772
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5561447143554688,
      "learning_rate": 5.6729634752510965e-06,
      "loss": 0.4867,
      "step": 20773
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7015782594680786,
      "learning_rate": 5.672045363759716e-06,
      "loss": 0.4133,
      "step": 20774
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5629791021347046,
      "learning_rate": 5.671127297156143e-06,
      "loss": 0.4538,
      "step": 20775
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5100970268249512,
      "learning_rate": 5.670209275449905e-06,
      "loss": 0.4119,
      "step": 20776
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6263047456741333,
      "learning_rate": 5.669291298650516e-06,
      "loss": 0.4817,
      "step": 20777
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.494350790977478,
      "learning_rate": 5.668373366767506e-06,
      "loss": 0.3939,
      "step": 20778
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5403156280517578,
      "learning_rate": 5.667455479810386e-06,
      "loss": 0.4147,
      "step": 20779
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6226465702056885,
      "learning_rate": 5.666537637788686e-06,
      "loss": 0.39,
      "step": 20780
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5278759002685547,
      "learning_rate": 5.6656198407119155e-06,
      "loss": 0.4299,
      "step": 20781
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5247459411621094,
      "learning_rate": 5.6647020885895986e-06,
      "loss": 0.3761,
      "step": 20782
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5978482961654663,
      "learning_rate": 5.663784381431257e-06,
      "loss": 0.4532,
      "step": 20783
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5835355520248413,
      "learning_rate": 5.662866719246406e-06,
      "loss": 0.4043,
      "step": 20784
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5169026851654053,
      "learning_rate": 5.6619491020445586e-06,
      "loss": 0.4413,
      "step": 20785
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.46071195602417,
      "learning_rate": 5.661031529835235e-06,
      "loss": 0.3814,
      "step": 20786
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5701611042022705,
      "learning_rate": 5.660114002627955e-06,
      "loss": 0.4936,
      "step": 20787
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5117535591125488,
      "learning_rate": 5.659196520432234e-06,
      "loss": 0.3738,
      "step": 20788
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4882837533950806,
      "learning_rate": 5.658279083257587e-06,
      "loss": 0.4156,
      "step": 20789
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7029083967208862,
      "learning_rate": 5.657361691113526e-06,
      "loss": 0.4397,
      "step": 20790
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4991647005081177,
      "learning_rate": 5.656444344009566e-06,
      "loss": 0.4095,
      "step": 20791
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5271048545837402,
      "learning_rate": 5.655527041955231e-06,
      "loss": 0.4068,
      "step": 20792
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6086289882659912,
      "learning_rate": 5.654609784960021e-06,
      "loss": 0.493,
      "step": 20793
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4981722831726074,
      "learning_rate": 5.653692573033463e-06,
      "loss": 0.3757,
      "step": 20794
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5170133113861084,
      "learning_rate": 5.652775406185056e-06,
      "loss": 0.4313,
      "step": 20795
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5697989463806152,
      "learning_rate": 5.651858284424325e-06,
      "loss": 0.3893,
      "step": 20796
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.737030029296875,
      "learning_rate": 5.650941207760774e-06,
      "loss": 0.4829,
      "step": 20797
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.519689679145813,
      "learning_rate": 5.6500241762039144e-06,
      "loss": 0.3959,
      "step": 20798
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6058894395828247,
      "learning_rate": 5.649107189763265e-06,
      "loss": 0.4351,
      "step": 20799
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5224450826644897,
      "learning_rate": 5.6481902484483334e-06,
      "loss": 0.3798,
      "step": 20800
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7674261331558228,
      "learning_rate": 5.647273352268622e-06,
      "loss": 0.9486,
      "step": 20801
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4063342809677124,
      "learning_rate": 5.646356501233648e-06,
      "loss": 0.9393,
      "step": 20802
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.628050684928894,
      "learning_rate": 5.645439695352917e-06,
      "loss": 0.4182,
      "step": 20803
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5376465320587158,
      "learning_rate": 5.644522934635943e-06,
      "loss": 0.3843,
      "step": 20804
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.562659502029419,
      "learning_rate": 5.643606219092233e-06,
      "loss": 0.4551,
      "step": 20805
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5112369060516357,
      "learning_rate": 5.642689548731288e-06,
      "loss": 0.412,
      "step": 20806
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6020212173461914,
      "learning_rate": 5.64177292356262e-06,
      "loss": 0.4097,
      "step": 20807
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.515394687652588,
      "learning_rate": 5.64085634359574e-06,
      "loss": 0.3981,
      "step": 20808
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5673283338546753,
      "learning_rate": 5.639939808840146e-06,
      "loss": 0.4735,
      "step": 20809
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4954434633255005,
      "learning_rate": 5.639023319305352e-06,
      "loss": 0.3827,
      "step": 20810
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6012461185455322,
      "learning_rate": 5.638106875000857e-06,
      "loss": 0.4833,
      "step": 20811
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4628716707229614,
      "learning_rate": 5.637190475936172e-06,
      "loss": 0.3822,
      "step": 20812
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4853681325912476,
      "learning_rate": 5.636274122120794e-06,
      "loss": 0.409,
      "step": 20813
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5174208879470825,
      "learning_rate": 5.635357813564232e-06,
      "loss": 0.3868,
      "step": 20814
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5706955194473267,
      "learning_rate": 5.634441550275992e-06,
      "loss": 0.459,
      "step": 20815
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.530177116394043,
      "learning_rate": 5.633525332265575e-06,
      "loss": 0.3821,
      "step": 20816
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.595631718635559,
      "learning_rate": 5.632609159542477e-06,
      "loss": 0.4505,
      "step": 20817
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5385531187057495,
      "learning_rate": 5.631693032116208e-06,
      "loss": 0.4003,
      "step": 20818
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4840420484542847,
      "learning_rate": 5.6307769499962705e-06,
      "loss": 0.3902,
      "step": 20819
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4610230922698975,
      "learning_rate": 5.629860913192162e-06,
      "loss": 0.3919,
      "step": 20820
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.563448429107666,
      "learning_rate": 5.6289449217133815e-06,
      "loss": 0.4366,
      "step": 20821
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5638052225112915,
      "learning_rate": 5.6280289755694305e-06,
      "loss": 0.3858,
      "step": 20822
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.595198631286621,
      "learning_rate": 5.627113074769811e-06,
      "loss": 0.5108,
      "step": 20823
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.560164451599121,
      "learning_rate": 5.6261972193240255e-06,
      "loss": 0.3855,
      "step": 20824
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6159483194351196,
      "learning_rate": 5.625281409241569e-06,
      "loss": 0.4136,
      "step": 20825
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.477874517440796,
      "learning_rate": 5.624365644531935e-06,
      "loss": 0.3744,
      "step": 20826
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5990163087844849,
      "learning_rate": 5.623449925204627e-06,
      "loss": 0.4344,
      "step": 20827
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.494215726852417,
      "learning_rate": 5.622534251269145e-06,
      "loss": 0.3961,
      "step": 20828
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5560519695281982,
      "learning_rate": 5.621618622734979e-06,
      "loss": 0.4213,
      "step": 20829
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5976938009262085,
      "learning_rate": 5.620703039611635e-06,
      "loss": 0.4268,
      "step": 20830
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7767659425735474,
      "learning_rate": 5.619787501908596e-06,
      "loss": 0.4527,
      "step": 20831
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5473430156707764,
      "learning_rate": 5.618872009635372e-06,
      "loss": 0.3965,
      "step": 20832
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5280452966690063,
      "learning_rate": 5.617956562801446e-06,
      "loss": 0.4266,
      "step": 20833
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5554888248443604,
      "learning_rate": 5.617041161416317e-06,
      "loss": 0.4104,
      "step": 20834
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6452635526657104,
      "learning_rate": 5.616125805489484e-06,
      "loss": 0.4517,
      "step": 20835
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4797208309173584,
      "learning_rate": 5.615210495030436e-06,
      "loss": 0.3768,
      "step": 20836
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7409026622772217,
      "learning_rate": 5.614295230048665e-06,
      "loss": 0.5168,
      "step": 20837
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5748291015625,
      "learning_rate": 5.613380010553664e-06,
      "loss": 0.395,
      "step": 20838
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5862282514572144,
      "learning_rate": 5.612464836554927e-06,
      "loss": 0.4366,
      "step": 20839
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.53199303150177,
      "learning_rate": 5.611549708061949e-06,
      "loss": 0.3704,
      "step": 20840
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5807623863220215,
      "learning_rate": 5.6106346250842195e-06,
      "loss": 0.4613,
      "step": 20841
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5101635456085205,
      "learning_rate": 5.609719587631224e-06,
      "loss": 0.3885,
      "step": 20842
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.601688265800476,
      "learning_rate": 5.608804595712456e-06,
      "loss": 0.4376,
      "step": 20843
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.540025234222412,
      "learning_rate": 5.60788964933741e-06,
      "loss": 0.3725,
      "step": 20844
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.54579496383667,
      "learning_rate": 5.606974748515568e-06,
      "loss": 0.4267,
      "step": 20845
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4447911977767944,
      "learning_rate": 5.606059893256426e-06,
      "loss": 0.3992,
      "step": 20846
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6477229595184326,
      "learning_rate": 5.605145083569465e-06,
      "loss": 0.4797,
      "step": 20847
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5625298023223877,
      "learning_rate": 5.604230319464183e-06,
      "loss": 0.3765,
      "step": 20848
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5606069564819336,
      "learning_rate": 5.603315600950057e-06,
      "loss": 0.4383,
      "step": 20849
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0568342208862305,
      "learning_rate": 5.602400928036577e-06,
      "loss": 0.3909,
      "step": 20850
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6529065370559692,
      "learning_rate": 5.601486300733239e-06,
      "loss": 0.4631,
      "step": 20851
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.455946445465088,
      "learning_rate": 5.600571719049519e-06,
      "loss": 0.3768,
      "step": 20852
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6190871000289917,
      "learning_rate": 5.599657182994902e-06,
      "loss": 0.4563,
      "step": 20853
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5484604835510254,
      "learning_rate": 5.598742692578876e-06,
      "loss": 0.3878,
      "step": 20854
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5928621292114258,
      "learning_rate": 5.597828247810931e-06,
      "loss": 0.4721,
      "step": 20855
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6622532606124878,
      "learning_rate": 5.596913848700547e-06,
      "loss": 0.3811,
      "step": 20856
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5142791271209717,
      "learning_rate": 5.595999495257203e-06,
      "loss": 0.4339,
      "step": 20857
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.503146767616272,
      "learning_rate": 5.595085187490385e-06,
      "loss": 0.3914,
      "step": 20858
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5124337673187256,
      "learning_rate": 5.594170925409579e-06,
      "loss": 0.4814,
      "step": 20859
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6036219596862793,
      "learning_rate": 5.593256709024268e-06,
      "loss": 0.4056,
      "step": 20860
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5388928651809692,
      "learning_rate": 5.592342538343933e-06,
      "loss": 0.4328,
      "step": 20861
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5833567380905151,
      "learning_rate": 5.59142841337805e-06,
      "loss": 0.3747,
      "step": 20862
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6331157684326172,
      "learning_rate": 5.590514334136105e-06,
      "loss": 0.462,
      "step": 20863
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.613734483718872,
      "learning_rate": 5.589600300627582e-06,
      "loss": 0.4377,
      "step": 20864
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5100773572921753,
      "learning_rate": 5.588686312861951e-06,
      "loss": 0.4291,
      "step": 20865
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5402445793151855,
      "learning_rate": 5.587772370848701e-06,
      "loss": 0.4125,
      "step": 20866
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.568333625793457,
      "learning_rate": 5.586858474597302e-06,
      "loss": 0.4378,
      "step": 20867
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5583661794662476,
      "learning_rate": 5.585944624117243e-06,
      "loss": 0.4185,
      "step": 20868
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5229814052581787,
      "learning_rate": 5.585030819417991e-06,
      "loss": 0.432,
      "step": 20869
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5685982704162598,
      "learning_rate": 5.584117060509031e-06,
      "loss": 0.3853,
      "step": 20870
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5286264419555664,
      "learning_rate": 5.583203347399843e-06,
      "loss": 0.4333,
      "step": 20871
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5335447788238525,
      "learning_rate": 5.582289680099898e-06,
      "loss": 0.4062,
      "step": 20872
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6280404329299927,
      "learning_rate": 5.581376058618669e-06,
      "loss": 0.4229,
      "step": 20873
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5487043857574463,
      "learning_rate": 5.580462482965637e-06,
      "loss": 0.364,
      "step": 20874
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6021978855133057,
      "learning_rate": 5.579548953150275e-06,
      "loss": 0.4813,
      "step": 20875
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4927668571472168,
      "learning_rate": 5.5786354691820656e-06,
      "loss": 0.398,
      "step": 20876
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5332342386245728,
      "learning_rate": 5.577722031070474e-06,
      "loss": 0.4432,
      "step": 20877
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4365041255950928,
      "learning_rate": 5.576808638824973e-06,
      "loss": 0.3747,
      "step": 20878
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6415009498596191,
      "learning_rate": 5.57589529245504e-06,
      "loss": 0.4035,
      "step": 20879
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4908185005187988,
      "learning_rate": 5.57498199197015e-06,
      "loss": 0.3902,
      "step": 20880
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4963222742080688,
      "learning_rate": 5.57406873737977e-06,
      "loss": 0.4325,
      "step": 20881
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4620977640151978,
      "learning_rate": 5.573155528693379e-06,
      "loss": 0.3912,
      "step": 20882
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6607577800750732,
      "learning_rate": 5.572242365920439e-06,
      "loss": 0.4616,
      "step": 20883
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5430335998535156,
      "learning_rate": 5.57132924907043e-06,
      "loss": 0.3944,
      "step": 20884
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6137893199920654,
      "learning_rate": 5.570416178152816e-06,
      "loss": 0.4781,
      "step": 20885
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.448427438735962,
      "learning_rate": 5.569503153177067e-06,
      "loss": 0.3628,
      "step": 20886
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.948047161102295,
      "learning_rate": 5.56859017415266e-06,
      "loss": 0.4531,
      "step": 20887
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.505845546722412,
      "learning_rate": 5.5676772410890604e-06,
      "loss": 0.39,
      "step": 20888
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5490832328796387,
      "learning_rate": 5.566764353995731e-06,
      "loss": 0.4566,
      "step": 20889
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5182517766952515,
      "learning_rate": 5.565851512882143e-06,
      "loss": 0.3969,
      "step": 20890
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.710849404335022,
      "learning_rate": 5.564938717757766e-06,
      "loss": 0.4604,
      "step": 20891
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4758102893829346,
      "learning_rate": 5.564025968632074e-06,
      "loss": 0.3611,
      "step": 20892
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5590693950653076,
      "learning_rate": 5.5631132655145195e-06,
      "loss": 0.4775,
      "step": 20893
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4602774381637573,
      "learning_rate": 5.562200608414576e-06,
      "loss": 0.3989,
      "step": 20894
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5236163139343262,
      "learning_rate": 5.561287997341707e-06,
      "loss": 0.403,
      "step": 20895
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5293282270431519,
      "learning_rate": 5.5603754323053825e-06,
      "loss": 0.4514,
      "step": 20896
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4873707294464111,
      "learning_rate": 5.559462913315066e-06,
      "loss": 0.4025,
      "step": 20897
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5003197193145752,
      "learning_rate": 5.558550440380215e-06,
      "loss": 0.3778,
      "step": 20898
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5622222423553467,
      "learning_rate": 5.557638013510298e-06,
      "loss": 0.4623,
      "step": 20899
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4954257011413574,
      "learning_rate": 5.5567256327147835e-06,
      "loss": 0.3725,
      "step": 20900
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.492925763130188,
      "learning_rate": 5.555813298003125e-06,
      "loss": 0.4772,
      "step": 20901
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5411500930786133,
      "learning_rate": 5.554901009384792e-06,
      "loss": 0.4071,
      "step": 20902
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5707542896270752,
      "learning_rate": 5.55398876686924e-06,
      "loss": 0.4307,
      "step": 20903
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5432312488555908,
      "learning_rate": 5.55307657046594e-06,
      "loss": 0.3939,
      "step": 20904
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4961282014846802,
      "learning_rate": 5.552164420184341e-06,
      "loss": 0.4097,
      "step": 20905
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4981956481933594,
      "learning_rate": 5.55125231603391e-06,
      "loss": 0.3807,
      "step": 20906
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.57656729221344,
      "learning_rate": 5.550340258024109e-06,
      "loss": 0.452,
      "step": 20907
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5109686851501465,
      "learning_rate": 5.549428246164396e-06,
      "loss": 0.3825,
      "step": 20908
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5504826307296753,
      "learning_rate": 5.548516280464227e-06,
      "loss": 0.4437,
      "step": 20909
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4630727767944336,
      "learning_rate": 5.54760436093306e-06,
      "loss": 0.3878,
      "step": 20910
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.522772192955017,
      "learning_rate": 5.546692487580357e-06,
      "loss": 0.4682,
      "step": 20911
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5752472877502441,
      "learning_rate": 5.5457806604155764e-06,
      "loss": 0.3995,
      "step": 20912
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6469365358352661,
      "learning_rate": 5.544868879448173e-06,
      "loss": 0.4467,
      "step": 20913
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5162663459777832,
      "learning_rate": 5.5439571446876e-06,
      "loss": 0.4144,
      "step": 20914
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6322095394134521,
      "learning_rate": 5.543045456143318e-06,
      "loss": 0.4367,
      "step": 20915
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5238091945648193,
      "learning_rate": 5.5421338138247835e-06,
      "loss": 0.403,
      "step": 20916
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.571710228919983,
      "learning_rate": 5.541222217741449e-06,
      "loss": 0.4656,
      "step": 20917
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6215252876281738,
      "learning_rate": 5.540310667902771e-06,
      "loss": 0.4431,
      "step": 20918
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7615348100662231,
      "learning_rate": 5.5393991643181996e-06,
      "loss": 0.9578,
      "step": 20919
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5146234035491943,
      "learning_rate": 5.538487706997198e-06,
      "loss": 1.0352,
      "step": 20920
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4845046997070312,
      "learning_rate": 5.5375762959492065e-06,
      "loss": 0.4246,
      "step": 20921
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6145685911178589,
      "learning_rate": 5.536664931183685e-06,
      "loss": 0.4197,
      "step": 20922
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5872228145599365,
      "learning_rate": 5.535753612710091e-06,
      "loss": 0.4534,
      "step": 20923
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5064536333084106,
      "learning_rate": 5.534842340537869e-06,
      "loss": 0.4005,
      "step": 20924
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5329698324203491,
      "learning_rate": 5.533931114676467e-06,
      "loss": 0.4178,
      "step": 20925
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.579285740852356,
      "learning_rate": 5.533019935135344e-06,
      "loss": 0.3822,
      "step": 20926
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.476210594177246,
      "learning_rate": 5.532108801923945e-06,
      "loss": 0.4489,
      "step": 20927
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5474497079849243,
      "learning_rate": 5.531197715051732e-06,
      "loss": 0.3956,
      "step": 20928
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5287690162658691,
      "learning_rate": 5.530286674528136e-06,
      "loss": 0.4242,
      "step": 20929
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.509921908378601,
      "learning_rate": 5.529375680362614e-06,
      "loss": 0.3851,
      "step": 20930
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5124967098236084,
      "learning_rate": 5.528464732564615e-06,
      "loss": 0.3939,
      "step": 20931
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6564807891845703,
      "learning_rate": 5.5275538311435926e-06,
      "loss": 0.4693,
      "step": 20932
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5841671228408813,
      "learning_rate": 5.526642976108987e-06,
      "loss": 0.4703,
      "step": 20933
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5593317747116089,
      "learning_rate": 5.525732167470244e-06,
      "loss": 0.4016,
      "step": 20934
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5412360429763794,
      "learning_rate": 5.524821405236813e-06,
      "loss": 0.424,
      "step": 20935
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4920899868011475,
      "learning_rate": 5.523910689418144e-06,
      "loss": 0.3608,
      "step": 20936
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6911067962646484,
      "learning_rate": 5.5230000200236745e-06,
      "loss": 1.0025,
      "step": 20937
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.435011625289917,
      "learning_rate": 5.522089397062859e-06,
      "loss": 0.9731,
      "step": 20938
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6241419315338135,
      "learning_rate": 5.521178820545132e-06,
      "loss": 0.4691,
      "step": 20939
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.546504259109497,
      "learning_rate": 5.520268290479946e-06,
      "loss": 0.416,
      "step": 20940
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6489440202713013,
      "learning_rate": 5.51935780687674e-06,
      "loss": 0.4685,
      "step": 20941
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4887397289276123,
      "learning_rate": 5.518447369744956e-06,
      "loss": 0.3766,
      "step": 20942
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6117699146270752,
      "learning_rate": 5.517536979094043e-06,
      "loss": 0.469,
      "step": 20943
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4527393579483032,
      "learning_rate": 5.516626634933441e-06,
      "loss": 0.3781,
      "step": 20944
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6734334230422974,
      "learning_rate": 5.515716337272585e-06,
      "loss": 0.4159,
      "step": 20945
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.581013798713684,
      "learning_rate": 5.514806086120921e-06,
      "loss": 0.382,
      "step": 20946
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5113496780395508,
      "learning_rate": 5.513895881487891e-06,
      "loss": 0.4227,
      "step": 20947
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.594228744506836,
      "learning_rate": 5.512985723382937e-06,
      "loss": 0.4094,
      "step": 20948
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5460478067398071,
      "learning_rate": 5.512075611815497e-06,
      "loss": 0.4544,
      "step": 20949
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5927793979644775,
      "learning_rate": 5.5111655467950065e-06,
      "loss": 0.4052,
      "step": 20950
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6919581890106201,
      "learning_rate": 5.510255528330905e-06,
      "loss": 0.4932,
      "step": 20951
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5377708673477173,
      "learning_rate": 5.5093455564326394e-06,
      "loss": 0.3991,
      "step": 20952
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6536012887954712,
      "learning_rate": 5.508435631109636e-06,
      "loss": 0.4849,
      "step": 20953
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6557103395462036,
      "learning_rate": 5.507525752371343e-06,
      "loss": 0.3962,
      "step": 20954
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4897516965866089,
      "learning_rate": 5.506615920227186e-06,
      "loss": 0.4371,
      "step": 20955
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4980510473251343,
      "learning_rate": 5.505706134686613e-06,
      "loss": 0.3893,
      "step": 20956
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.384136199951172,
      "learning_rate": 5.504796395759048e-06,
      "loss": 0.5017,
      "step": 20957
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.518133282661438,
      "learning_rate": 5.503886703453933e-06,
      "loss": 0.3974,
      "step": 20958
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5093026161193848,
      "learning_rate": 5.5029770577807075e-06,
      "loss": 0.4173,
      "step": 20959
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.542576551437378,
      "learning_rate": 5.502067458748796e-06,
      "loss": 0.3937,
      "step": 20960
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5350602865219116,
      "learning_rate": 5.501157906367642e-06,
      "loss": 0.4188,
      "step": 20961
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4999265670776367,
      "learning_rate": 5.50024840064667e-06,
      "loss": 0.3846,
      "step": 20962
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.556929349899292,
      "learning_rate": 5.499338941595319e-06,
      "loss": 0.39,
      "step": 20963
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6176663637161255,
      "learning_rate": 5.498429529223024e-06,
      "loss": 0.4072,
      "step": 20964
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5560896396636963,
      "learning_rate": 5.497520163539212e-06,
      "loss": 0.4473,
      "step": 20965
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5396349430084229,
      "learning_rate": 5.4966108445533115e-06,
      "loss": 0.3889,
      "step": 20966
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5530956983566284,
      "learning_rate": 5.495701572274759e-06,
      "loss": 0.4577,
      "step": 20967
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.522274136543274,
      "learning_rate": 5.4947923467129835e-06,
      "loss": 0.3871,
      "step": 20968
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5871740579605103,
      "learning_rate": 5.493883167877423e-06,
      "loss": 0.4417,
      "step": 20969
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5013970136642456,
      "learning_rate": 5.492974035777492e-06,
      "loss": 0.3715,
      "step": 20970
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5595464706420898,
      "learning_rate": 5.492064950422628e-06,
      "loss": 0.4259,
      "step": 20971
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5640255212783813,
      "learning_rate": 5.491155911822259e-06,
      "loss": 0.4098,
      "step": 20972
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.573706865310669,
      "learning_rate": 5.490246919985815e-06,
      "loss": 0.4485,
      "step": 20973
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.521859049797058,
      "learning_rate": 5.489337974922725e-06,
      "loss": 0.3774,
      "step": 20974
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6112966537475586,
      "learning_rate": 5.4884290766424074e-06,
      "loss": 0.4408,
      "step": 20975
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5558112859725952,
      "learning_rate": 5.487520225154295e-06,
      "loss": 0.4206,
      "step": 20976
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5820297002792358,
      "learning_rate": 5.486611420467818e-06,
      "loss": 0.4435,
      "step": 20977
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5780781507492065,
      "learning_rate": 5.485702662592394e-06,
      "loss": 0.4134,
      "step": 20978
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5888134241104126,
      "learning_rate": 5.484793951537455e-06,
      "loss": 0.4564,
      "step": 20979
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5631805658340454,
      "learning_rate": 5.4838852873124194e-06,
      "loss": 0.4062,
      "step": 20980
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5463687181472778,
      "learning_rate": 5.48297666992672e-06,
      "loss": 0.4766,
      "step": 20981
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4966745376586914,
      "learning_rate": 5.48206809938977e-06,
      "loss": 0.3962,
      "step": 20982
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.530841588973999,
      "learning_rate": 5.481159575710999e-06,
      "loss": 0.402,
      "step": 20983
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5356563329696655,
      "learning_rate": 5.480251098899834e-06,
      "loss": 0.3976,
      "step": 20984
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5883171558380127,
      "learning_rate": 5.479342668965692e-06,
      "loss": 0.5003,
      "step": 20985
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5822631120681763,
      "learning_rate": 5.47843428591799e-06,
      "loss": 0.3917,
      "step": 20986
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4854906797409058,
      "learning_rate": 5.477525949766156e-06,
      "loss": 0.4226,
      "step": 20987
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7123626470565796,
      "learning_rate": 5.47661766051961e-06,
      "loss": 0.4557,
      "step": 20988
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5759410858154297,
      "learning_rate": 5.475709418187775e-06,
      "loss": 0.4319,
      "step": 20989
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4981616735458374,
      "learning_rate": 5.474801222780071e-06,
      "loss": 0.373,
      "step": 20990
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5645109415054321,
      "learning_rate": 5.473893074305907e-06,
      "loss": 0.4605,
      "step": 20991
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.619794487953186,
      "learning_rate": 5.472984972774711e-06,
      "loss": 0.4057,
      "step": 20992
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.588724970817566,
      "learning_rate": 5.4720769181959035e-06,
      "loss": 0.4301,
      "step": 20993
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4565194845199585,
      "learning_rate": 5.471168910578896e-06,
      "loss": 0.3594,
      "step": 20994
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9036917686462402,
      "learning_rate": 5.4702609499331115e-06,
      "loss": 0.4542,
      "step": 20995
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5669517517089844,
      "learning_rate": 5.469353036267961e-06,
      "loss": 0.3994,
      "step": 20996
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4787184000015259,
      "learning_rate": 5.468445169592869e-06,
      "loss": 0.4355,
      "step": 20997
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6095219850540161,
      "learning_rate": 5.467537349917241e-06,
      "loss": 0.3978,
      "step": 20998
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5586340427398682,
      "learning_rate": 5.466629577250499e-06,
      "loss": 0.4354,
      "step": 20999
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.598150610923767,
      "learning_rate": 5.465721851602064e-06,
      "loss": 0.3632,
      "step": 21000
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5791683197021484,
      "learning_rate": 5.464814172981341e-06,
      "loss": 0.424,
      "step": 21001
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5662614107131958,
      "learning_rate": 5.463906541397744e-06,
      "loss": 0.4072,
      "step": 21002
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.484864354133606,
      "learning_rate": 5.46299895686069e-06,
      "loss": 0.4165,
      "step": 21003
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5399826765060425,
      "learning_rate": 5.462091419379593e-06,
      "loss": 0.4285,
      "step": 21004
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5463565587997437,
      "learning_rate": 5.4611839289638715e-06,
      "loss": 0.4347,
      "step": 21005
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5034639835357666,
      "learning_rate": 5.460276485622922e-06,
      "loss": 0.3904,
      "step": 21006
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.486423373222351,
      "learning_rate": 5.4593690893661645e-06,
      "loss": 0.4303,
      "step": 21007
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5851765871047974,
      "learning_rate": 5.458461740203012e-06,
      "loss": 0.396,
      "step": 21008
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5712294578552246,
      "learning_rate": 5.457554438142876e-06,
      "loss": 0.4601,
      "step": 21009
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5335941314697266,
      "learning_rate": 5.456647183195165e-06,
      "loss": 0.3793,
      "step": 21010
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.67162024974823,
      "learning_rate": 5.455739975369284e-06,
      "loss": 0.4426,
      "step": 21011
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4493191242218018,
      "learning_rate": 5.454832814674646e-06,
      "loss": 0.3672,
      "step": 21012
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7440831661224365,
      "learning_rate": 5.453925701120664e-06,
      "loss": 0.4481,
      "step": 21013
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5758602619171143,
      "learning_rate": 5.453018634716738e-06,
      "loss": 0.407,
      "step": 21014
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5933773517608643,
      "learning_rate": 5.452111615472282e-06,
      "loss": 0.5091,
      "step": 21015
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.578840970993042,
      "learning_rate": 5.451204643396699e-06,
      "loss": 0.3819,
      "step": 21016
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5357964038848877,
      "learning_rate": 5.450297718499401e-06,
      "loss": 0.4218,
      "step": 21017
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5232999324798584,
      "learning_rate": 5.449390840789788e-06,
      "loss": 0.3809,
      "step": 21018
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5759943723678589,
      "learning_rate": 5.448484010277268e-06,
      "loss": 0.4706,
      "step": 21019
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5672149658203125,
      "learning_rate": 5.44757722697125e-06,
      "loss": 0.3988,
      "step": 21020
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.585281491279602,
      "learning_rate": 5.446670490881136e-06,
      "loss": 0.442,
      "step": 21021
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.638607382774353,
      "learning_rate": 5.4457638020163285e-06,
      "loss": 0.4194,
      "step": 21022
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5485408306121826,
      "learning_rate": 5.444857160386231e-06,
      "loss": 0.4402,
      "step": 21023
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.514021396636963,
      "learning_rate": 5.443950566000249e-06,
      "loss": 0.3876,
      "step": 21024
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6243902444839478,
      "learning_rate": 5.443044018867789e-06,
      "loss": 0.4858,
      "step": 21025
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.531530499458313,
      "learning_rate": 5.44213751899825e-06,
      "loss": 0.39,
      "step": 21026
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.503613829612732,
      "learning_rate": 5.441231066401029e-06,
      "loss": 0.4401,
      "step": 21027
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5837312936782837,
      "learning_rate": 5.440324661085532e-06,
      "loss": 0.4135,
      "step": 21028
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4771219491958618,
      "learning_rate": 5.439418303061163e-06,
      "loss": 0.4038,
      "step": 21029
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6680501699447632,
      "learning_rate": 5.4385119923373165e-06,
      "loss": 0.4559,
      "step": 21030
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5037307739257812,
      "learning_rate": 5.437605728923397e-06,
      "loss": 0.4288,
      "step": 21031
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.587398648262024,
      "learning_rate": 5.436699512828798e-06,
      "loss": 0.4159,
      "step": 21032
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6457087993621826,
      "learning_rate": 5.435793344062927e-06,
      "loss": 0.4719,
      "step": 21033
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6236371994018555,
      "learning_rate": 5.434887222635174e-06,
      "loss": 0.3937,
      "step": 21034
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5540770292282104,
      "learning_rate": 5.4339811485549385e-06,
      "loss": 0.4457,
      "step": 21035
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5872644186019897,
      "learning_rate": 5.433075121831626e-06,
      "loss": 0.4024,
      "step": 21036
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7377877235412598,
      "learning_rate": 5.432169142474626e-06,
      "loss": 0.425,
      "step": 21037
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4904818534851074,
      "learning_rate": 5.431263210493332e-06,
      "loss": 0.3959,
      "step": 21038
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.579474925994873,
      "learning_rate": 5.430357325897146e-06,
      "loss": 0.4515,
      "step": 21039
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7032700777053833,
      "learning_rate": 5.429451488695461e-06,
      "loss": 0.3822,
      "step": 21040
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.554399013519287,
      "learning_rate": 5.4285456988976755e-06,
      "loss": 0.414,
      "step": 21041
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.519392490386963,
      "learning_rate": 5.427639956513182e-06,
      "loss": 0.4016,
      "step": 21042
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6627960205078125,
      "learning_rate": 5.42673426155137e-06,
      "loss": 0.4714,
      "step": 21043
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5228086709976196,
      "learning_rate": 5.425828614021638e-06,
      "loss": 0.3858,
      "step": 21044
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5091580152511597,
      "learning_rate": 5.4249230139333785e-06,
      "loss": 0.4011,
      "step": 21045
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5927236080169678,
      "learning_rate": 5.424017461295985e-06,
      "loss": 0.3925,
      "step": 21046
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5333516597747803,
      "learning_rate": 5.423111956118845e-06,
      "loss": 0.4066,
      "step": 21047
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6179553270339966,
      "learning_rate": 5.4222064984113505e-06,
      "loss": 0.3945,
      "step": 21048
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.59085214138031,
      "learning_rate": 5.421301088182902e-06,
      "loss": 0.4645,
      "step": 21049
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.508198857307434,
      "learning_rate": 5.4203957254428765e-06,
      "loss": 0.3854,
      "step": 21050
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5948821306228638,
      "learning_rate": 5.419490410200675e-06,
      "loss": 0.4728,
      "step": 21051
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.503868818283081,
      "learning_rate": 5.418585142465679e-06,
      "loss": 0.3798,
      "step": 21052
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5996521711349487,
      "learning_rate": 5.417679922247284e-06,
      "loss": 0.9095,
      "step": 21053
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4006041288375854,
      "learning_rate": 5.416774749554873e-06,
      "loss": 1.0212,
      "step": 21054
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.626574993133545,
      "learning_rate": 5.415869624397836e-06,
      "loss": 0.4612,
      "step": 21055
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4510473012924194,
      "learning_rate": 5.414964546785566e-06,
      "loss": 0.3732,
      "step": 21056
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5184012651443481,
      "learning_rate": 5.414059516727444e-06,
      "loss": 0.4597,
      "step": 21057
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4962719678878784,
      "learning_rate": 5.413154534232856e-06,
      "loss": 0.384,
      "step": 21058
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5784550905227661,
      "learning_rate": 5.41224959931119e-06,
      "loss": 0.4326,
      "step": 21059
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5915141105651855,
      "learning_rate": 5.411344711971831e-06,
      "loss": 0.4279,
      "step": 21060
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6068743467330933,
      "learning_rate": 5.4104398722241695e-06,
      "loss": 0.4394,
      "step": 21061
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5078908205032349,
      "learning_rate": 5.409535080077585e-06,
      "loss": 0.4009,
      "step": 21062
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8155971765518188,
      "learning_rate": 5.408630335541458e-06,
      "loss": 0.4899,
      "step": 21063
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.512697458267212,
      "learning_rate": 5.407725638625177e-06,
      "loss": 0.3861,
      "step": 21064
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.564002275466919,
      "learning_rate": 5.406820989338128e-06,
      "loss": 0.4615,
      "step": 21065
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5210355520248413,
      "learning_rate": 5.405916387689688e-06,
      "loss": 0.392,
      "step": 21066
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5968916416168213,
      "learning_rate": 5.405011833689244e-06,
      "loss": 0.4387,
      "step": 21067
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.568879246711731,
      "learning_rate": 5.404107327346171e-06,
      "loss": 0.3671,
      "step": 21068
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.613551139831543,
      "learning_rate": 5.403202868669859e-06,
      "loss": 0.4258,
      "step": 21069
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5331175327301025,
      "learning_rate": 5.402298457669679e-06,
      "loss": 0.4129,
      "step": 21070
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.579215407371521,
      "learning_rate": 5.4013940943550165e-06,
      "loss": 0.5098,
      "step": 21071
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5684022903442383,
      "learning_rate": 5.400489778735254e-06,
      "loss": 0.3909,
      "step": 21072
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5410957336425781,
      "learning_rate": 5.39958551081977e-06,
      "loss": 0.4506,
      "step": 21073
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5873030424118042,
      "learning_rate": 5.398681290617933e-06,
      "loss": 0.4,
      "step": 21074
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5893568992614746,
      "learning_rate": 5.3977771181391315e-06,
      "loss": 0.4525,
      "step": 21075
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4698365926742554,
      "learning_rate": 5.396872993392739e-06,
      "loss": 0.4271,
      "step": 21076
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6088947057724,
      "learning_rate": 5.395968916388139e-06,
      "loss": 0.4393,
      "step": 21077
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5530917644500732,
      "learning_rate": 5.395064887134704e-06,
      "loss": 0.3766,
      "step": 21078
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5067400932312012,
      "learning_rate": 5.394160905641805e-06,
      "loss": 0.4194,
      "step": 21079
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5949386358261108,
      "learning_rate": 5.393256971918824e-06,
      "loss": 0.4289,
      "step": 21080
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6226842403411865,
      "learning_rate": 5.3923530859751376e-06,
      "loss": 0.4752,
      "step": 21081
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.571377992630005,
      "learning_rate": 5.3914492478201144e-06,
      "loss": 0.38,
      "step": 21082
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.519060492515564,
      "learning_rate": 5.390545457463134e-06,
      "loss": 0.4202,
      "step": 21083
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6644785404205322,
      "learning_rate": 5.389641714913567e-06,
      "loss": 0.442,
      "step": 21084
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5901654958724976,
      "learning_rate": 5.388738020180791e-06,
      "loss": 0.446,
      "step": 21085
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5340794324874878,
      "learning_rate": 5.38783437327417e-06,
      "loss": 0.3783,
      "step": 21086
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5512295961380005,
      "learning_rate": 5.386930774203086e-06,
      "loss": 0.4994,
      "step": 21087
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4354966878890991,
      "learning_rate": 5.386027222976904e-06,
      "loss": 0.3884,
      "step": 21088
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.528573989868164,
      "learning_rate": 5.385123719605001e-06,
      "loss": 0.4238,
      "step": 21089
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.461978554725647,
      "learning_rate": 5.38422026409674e-06,
      "loss": 0.364,
      "step": 21090
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5917295217514038,
      "learning_rate": 5.383316856461496e-06,
      "loss": 0.4967,
      "step": 21091
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.48965322971344,
      "learning_rate": 5.382413496708643e-06,
      "loss": 0.4012,
      "step": 21092
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5316458940505981,
      "learning_rate": 5.381510184847547e-06,
      "loss": 0.4419,
      "step": 21093
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6255135536193848,
      "learning_rate": 5.38060692088757e-06,
      "loss": 0.4279,
      "step": 21094
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.61769437789917,
      "learning_rate": 5.379703704838086e-06,
      "loss": 0.4462,
      "step": 21095
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4838300943374634,
      "learning_rate": 5.378800536708464e-06,
      "loss": 0.3848,
      "step": 21096
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5303808450698853,
      "learning_rate": 5.377897416508073e-06,
      "loss": 0.4744,
      "step": 21097
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5454297065734863,
      "learning_rate": 5.376994344246278e-06,
      "loss": 0.4046,
      "step": 21098
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5321000814437866,
      "learning_rate": 5.376091319932441e-06,
      "loss": 0.5042,
      "step": 21099
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4788528680801392,
      "learning_rate": 5.375188343575929e-06,
      "loss": 0.3779,
      "step": 21100
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5583220720291138,
      "learning_rate": 5.374285415186113e-06,
      "loss": 0.4436,
      "step": 21101
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.499531626701355,
      "learning_rate": 5.373382534772352e-06,
      "loss": 0.3985,
      "step": 21102
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5076085329055786,
      "learning_rate": 5.372479702344016e-06,
      "loss": 0.4575,
      "step": 21103
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.53536856174469,
      "learning_rate": 5.371576917910462e-06,
      "loss": 0.3792,
      "step": 21104
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6572340726852417,
      "learning_rate": 5.370674181481059e-06,
      "loss": 0.4165,
      "step": 21105
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5838137865066528,
      "learning_rate": 5.369771493065164e-06,
      "loss": 0.3847,
      "step": 21106
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5592315196990967,
      "learning_rate": 5.368868852672142e-06,
      "loss": 0.4079,
      "step": 21107
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5283985137939453,
      "learning_rate": 5.36796626031136e-06,
      "loss": 0.3829,
      "step": 21108
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5119702816009521,
      "learning_rate": 5.367063715992176e-06,
      "loss": 0.4178,
      "step": 21109
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5450260639190674,
      "learning_rate": 5.366161219723944e-06,
      "loss": 0.4151,
      "step": 21110
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7779958248138428,
      "learning_rate": 5.365258771516031e-06,
      "loss": 0.9889,
      "step": 21111
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4946221113204956,
      "learning_rate": 5.364356371377795e-06,
      "loss": 1.0245,
      "step": 21112
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5976712703704834,
      "learning_rate": 5.363454019318601e-06,
      "loss": 0.4566,
      "step": 21113
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6859828233718872,
      "learning_rate": 5.362551715347803e-06,
      "loss": 0.3865,
      "step": 21114
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5473555326461792,
      "learning_rate": 5.361649459474756e-06,
      "loss": 0.4514,
      "step": 21115
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5596081018447876,
      "learning_rate": 5.36074725170882e-06,
      "loss": 0.439,
      "step": 21116
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5431171655654907,
      "learning_rate": 5.359845092059359e-06,
      "loss": 0.4169,
      "step": 21117
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.555132508277893,
      "learning_rate": 5.35894298053572e-06,
      "loss": 0.3898,
      "step": 21118
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7920712232589722,
      "learning_rate": 5.3580409171472654e-06,
      "loss": 0.9358,
      "step": 21119
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4867404699325562,
      "learning_rate": 5.357138901903347e-06,
      "loss": 1.0086,
      "step": 21120
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4993411302566528,
      "learning_rate": 5.356236934813328e-06,
      "loss": 0.4345,
      "step": 21121
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5347974300384521,
      "learning_rate": 5.355335015886551e-06,
      "loss": 0.4119,
      "step": 21122
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.681875228881836,
      "learning_rate": 5.354433145132383e-06,
      "loss": 0.8868,
      "step": 21123
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.350050449371338,
      "learning_rate": 5.353531322560168e-06,
      "loss": 0.976,
      "step": 21124
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6454945802688599,
      "learning_rate": 5.352629548179266e-06,
      "loss": 0.4904,
      "step": 21125
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4460456371307373,
      "learning_rate": 5.3517278219990245e-06,
      "loss": 0.3699,
      "step": 21126
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7011780738830566,
      "learning_rate": 5.350826144028797e-06,
      "loss": 0.4889,
      "step": 21127
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.60326087474823,
      "learning_rate": 5.349924514277942e-06,
      "loss": 0.3741,
      "step": 21128
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5325640439987183,
      "learning_rate": 5.3490229327558055e-06,
      "loss": 0.4704,
      "step": 21129
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4513872861862183,
      "learning_rate": 5.348121399471733e-06,
      "loss": 0.3651,
      "step": 21130
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.588064193725586,
      "learning_rate": 5.347219914435081e-06,
      "loss": 0.4434,
      "step": 21131
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.472721815109253,
      "learning_rate": 5.3463184776551994e-06,
      "loss": 0.3851,
      "step": 21132
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5104867219924927,
      "learning_rate": 5.34541708914144e-06,
      "loss": 0.4101,
      "step": 21133
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5326329469680786,
      "learning_rate": 5.34451574890315e-06,
      "loss": 0.4187,
      "step": 21134
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.546305537223816,
      "learning_rate": 5.34361445694967e-06,
      "loss": 0.4478,
      "step": 21135
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4836342334747314,
      "learning_rate": 5.342713213290356e-06,
      "loss": 0.3996,
      "step": 21136
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6889113187789917,
      "learning_rate": 5.341812017934555e-06,
      "loss": 0.3987,
      "step": 21137
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5126014947891235,
      "learning_rate": 5.340910870891611e-06,
      "loss": 0.4106,
      "step": 21138
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5446841716766357,
      "learning_rate": 5.340009772170874e-06,
      "loss": 0.4487,
      "step": 21139
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5019762516021729,
      "learning_rate": 5.339108721781683e-06,
      "loss": 0.3957,
      "step": 21140
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5270787477493286,
      "learning_rate": 5.338207719733392e-06,
      "loss": 0.4248,
      "step": 21141
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5043315887451172,
      "learning_rate": 5.337306766035338e-06,
      "loss": 0.3634,
      "step": 21142
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5331857204437256,
      "learning_rate": 5.336405860696869e-06,
      "loss": 0.4399,
      "step": 21143
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7901090383529663,
      "learning_rate": 5.335505003727332e-06,
      "loss": 0.3881,
      "step": 21144
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5265378952026367,
      "learning_rate": 5.334604195136068e-06,
      "loss": 0.418,
      "step": 21145
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5954829454421997,
      "learning_rate": 5.333703434932414e-06,
      "loss": 0.4069,
      "step": 21146
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.576479196548462,
      "learning_rate": 5.3328027231257165e-06,
      "loss": 0.4156,
      "step": 21147
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6015326976776123,
      "learning_rate": 5.33190205972532e-06,
      "loss": 0.4065,
      "step": 21148
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.557829737663269,
      "learning_rate": 5.331001444740567e-06,
      "loss": 0.4428,
      "step": 21149
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6241992712020874,
      "learning_rate": 5.330100878180794e-06,
      "loss": 0.4247,
      "step": 21150
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.483743667602539,
      "learning_rate": 5.3292003600553385e-06,
      "loss": 0.4247,
      "step": 21151
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5065466165542603,
      "learning_rate": 5.328299890373545e-06,
      "loss": 0.4043,
      "step": 21152
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6603505611419678,
      "learning_rate": 5.327399469144756e-06,
      "loss": 0.4718,
      "step": 21153
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5235130786895752,
      "learning_rate": 5.326499096378303e-06,
      "loss": 0.4197,
      "step": 21154
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8641729354858398,
      "learning_rate": 5.3255987720835315e-06,
      "loss": 0.4354,
      "step": 21155
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5639623403549194,
      "learning_rate": 5.324698496269771e-06,
      "loss": 0.4433,
      "step": 21156
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.618320107460022,
      "learning_rate": 5.323798268946368e-06,
      "loss": 0.3945,
      "step": 21157
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.51616632938385,
      "learning_rate": 5.322898090122648e-06,
      "loss": 0.3904,
      "step": 21158
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4725911617279053,
      "learning_rate": 5.321997959807957e-06,
      "loss": 0.4358,
      "step": 21159
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.505345106124878,
      "learning_rate": 5.32109787801163e-06,
      "loss": 0.3603,
      "step": 21160
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5623691082000732,
      "learning_rate": 5.320197844743e-06,
      "loss": 0.4416,
      "step": 21161
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6081830263137817,
      "learning_rate": 5.319297860011399e-06,
      "loss": 0.4193,
      "step": 21162
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0991265773773193,
      "learning_rate": 5.318397923826162e-06,
      "loss": 0.4847,
      "step": 21163
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5811775922775269,
      "learning_rate": 5.3174980361966315e-06,
      "loss": 0.4082,
      "step": 21164
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5591695308685303,
      "learning_rate": 5.316598197132131e-06,
      "loss": 0.4107,
      "step": 21165
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5677802562713623,
      "learning_rate": 5.3156984066419935e-06,
      "loss": 0.3997,
      "step": 21166
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4545797109603882,
      "learning_rate": 5.314798664735555e-06,
      "loss": 0.3948,
      "step": 21167
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5493403673171997,
      "learning_rate": 5.313898971422146e-06,
      "loss": 0.3827,
      "step": 21168
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5864386558532715,
      "learning_rate": 5.312999326711101e-06,
      "loss": 0.481,
      "step": 21169
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5540663003921509,
      "learning_rate": 5.312099730611749e-06,
      "loss": 0.3962,
      "step": 21170
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5873956680297852,
      "learning_rate": 5.311200183133416e-06,
      "loss": 0.4359,
      "step": 21171
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.461702585220337,
      "learning_rate": 5.310300684285433e-06,
      "loss": 0.3866,
      "step": 21172
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5239899158477783,
      "learning_rate": 5.309401234077136e-06,
      "loss": 0.479,
      "step": 21173
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4793550968170166,
      "learning_rate": 5.308501832517845e-06,
      "loss": 0.3749,
      "step": 21174
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5422388315200806,
      "learning_rate": 5.307602479616896e-06,
      "loss": 0.4542,
      "step": 21175
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7168806791305542,
      "learning_rate": 5.306703175383608e-06,
      "loss": 0.3791,
      "step": 21176
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5431909561157227,
      "learning_rate": 5.305803919827318e-06,
      "loss": 0.4249,
      "step": 21177
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4926178455352783,
      "learning_rate": 5.304904712957344e-06,
      "loss": 0.3901,
      "step": 21178
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5621471405029297,
      "learning_rate": 5.304005554783015e-06,
      "loss": 0.4553,
      "step": 21179
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5025966167449951,
      "learning_rate": 5.303106445313662e-06,
      "loss": 0.3712,
      "step": 21180
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4837665557861328,
      "learning_rate": 5.302207384558607e-06,
      "loss": 0.4361,
      "step": 21181
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5234519243240356,
      "learning_rate": 5.301308372527168e-06,
      "loss": 0.3777,
      "step": 21182
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5761446952819824,
      "learning_rate": 5.300409409228675e-06,
      "loss": 0.4657,
      "step": 21183
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4975332021713257,
      "learning_rate": 5.299510494672451e-06,
      "loss": 0.3636,
      "step": 21184
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5469974279403687,
      "learning_rate": 5.298611628867825e-06,
      "loss": 0.4259,
      "step": 21185
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5005115270614624,
      "learning_rate": 5.297712811824114e-06,
      "loss": 0.3924,
      "step": 21186
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5329238176345825,
      "learning_rate": 5.296814043550636e-06,
      "loss": 0.4212,
      "step": 21187
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5521482229232788,
      "learning_rate": 5.295915324056717e-06,
      "loss": 0.4384,
      "step": 21188
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.545738935470581,
      "learning_rate": 5.2950166533516824e-06,
      "loss": 0.4517,
      "step": 21189
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.49820876121521,
      "learning_rate": 5.2941180314448445e-06,
      "loss": 0.3811,
      "step": 21190
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.652996301651001,
      "learning_rate": 5.293219458345533e-06,
      "loss": 0.4519,
      "step": 21191
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4676488637924194,
      "learning_rate": 5.2923209340630566e-06,
      "loss": 0.3757,
      "step": 21192
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5711525678634644,
      "learning_rate": 5.291422458606746e-06,
      "loss": 0.5326,
      "step": 21193
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5427018404006958,
      "learning_rate": 5.290524031985908e-06,
      "loss": 0.3858,
      "step": 21194
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5730602741241455,
      "learning_rate": 5.289625654209867e-06,
      "loss": 0.4671,
      "step": 21195
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5500450134277344,
      "learning_rate": 5.288727325287945e-06,
      "loss": 0.4064,
      "step": 21196
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.508172869682312,
      "learning_rate": 5.287829045229453e-06,
      "loss": 0.4041,
      "step": 21197
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4478168487548828,
      "learning_rate": 5.286930814043707e-06,
      "loss": 0.4208,
      "step": 21198
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6643768548965454,
      "learning_rate": 5.286032631740023e-06,
      "loss": 0.4536,
      "step": 21199
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4925541877746582,
      "learning_rate": 5.28513449832772e-06,
      "loss": 0.3728,
      "step": 21200
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5564055442810059,
      "learning_rate": 5.284236413816115e-06,
      "loss": 0.4497,
      "step": 21201
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5919612646102905,
      "learning_rate": 5.2833383782145195e-06,
      "loss": 0.413,
      "step": 21202
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6205230951309204,
      "learning_rate": 5.282440391532244e-06,
      "loss": 0.4857,
      "step": 21203
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.537329912185669,
      "learning_rate": 5.281542453778605e-06,
      "loss": 0.3986,
      "step": 21204
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5795331001281738,
      "learning_rate": 5.28064456496292e-06,
      "loss": 0.4476,
      "step": 21205
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4719053506851196,
      "learning_rate": 5.279746725094496e-06,
      "loss": 0.3735,
      "step": 21206
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4806921482086182,
      "learning_rate": 5.278848934182644e-06,
      "loss": 0.4514,
      "step": 21207
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4349390268325806,
      "learning_rate": 5.277951192236679e-06,
      "loss": 0.3704,
      "step": 21208
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5215505361557007,
      "learning_rate": 5.27705349926591e-06,
      "loss": 0.4498,
      "step": 21209
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.70729660987854,
      "learning_rate": 5.276155855279652e-06,
      "loss": 0.4073,
      "step": 21210
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5221883058547974,
      "learning_rate": 5.275258260287212e-06,
      "loss": 0.4427,
      "step": 21211
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.564995527267456,
      "learning_rate": 5.274360714297895e-06,
      "loss": 0.4171,
      "step": 21212
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.46054208278656,
      "learning_rate": 5.273463217321014e-06,
      "loss": 0.3956,
      "step": 21213
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5120962858200073,
      "learning_rate": 5.272565769365883e-06,
      "loss": 0.3901,
      "step": 21214
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5101995468139648,
      "learning_rate": 5.271668370441798e-06,
      "loss": 0.4897,
      "step": 21215
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.676001787185669,
      "learning_rate": 5.270771020558079e-06,
      "loss": 0.4153,
      "step": 21216
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5910941362380981,
      "learning_rate": 5.269873719724021e-06,
      "loss": 0.4408,
      "step": 21217
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5560965538024902,
      "learning_rate": 5.268976467948943e-06,
      "loss": 0.4143,
      "step": 21218
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6069222688674927,
      "learning_rate": 5.268079265242139e-06,
      "loss": 0.4619,
      "step": 21219
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5013108253479004,
      "learning_rate": 5.267182111612918e-06,
      "loss": 0.3785,
      "step": 21220
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.547454833984375,
      "learning_rate": 5.266285007070593e-06,
      "loss": 0.4408,
      "step": 21221
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5136606693267822,
      "learning_rate": 5.265387951624461e-06,
      "loss": 0.3845,
      "step": 21222
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5682603120803833,
      "learning_rate": 5.2644909452838245e-06,
      "loss": 0.4322,
      "step": 21223
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5161833763122559,
      "learning_rate": 5.2635939880579876e-06,
      "loss": 0.38,
      "step": 21224
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6136643886566162,
      "learning_rate": 5.262697079956256e-06,
      "loss": 0.4549,
      "step": 21225
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4636367559432983,
      "learning_rate": 5.261800220987934e-06,
      "loss": 0.358,
      "step": 21226
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.176011800765991,
      "learning_rate": 5.260903411162322e-06,
      "loss": 0.4618,
      "step": 21227
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5271328687667847,
      "learning_rate": 5.260006650488714e-06,
      "loss": 0.4007,
      "step": 21228
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.478654384613037,
      "learning_rate": 5.259109938976418e-06,
      "loss": 0.4185,
      "step": 21229
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5827134847640991,
      "learning_rate": 5.258213276634736e-06,
      "loss": 0.4328,
      "step": 21230
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6609697341918945,
      "learning_rate": 5.2573166634729625e-06,
      "loss": 0.455,
      "step": 21231
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4955617189407349,
      "learning_rate": 5.256420099500401e-06,
      "loss": 0.3987,
      "step": 21232
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6058835983276367,
      "learning_rate": 5.255523584726345e-06,
      "loss": 0.4342,
      "step": 21233
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.51195228099823,
      "learning_rate": 5.2546271191601005e-06,
      "loss": 0.3826,
      "step": 21234
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5326809883117676,
      "learning_rate": 5.253730702810958e-06,
      "loss": 0.436,
      "step": 21235
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5634381771087646,
      "learning_rate": 5.252834335688217e-06,
      "loss": 0.371,
      "step": 21236
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5529366731643677,
      "learning_rate": 5.2519380178011795e-06,
      "loss": 0.4262,
      "step": 21237
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.623410940170288,
      "learning_rate": 5.251041749159136e-06,
      "loss": 0.3877,
      "step": 21238
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5136107206344604,
      "learning_rate": 5.25014552977138e-06,
      "loss": 0.4399,
      "step": 21239
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.493272304534912,
      "learning_rate": 5.249249359647211e-06,
      "loss": 0.3925,
      "step": 21240
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5950202941894531,
      "learning_rate": 5.248353238795927e-06,
      "loss": 0.437,
      "step": 21241
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4947967529296875,
      "learning_rate": 5.247457167226817e-06,
      "loss": 0.3797,
      "step": 21242
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5144151449203491,
      "learning_rate": 5.246561144949174e-06,
      "loss": 0.4254,
      "step": 21243
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5252457857131958,
      "learning_rate": 5.245665171972291e-06,
      "loss": 0.383,
      "step": 21244
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5132777690887451,
      "learning_rate": 5.244769248305463e-06,
      "loss": 0.4373,
      "step": 21245
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.610939383506775,
      "learning_rate": 5.243873373957986e-06,
      "loss": 0.3891,
      "step": 21246
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5851490497589111,
      "learning_rate": 5.242977548939148e-06,
      "loss": 0.4213,
      "step": 21247
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5478709936141968,
      "learning_rate": 5.242081773258237e-06,
      "loss": 0.396,
      "step": 21248
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6228060722351074,
      "learning_rate": 5.241186046924544e-06,
      "loss": 0.5367,
      "step": 21249
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5001254081726074,
      "learning_rate": 5.240290369947366e-06,
      "loss": 0.3897,
      "step": 21250
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5623763799667358,
      "learning_rate": 5.239394742335984e-06,
      "loss": 0.4263,
      "step": 21251
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7378901243209839,
      "learning_rate": 5.238499164099695e-06,
      "loss": 0.3787,
      "step": 21252
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5871621370315552,
      "learning_rate": 5.237603635247781e-06,
      "loss": 0.5134,
      "step": 21253
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4443942308425903,
      "learning_rate": 5.236708155789536e-06,
      "loss": 0.3971,
      "step": 21254
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4735066890716553,
      "learning_rate": 5.23581272573424e-06,
      "loss": 0.4482,
      "step": 21255
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.860360026359558,
      "learning_rate": 5.234917345091184e-06,
      "loss": 0.3898,
      "step": 21256
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5841963291168213,
      "learning_rate": 5.234022013869659e-06,
      "loss": 0.4502,
      "step": 21257
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4979908466339111,
      "learning_rate": 5.2331267320789485e-06,
      "loss": 0.4019,
      "step": 21258
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4849472045898438,
      "learning_rate": 5.232231499728331e-06,
      "loss": 0.4589,
      "step": 21259
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5051274299621582,
      "learning_rate": 5.2313363168270966e-06,
      "loss": 0.3924,
      "step": 21260
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5318739414215088,
      "learning_rate": 5.230441183384532e-06,
      "loss": 0.4017,
      "step": 21261
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5450016260147095,
      "learning_rate": 5.229546099409921e-06,
      "loss": 0.4064,
      "step": 21262
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5045180320739746,
      "learning_rate": 5.228651064912547e-06,
      "loss": 0.4243,
      "step": 21263
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.564351201057434,
      "learning_rate": 5.227756079901685e-06,
      "loss": 0.4217,
      "step": 21264
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5252991914749146,
      "learning_rate": 5.226861144386626e-06,
      "loss": 0.4785,
      "step": 21265
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5434142351150513,
      "learning_rate": 5.225966258376652e-06,
      "loss": 0.4,
      "step": 21266
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6039210557937622,
      "learning_rate": 5.2250714218810385e-06,
      "loss": 0.4486,
      "step": 21267
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5236425399780273,
      "learning_rate": 5.2241766349090735e-06,
      "loss": 0.3726,
      "step": 21268
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6932862997055054,
      "learning_rate": 5.223281897470031e-06,
      "loss": 0.482,
      "step": 21269
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4983261823654175,
      "learning_rate": 5.222387209573196e-06,
      "loss": 0.3641,
      "step": 21270
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5979516506195068,
      "learning_rate": 5.221492571227842e-06,
      "loss": 0.4343,
      "step": 21271
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4436193704605103,
      "learning_rate": 5.220597982443252e-06,
      "loss": 0.3773,
      "step": 21272
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5384997129440308,
      "learning_rate": 5.219703443228707e-06,
      "loss": 0.455,
      "step": 21273
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5307855606079102,
      "learning_rate": 5.218808953593482e-06,
      "loss": 0.3922,
      "step": 21274
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.603394865989685,
      "learning_rate": 5.217914513546848e-06,
      "loss": 0.4282,
      "step": 21275
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5148394107818604,
      "learning_rate": 5.2170201230980886e-06,
      "loss": 0.3764,
      "step": 21276
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5330952405929565,
      "learning_rate": 5.216125782256478e-06,
      "loss": 0.4425,
      "step": 21277
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.57979416847229,
      "learning_rate": 5.215231491031302e-06,
      "loss": 0.3944,
      "step": 21278
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5951943397521973,
      "learning_rate": 5.2143372494318175e-06,
      "loss": 0.4816,
      "step": 21279
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4813534021377563,
      "learning_rate": 5.21344305746731e-06,
      "loss": 0.3623,
      "step": 21280
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5848102569580078,
      "learning_rate": 5.212548915147051e-06,
      "loss": 0.4317,
      "step": 21281
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6603854894638062,
      "learning_rate": 5.21165482248032e-06,
      "loss": 0.3971,
      "step": 21282
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.708435297012329,
      "learning_rate": 5.210760779476385e-06,
      "loss": 0.4236,
      "step": 21283
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5605131387710571,
      "learning_rate": 5.209866786144515e-06,
      "loss": 0.407,
      "step": 21284
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6398885250091553,
      "learning_rate": 5.208972842493987e-06,
      "loss": 0.4193,
      "step": 21285
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4652462005615234,
      "learning_rate": 5.208078948534074e-06,
      "loss": 0.3699,
      "step": 21286
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.493362545967102,
      "learning_rate": 5.2071851042740415e-06,
      "loss": 0.4187,
      "step": 21287
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5073606967926025,
      "learning_rate": 5.206291309723168e-06,
      "loss": 0.3981,
      "step": 21288
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7694265842437744,
      "learning_rate": 5.205397564890716e-06,
      "loss": 0.4672,
      "step": 21289
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5026445388793945,
      "learning_rate": 5.204503869785962e-06,
      "loss": 0.3777,
      "step": 21290
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5689589977264404,
      "learning_rate": 5.203610224418166e-06,
      "loss": 0.45,
      "step": 21291
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4785246849060059,
      "learning_rate": 5.202716628796602e-06,
      "loss": 0.4179,
      "step": 21292
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.57405686378479,
      "learning_rate": 5.201823082930544e-06,
      "loss": 0.4487,
      "step": 21293
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5149285793304443,
      "learning_rate": 5.200929586829251e-06,
      "loss": 0.4075,
      "step": 21294
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6090714931488037,
      "learning_rate": 5.200036140501989e-06,
      "loss": 0.483,
      "step": 21295
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5893341302871704,
      "learning_rate": 5.199142743958028e-06,
      "loss": 0.4265,
      "step": 21296
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6116489171981812,
      "learning_rate": 5.198249397206634e-06,
      "loss": 0.3797,
      "step": 21297
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5320677757263184,
      "learning_rate": 5.197356100257075e-06,
      "loss": 0.403,
      "step": 21298
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5074169635772705,
      "learning_rate": 5.196462853118613e-06,
      "loss": 0.4068,
      "step": 21299
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5170818567276,
      "learning_rate": 5.19556965580051e-06,
      "loss": 0.3952,
      "step": 21300
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4459805488586426,
      "learning_rate": 5.194676508312031e-06,
      "loss": 0.4279,
      "step": 21301
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4974631071090698,
      "learning_rate": 5.193783410662445e-06,
      "loss": 0.3767,
      "step": 21302
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5424095392227173,
      "learning_rate": 5.1928903628610076e-06,
      "loss": 0.4398,
      "step": 21303
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5448709726333618,
      "learning_rate": 5.191997364916987e-06,
      "loss": 0.3797,
      "step": 21304
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5440562963485718,
      "learning_rate": 5.191104416839638e-06,
      "loss": 0.4679,
      "step": 21305
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5161328315734863,
      "learning_rate": 5.19021151863823e-06,
      "loss": 0.3774,
      "step": 21306
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5747408866882324,
      "learning_rate": 5.189318670322016e-06,
      "loss": 0.4436,
      "step": 21307
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6194244623184204,
      "learning_rate": 5.18842587190026e-06,
      "loss": 0.4083,
      "step": 21308
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5702252388000488,
      "learning_rate": 5.187533123382227e-06,
      "loss": 0.3985,
      "step": 21309
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.542455792427063,
      "learning_rate": 5.186640424777169e-06,
      "loss": 0.3911,
      "step": 21310
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8524225950241089,
      "learning_rate": 5.185747776094344e-06,
      "loss": 1.0388,
      "step": 21311
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4697059392929077,
      "learning_rate": 5.184855177343013e-06,
      "loss": 0.9733,
      "step": 21312
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5119837522506714,
      "learning_rate": 5.183962628532433e-06,
      "loss": 0.3921,
      "step": 21313
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5215981006622314,
      "learning_rate": 5.18307012967187e-06,
      "loss": 0.3829,
      "step": 21314
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5846312046051025,
      "learning_rate": 5.182177680770565e-06,
      "loss": 0.4111,
      "step": 21315
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5585179328918457,
      "learning_rate": 5.18128528183778e-06,
      "loss": 0.4118,
      "step": 21316
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5148307085037231,
      "learning_rate": 5.180392932882774e-06,
      "loss": 0.3953,
      "step": 21317
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5316489934921265,
      "learning_rate": 5.179500633914803e-06,
      "loss": 0.3732,
      "step": 21318
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6247469186782837,
      "learning_rate": 5.17860838494312e-06,
      "loss": 0.4267,
      "step": 21319
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.528147578239441,
      "learning_rate": 5.177716185976973e-06,
      "loss": 0.3824,
      "step": 21320
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6545474529266357,
      "learning_rate": 5.176824037025621e-06,
      "loss": 1.0064,
      "step": 21321
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4081621170043945,
      "learning_rate": 5.175931938098319e-06,
      "loss": 1.0594,
      "step": 21322
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.671007752418518,
      "learning_rate": 5.175039889204314e-06,
      "loss": 0.4409,
      "step": 21323
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.559178113937378,
      "learning_rate": 5.174147890352864e-06,
      "loss": 0.3842,
      "step": 21324
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5376259088516235,
      "learning_rate": 5.173255941553214e-06,
      "loss": 0.4186,
      "step": 21325
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6006267070770264,
      "learning_rate": 5.172364042814622e-06,
      "loss": 0.4085,
      "step": 21326
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5905725955963135,
      "learning_rate": 5.171472194146331e-06,
      "loss": 0.3952,
      "step": 21327
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5571767091751099,
      "learning_rate": 5.1705803955575936e-06,
      "loss": 0.3779,
      "step": 21328
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5832738876342773,
      "learning_rate": 5.169688647057665e-06,
      "loss": 0.462,
      "step": 21329
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4324966669082642,
      "learning_rate": 5.16879694865579e-06,
      "loss": 0.3623,
      "step": 21330
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5484527349472046,
      "learning_rate": 5.1679053003612115e-06,
      "loss": 0.4266,
      "step": 21331
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4734152555465698,
      "learning_rate": 5.167013702183181e-06,
      "loss": 0.4049,
      "step": 21332
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.509425163269043,
      "learning_rate": 5.166122154130947e-06,
      "loss": 0.3987,
      "step": 21333
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4921704530715942,
      "learning_rate": 5.16523065621376e-06,
      "loss": 0.3681,
      "step": 21334
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7058582305908203,
      "learning_rate": 5.164339208440863e-06,
      "loss": 0.4564,
      "step": 21335
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4658998250961304,
      "learning_rate": 5.163447810821496e-06,
      "loss": 0.3717,
      "step": 21336
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5502830743789673,
      "learning_rate": 5.162556463364911e-06,
      "loss": 0.4778,
      "step": 21337
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7001709938049316,
      "learning_rate": 5.161665166080354e-06,
      "loss": 0.4088,
      "step": 21338
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5662007331848145,
      "learning_rate": 5.160773918977061e-06,
      "loss": 0.4338,
      "step": 21339
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5583513975143433,
      "learning_rate": 5.1598827220642866e-06,
      "loss": 0.3898,
      "step": 21340
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5445606708526611,
      "learning_rate": 5.158991575351262e-06,
      "loss": 0.4396,
      "step": 21341
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6182217597961426,
      "learning_rate": 5.158100478847241e-06,
      "loss": 0.4135,
      "step": 21342
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5032074451446533,
      "learning_rate": 5.157209432561457e-06,
      "loss": 0.4296,
      "step": 21343
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5229504108428955,
      "learning_rate": 5.156318436503156e-06,
      "loss": 0.3871,
      "step": 21344
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.552884578704834,
      "learning_rate": 5.155427490681581e-06,
      "loss": 0.3914,
      "step": 21345
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5630407333374023,
      "learning_rate": 5.15453659510597e-06,
      "loss": 0.3896,
      "step": 21346
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.540121078491211,
      "learning_rate": 5.153645749785559e-06,
      "loss": 0.4101,
      "step": 21347
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.502057671546936,
      "learning_rate": 5.152754954729591e-06,
      "loss": 0.3841,
      "step": 21348
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5267881155014038,
      "learning_rate": 5.151864209947306e-06,
      "loss": 0.3908,
      "step": 21349
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5008872747421265,
      "learning_rate": 5.150973515447945e-06,
      "loss": 0.3807,
      "step": 21350
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6288771629333496,
      "learning_rate": 5.150082871240742e-06,
      "loss": 0.4802,
      "step": 21351
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5569266080856323,
      "learning_rate": 5.149192277334931e-06,
      "loss": 0.396,
      "step": 21352
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5895967483520508,
      "learning_rate": 5.148301733739754e-06,
      "loss": 0.4311,
      "step": 21353
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5965461730957031,
      "learning_rate": 5.14741124046445e-06,
      "loss": 0.3999,
      "step": 21354
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5977206230163574,
      "learning_rate": 5.146520797518251e-06,
      "loss": 0.4663,
      "step": 21355
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5014528036117554,
      "learning_rate": 5.1456304049103886e-06,
      "loss": 0.3896,
      "step": 21356
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7720764875411987,
      "learning_rate": 5.144740062650102e-06,
      "loss": 0.4663,
      "step": 21357
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5454809665679932,
      "learning_rate": 5.143849770746627e-06,
      "loss": 0.3906,
      "step": 21358
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6473939418792725,
      "learning_rate": 5.142959529209194e-06,
      "loss": 0.4761,
      "step": 21359
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4892284870147705,
      "learning_rate": 5.14206933804704e-06,
      "loss": 0.411,
      "step": 21360
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5376805067062378,
      "learning_rate": 5.14117919726939e-06,
      "loss": 0.3963,
      "step": 21361
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4916177988052368,
      "learning_rate": 5.140289106885488e-06,
      "loss": 0.4077,
      "step": 21362
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5101372003555298,
      "learning_rate": 5.139399066904553e-06,
      "loss": 0.4147,
      "step": 21363
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.503282904624939,
      "learning_rate": 5.138509077335823e-06,
      "loss": 0.3885,
      "step": 21364
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.50663423538208,
      "learning_rate": 5.137619138188531e-06,
      "loss": 0.4343,
      "step": 21365
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.49205482006073,
      "learning_rate": 5.136729249471905e-06,
      "loss": 0.3847,
      "step": 21366
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.580338716506958,
      "learning_rate": 5.135839411195169e-06,
      "loss": 0.4867,
      "step": 21367
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4482780694961548,
      "learning_rate": 5.134949623367555e-06,
      "loss": 0.3807,
      "step": 21368
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.539303183555603,
      "learning_rate": 5.134059885998295e-06,
      "loss": 0.4728,
      "step": 21369
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.584609031677246,
      "learning_rate": 5.133170199096619e-06,
      "loss": 0.4016,
      "step": 21370
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6540712118148804,
      "learning_rate": 5.13228056267175e-06,
      "loss": 0.4559,
      "step": 21371
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5604581832885742,
      "learning_rate": 5.1313909767329105e-06,
      "loss": 0.4049,
      "step": 21372
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5574287176132202,
      "learning_rate": 5.130501441289333e-06,
      "loss": 0.4588,
      "step": 21373
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5512444972991943,
      "learning_rate": 5.129611956350247e-06,
      "loss": 0.4146,
      "step": 21374
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.585731029510498,
      "learning_rate": 5.128722521924867e-06,
      "loss": 0.4355,
      "step": 21375
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.591162085533142,
      "learning_rate": 5.1278331380224286e-06,
      "loss": 0.4011,
      "step": 21376
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.536934733390808,
      "learning_rate": 5.126943804652149e-06,
      "loss": 0.4029,
      "step": 21377
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5558661222457886,
      "learning_rate": 5.126054521823257e-06,
      "loss": 0.3795,
      "step": 21378
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.500506043434143,
      "learning_rate": 5.125165289544971e-06,
      "loss": 0.427,
      "step": 21379
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5168259143829346,
      "learning_rate": 5.1242761078265145e-06,
      "loss": 0.3894,
      "step": 21380
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7518025636672974,
      "learning_rate": 5.123386976677116e-06,
      "loss": 0.4083,
      "step": 21381
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4922287464141846,
      "learning_rate": 5.122497896105993e-06,
      "loss": 0.3871,
      "step": 21382
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5773589611053467,
      "learning_rate": 5.121608866122363e-06,
      "loss": 0.4161,
      "step": 21383
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.493037462234497,
      "learning_rate": 5.120719886735449e-06,
      "loss": 0.3914,
      "step": 21384
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5711323022842407,
      "learning_rate": 5.119830957954472e-06,
      "loss": 0.4386,
      "step": 21385
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5645370483398438,
      "learning_rate": 5.1189420797886555e-06,
      "loss": 0.3804,
      "step": 21386
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6242456436157227,
      "learning_rate": 5.118053252247216e-06,
      "loss": 0.4459,
      "step": 21387
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.618795394897461,
      "learning_rate": 5.117164475339367e-06,
      "loss": 0.4106,
      "step": 21388
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5782986879348755,
      "learning_rate": 5.1162757490743285e-06,
      "loss": 0.4577,
      "step": 21389
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8158924579620361,
      "learning_rate": 5.1153870734613265e-06,
      "loss": 0.435,
      "step": 21390
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5347269773483276,
      "learning_rate": 5.11449844850957e-06,
      "loss": 0.4314,
      "step": 21391
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.530890703201294,
      "learning_rate": 5.113609874228273e-06,
      "loss": 0.4235,
      "step": 21392
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5553640127182007,
      "learning_rate": 5.112721350626655e-06,
      "loss": 0.4594,
      "step": 21393
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4414836168289185,
      "learning_rate": 5.111832877713936e-06,
      "loss": 0.3794,
      "step": 21394
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9393094778060913,
      "learning_rate": 5.110944455499323e-06,
      "loss": 1.047,
      "step": 21395
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5104960203170776,
      "learning_rate": 5.110056083992039e-06,
      "loss": 1.0756,
      "step": 21396
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7840533256530762,
      "learning_rate": 5.109167763201287e-06,
      "loss": 0.5278,
      "step": 21397
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4847532510757446,
      "learning_rate": 5.10827949313629e-06,
      "loss": 0.375,
      "step": 21398
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.531891942024231,
      "learning_rate": 5.107391273806255e-06,
      "loss": 0.4547,
      "step": 21399
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5821406841278076,
      "learning_rate": 5.106503105220394e-06,
      "loss": 0.3839,
      "step": 21400
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6253132820129395,
      "learning_rate": 5.105614987387926e-06,
      "loss": 0.9335,
      "step": 21401
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4202625751495361,
      "learning_rate": 5.104726920318057e-06,
      "loss": 1.0385,
      "step": 21402
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6227773427963257,
      "learning_rate": 5.103838904019993e-06,
      "loss": 0.4472,
      "step": 21403
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4637356996536255,
      "learning_rate": 5.102950938502949e-06,
      "loss": 0.3728,
      "step": 21404
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8340352773666382,
      "learning_rate": 5.102063023776136e-06,
      "loss": 0.981,
      "step": 21405
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4342694282531738,
      "learning_rate": 5.101175159848765e-06,
      "loss": 1.0072,
      "step": 21406
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.524039626121521,
      "learning_rate": 5.1002873467300405e-06,
      "loss": 0.4211,
      "step": 21407
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5272889137268066,
      "learning_rate": 5.099399584429168e-06,
      "loss": 0.3698,
      "step": 21408
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5931494235992432,
      "learning_rate": 5.0985118729553586e-06,
      "loss": 0.4583,
      "step": 21409
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5842714309692383,
      "learning_rate": 5.097624212317822e-06,
      "loss": 0.4076,
      "step": 21410
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5152605772018433,
      "learning_rate": 5.096736602525758e-06,
      "loss": 0.4162,
      "step": 21411
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4506957530975342,
      "learning_rate": 5.095849043588379e-06,
      "loss": 0.4311,
      "step": 21412
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4717553853988647,
      "learning_rate": 5.094961535514885e-06,
      "loss": 0.3827,
      "step": 21413
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4922555685043335,
      "learning_rate": 5.094074078314487e-06,
      "loss": 0.4044,
      "step": 21414
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5240870714187622,
      "learning_rate": 5.093186671996382e-06,
      "loss": 0.4138,
      "step": 21415
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5819871425628662,
      "learning_rate": 5.092299316569778e-06,
      "loss": 0.4236,
      "step": 21416
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5598068237304688,
      "learning_rate": 5.091412012043881e-06,
      "loss": 0.4369,
      "step": 21417
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.584700345993042,
      "learning_rate": 5.090524758427891e-06,
      "loss": 0.3911,
      "step": 21418
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6404197216033936,
      "learning_rate": 5.089637555731006e-06,
      "loss": 0.4217,
      "step": 21419
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5372430086135864,
      "learning_rate": 5.088750403962431e-06,
      "loss": 0.3889,
      "step": 21420
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.574313998222351,
      "learning_rate": 5.0878633031313685e-06,
      "loss": 0.4342,
      "step": 21421
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.564560055732727,
      "learning_rate": 5.086976253247023e-06,
      "loss": 0.4129,
      "step": 21422
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.444109559059143,
      "learning_rate": 5.086089254318589e-06,
      "loss": 0.3951,
      "step": 21423
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5283970832824707,
      "learning_rate": 5.085202306355264e-06,
      "loss": 0.3993,
      "step": 21424
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5637634992599487,
      "learning_rate": 5.0843154093662495e-06,
      "loss": 0.4808,
      "step": 21425
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.491215705871582,
      "learning_rate": 5.083428563360749e-06,
      "loss": 0.3989,
      "step": 21426
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.542123556137085,
      "learning_rate": 5.082541768347952e-06,
      "loss": 0.4243,
      "step": 21427
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8482670783996582,
      "learning_rate": 5.081655024337064e-06,
      "loss": 0.4182,
      "step": 21428
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6105897426605225,
      "learning_rate": 5.080768331337276e-06,
      "loss": 0.408,
      "step": 21429
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5187405347824097,
      "learning_rate": 5.079881689357789e-06,
      "loss": 0.3953,
      "step": 21430
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5232930183410645,
      "learning_rate": 5.078995098407793e-06,
      "loss": 0.4268,
      "step": 21431
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5972403287887573,
      "learning_rate": 5.0781085584964905e-06,
      "loss": 0.4203,
      "step": 21432
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5931153297424316,
      "learning_rate": 5.07722206963307e-06,
      "loss": 0.44,
      "step": 21433
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4959235191345215,
      "learning_rate": 5.076335631826731e-06,
      "loss": 0.383,
      "step": 21434
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.59663724899292,
      "learning_rate": 5.075449245086661e-06,
      "loss": 0.4522,
      "step": 21435
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6241971254348755,
      "learning_rate": 5.074562909422058e-06,
      "loss": 0.3821,
      "step": 21436
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7728838920593262,
      "learning_rate": 5.073676624842119e-06,
      "loss": 0.4072,
      "step": 21437
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5213404893875122,
      "learning_rate": 5.072790391356029e-06,
      "loss": 0.4014,
      "step": 21438
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7181979417800903,
      "learning_rate": 5.071904208972977e-06,
      "loss": 0.44,
      "step": 21439
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.537357211112976,
      "learning_rate": 5.071018077702161e-06,
      "loss": 0.4018,
      "step": 21440
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5729060173034668,
      "learning_rate": 5.070131997552768e-06,
      "loss": 0.4194,
      "step": 21441
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5171607732772827,
      "learning_rate": 5.069245968533994e-06,
      "loss": 0.3748,
      "step": 21442
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.544538140296936,
      "learning_rate": 5.068359990655023e-06,
      "loss": 0.4126,
      "step": 21443
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4993836879730225,
      "learning_rate": 5.067474063925042e-06,
      "loss": 0.3847,
      "step": 21444
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6302616596221924,
      "learning_rate": 5.066588188353242e-06,
      "loss": 0.4444,
      "step": 21445
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5006989240646362,
      "learning_rate": 5.065702363948812e-06,
      "loss": 0.3896,
      "step": 21446
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5486565828323364,
      "learning_rate": 5.064816590720942e-06,
      "loss": 0.4593,
      "step": 21447
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5135157108306885,
      "learning_rate": 5.063930868678817e-06,
      "loss": 0.3945,
      "step": 21448
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.559643030166626,
      "learning_rate": 5.063045197831616e-06,
      "loss": 0.4228,
      "step": 21449
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5210258960723877,
      "learning_rate": 5.062159578188533e-06,
      "loss": 0.3687,
      "step": 21450
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.650424599647522,
      "learning_rate": 5.0612740097587545e-06,
      "loss": 0.4468,
      "step": 21451
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6563701629638672,
      "learning_rate": 5.060388492551457e-06,
      "loss": 0.3628,
      "step": 21452
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6448911428451538,
      "learning_rate": 5.0595030265758335e-06,
      "loss": 0.423,
      "step": 21453
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5039854049682617,
      "learning_rate": 5.058617611841061e-06,
      "loss": 0.3843,
      "step": 21454
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5022692680358887,
      "learning_rate": 5.05773224835633e-06,
      "loss": 0.4487,
      "step": 21455
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.421051263809204,
      "learning_rate": 5.056846936130814e-06,
      "loss": 0.3926,
      "step": 21456
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5554742813110352,
      "learning_rate": 5.0559616751737e-06,
      "loss": 0.4092,
      "step": 21457
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5839459896087646,
      "learning_rate": 5.055076465494173e-06,
      "loss": 0.3896,
      "step": 21458
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5564076900482178,
      "learning_rate": 5.054191307101411e-06,
      "loss": 0.4578,
      "step": 21459
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5384336709976196,
      "learning_rate": 5.053306200004588e-06,
      "loss": 0.3996,
      "step": 21460
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7491310834884644,
      "learning_rate": 5.052421144212891e-06,
      "loss": 0.4845,
      "step": 21461
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5434132814407349,
      "learning_rate": 5.0515361397355005e-06,
      "loss": 0.3976,
      "step": 21462
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5835487842559814,
      "learning_rate": 5.050651186581594e-06,
      "loss": 0.4661,
      "step": 21463
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6124428510665894,
      "learning_rate": 5.04976628476035e-06,
      "loss": 0.3913,
      "step": 21464
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5656651258468628,
      "learning_rate": 5.0488814342809425e-06,
      "loss": 0.4834,
      "step": 21465
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5409307479858398,
      "learning_rate": 5.04799663515255e-06,
      "loss": 0.3843,
      "step": 21466
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7503379583358765,
      "learning_rate": 5.047111887384357e-06,
      "loss": 0.9479,
      "step": 21467
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5071840286254883,
      "learning_rate": 5.0462271909855286e-06,
      "loss": 1.0574,
      "step": 21468
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6058578491210938,
      "learning_rate": 5.04534254596525e-06,
      "loss": 0.4522,
      "step": 21469
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6236144304275513,
      "learning_rate": 5.044457952332689e-06,
      "loss": 0.3937,
      "step": 21470
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6124398708343506,
      "learning_rate": 5.043573410097026e-06,
      "loss": 0.4632,
      "step": 21471
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5286558866500854,
      "learning_rate": 5.042688919267429e-06,
      "loss": 0.4047,
      "step": 21472
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5375163555145264,
      "learning_rate": 5.041804479853079e-06,
      "loss": 0.4237,
      "step": 21473
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4797569513320923,
      "learning_rate": 5.0409200918631406e-06,
      "loss": 0.3784,
      "step": 21474
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.2592618465423584,
      "learning_rate": 5.0400357553067956e-06,
      "loss": 0.4101,
      "step": 21475
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4784475564956665,
      "learning_rate": 5.039151470193207e-06,
      "loss": 0.3724,
      "step": 21476
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.612821102142334,
      "learning_rate": 5.038267236531551e-06,
      "loss": 0.4662,
      "step": 21477
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.582823395729065,
      "learning_rate": 5.037383054331e-06,
      "loss": 0.4026,
      "step": 21478
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6758960485458374,
      "learning_rate": 5.036498923600725e-06,
      "loss": 0.4771,
      "step": 21479
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5403748750686646,
      "learning_rate": 5.035614844349888e-06,
      "loss": 0.3934,
      "step": 21480
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.563146710395813,
      "learning_rate": 5.034730816587663e-06,
      "loss": 0.4435,
      "step": 21481
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5684492588043213,
      "learning_rate": 5.03384684032322e-06,
      "loss": 0.4013,
      "step": 21482
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.572433590888977,
      "learning_rate": 5.032962915565729e-06,
      "loss": 0.428,
      "step": 21483
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.472658634185791,
      "learning_rate": 5.032079042324357e-06,
      "loss": 0.3798,
      "step": 21484
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7089636325836182,
      "learning_rate": 5.031195220608265e-06,
      "loss": 0.4232,
      "step": 21485
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.444652795791626,
      "learning_rate": 5.030311450426624e-06,
      "loss": 0.3613,
      "step": 21486
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6230649948120117,
      "learning_rate": 5.029427731788604e-06,
      "loss": 0.4593,
      "step": 21487
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.514858603477478,
      "learning_rate": 5.0285440647033634e-06,
      "loss": 0.3948,
      "step": 21488
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5676112174987793,
      "learning_rate": 5.027660449180074e-06,
      "loss": 0.4284,
      "step": 21489
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5410279035568237,
      "learning_rate": 5.026776885227893e-06,
      "loss": 0.3962,
      "step": 21490
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5128806829452515,
      "learning_rate": 5.0258933728559926e-06,
      "loss": 0.3883,
      "step": 21491
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6250014305114746,
      "learning_rate": 5.025009912073528e-06,
      "loss": 0.4218,
      "step": 21492
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5541143417358398,
      "learning_rate": 5.0241265028896655e-06,
      "loss": 0.4669,
      "step": 21493
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5136181116104126,
      "learning_rate": 5.023243145313572e-06,
      "loss": 0.3964,
      "step": 21494
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5789577960968018,
      "learning_rate": 5.022359839354406e-06,
      "loss": 0.4315,
      "step": 21495
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5203090906143188,
      "learning_rate": 5.021476585021323e-06,
      "loss": 0.4099,
      "step": 21496
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6023222208023071,
      "learning_rate": 5.020593382323491e-06,
      "loss": 0.4185,
      "step": 21497
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5685365200042725,
      "learning_rate": 5.019710231270066e-06,
      "loss": 0.3588,
      "step": 21498
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5439926385879517,
      "learning_rate": 5.018827131870214e-06,
      "loss": 0.4536,
      "step": 21499
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5763626098632812,
      "learning_rate": 5.017944084133091e-06,
      "loss": 0.3946,
      "step": 21500
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6481928825378418,
      "learning_rate": 5.017061088067849e-06,
      "loss": 0.45,
      "step": 21501
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5953304767608643,
      "learning_rate": 5.016178143683654e-06,
      "loss": 0.4118,
      "step": 21502
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5625194311141968,
      "learning_rate": 5.0152952509896636e-06,
      "loss": 0.4757,
      "step": 21503
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5055063962936401,
      "learning_rate": 5.0144124099950285e-06,
      "loss": 0.4066,
      "step": 21504
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.583572506904602,
      "learning_rate": 5.013529620708912e-06,
      "loss": 0.4767,
      "step": 21505
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5632860660552979,
      "learning_rate": 5.012646883140465e-06,
      "loss": 0.3936,
      "step": 21506
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5927808284759521,
      "learning_rate": 5.011764197298848e-06,
      "loss": 0.4613,
      "step": 21507
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5978987216949463,
      "learning_rate": 5.01088156319321e-06,
      "loss": 0.4393,
      "step": 21508
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7914667129516602,
      "learning_rate": 5.009998980832711e-06,
      "loss": 0.431,
      "step": 21509
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5297054052352905,
      "learning_rate": 5.0091164502264985e-06,
      "loss": 0.4085,
      "step": 21510
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6564216613769531,
      "learning_rate": 5.0082339713837325e-06,
      "loss": 0.4151,
      "step": 21511
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.54961097240448,
      "learning_rate": 5.007351544313559e-06,
      "loss": 0.4059,
      "step": 21512
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5462745428085327,
      "learning_rate": 5.0064691690251346e-06,
      "loss": 0.4216,
      "step": 21513
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5230225324630737,
      "learning_rate": 5.005586845527612e-06,
      "loss": 0.3831,
      "step": 21514
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4954432249069214,
      "learning_rate": 5.0047045738301415e-06,
      "loss": 0.4089,
      "step": 21515
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5826919078826904,
      "learning_rate": 5.003822353941869e-06,
      "loss": 0.3937,
      "step": 21516
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5929940938949585,
      "learning_rate": 5.002940185871947e-06,
      "loss": 0.4163,
      "step": 21517
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5286877155303955,
      "learning_rate": 5.002058069629525e-06,
      "loss": 0.3962,
      "step": 21518
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5610558986663818,
      "learning_rate": 5.001176005223759e-06,
      "loss": 0.4719,
      "step": 21519
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5295268297195435,
      "learning_rate": 5.00029399266379e-06,
      "loss": 0.3778,
      "step": 21520
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5653226375579834,
      "learning_rate": 4.9994120319587634e-06,
      "loss": 0.4258,
      "step": 21521
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5104111433029175,
      "learning_rate": 4.998530123117832e-06,
      "loss": 0.3946,
      "step": 21522
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5621216297149658,
      "learning_rate": 4.997648266150142e-06,
      "loss": 0.4606,
      "step": 21523
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5670758485794067,
      "learning_rate": 4.996766461064837e-06,
      "loss": 0.3914,
      "step": 21524
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5806300640106201,
      "learning_rate": 4.995884707871068e-06,
      "loss": 0.4221,
      "step": 21525
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5575908422470093,
      "learning_rate": 4.995003006577972e-06,
      "loss": 0.3949,
      "step": 21526
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5495915412902832,
      "learning_rate": 4.994121357194702e-06,
      "loss": 0.443,
      "step": 21527
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8161702156066895,
      "learning_rate": 4.993239759730395e-06,
      "loss": 0.4327,
      "step": 21528
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.540852427482605,
      "learning_rate": 4.9923582141941985e-06,
      "loss": 0.395,
      "step": 21529
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.499072790145874,
      "learning_rate": 4.991476720595257e-06,
      "loss": 0.3758,
      "step": 21530
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5924383401870728,
      "learning_rate": 4.990595278942712e-06,
      "loss": 0.4387,
      "step": 21531
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4627406597137451,
      "learning_rate": 4.989713889245702e-06,
      "loss": 0.3804,
      "step": 21532
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6276146173477173,
      "learning_rate": 4.9888325515133685e-06,
      "loss": 0.4339,
      "step": 21533
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.475089430809021,
      "learning_rate": 4.9879512657548555e-06,
      "loss": 0.3617,
      "step": 21534
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5730215311050415,
      "learning_rate": 4.987070031979307e-06,
      "loss": 0.4198,
      "step": 21535
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5473579168319702,
      "learning_rate": 4.9861888501958575e-06,
      "loss": 0.3801,
      "step": 21536
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.583486795425415,
      "learning_rate": 4.985307720413644e-06,
      "loss": 0.4704,
      "step": 21537
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4789915084838867,
      "learning_rate": 4.984426642641808e-06,
      "loss": 0.3557,
      "step": 21538
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8062801361083984,
      "learning_rate": 4.9835456168894914e-06,
      "loss": 0.9639,
      "step": 21539
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4442806243896484,
      "learning_rate": 4.982664643165826e-06,
      "loss": 1.0168,
      "step": 21540
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6099013090133667,
      "learning_rate": 4.9817837214799535e-06,
      "loss": 0.4504,
      "step": 21541
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.512005090713501,
      "learning_rate": 4.980902851841005e-06,
      "loss": 0.4045,
      "step": 21542
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6289170980453491,
      "learning_rate": 4.980022034258124e-06,
      "loss": 0.4729,
      "step": 21543
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.452085018157959,
      "learning_rate": 4.979141268740439e-06,
      "loss": 0.3959,
      "step": 21544
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6997451782226562,
      "learning_rate": 4.978260555297086e-06,
      "loss": 0.4302,
      "step": 21545
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.462090253829956,
      "learning_rate": 4.977379893937205e-06,
      "loss": 0.3751,
      "step": 21546
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5455156564712524,
      "learning_rate": 4.976499284669928e-06,
      "loss": 0.4377,
      "step": 21547
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6907176971435547,
      "learning_rate": 4.975618727504381e-06,
      "loss": 0.4171,
      "step": 21548
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1457877159118652,
      "learning_rate": 4.974738222449703e-06,
      "loss": 0.4429,
      "step": 21549
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.588443398475647,
      "learning_rate": 4.9738577695150294e-06,
      "loss": 0.4004,
      "step": 21550
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4563446044921875,
      "learning_rate": 4.972977368709487e-06,
      "loss": 0.4112,
      "step": 21551
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6003261804580688,
      "learning_rate": 4.972097020042205e-06,
      "loss": 0.3951,
      "step": 21552
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5079290866851807,
      "learning_rate": 4.971216723522317e-06,
      "loss": 0.421,
      "step": 21553
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5245780944824219,
      "learning_rate": 4.970336479158951e-06,
      "loss": 0.3689,
      "step": 21554
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5858585834503174,
      "learning_rate": 4.969456286961244e-06,
      "loss": 0.4733,
      "step": 21555
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5007822513580322,
      "learning_rate": 4.96857614693832e-06,
      "loss": 0.3656,
      "step": 21556
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6846275329589844,
      "learning_rate": 4.967696059099302e-06,
      "loss": 0.4796,
      "step": 21557
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5723607540130615,
      "learning_rate": 4.966816023453323e-06,
      "loss": 0.3544,
      "step": 21558
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.688446283340454,
      "learning_rate": 4.965936040009516e-06,
      "loss": 0.4927,
      "step": 21559
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.45577871799469,
      "learning_rate": 4.965056108776997e-06,
      "loss": 0.3644,
      "step": 21560
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5643188953399658,
      "learning_rate": 4.964176229764901e-06,
      "loss": 0.4085,
      "step": 21561
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5475828647613525,
      "learning_rate": 4.963296402982346e-06,
      "loss": 0.3853,
      "step": 21562
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.590775966644287,
      "learning_rate": 4.962416628438466e-06,
      "loss": 0.4804,
      "step": 21563
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.521031379699707,
      "learning_rate": 4.961536906142377e-06,
      "loss": 0.3888,
      "step": 21564
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.580133318901062,
      "learning_rate": 4.960657236103207e-06,
      "loss": 0.4379,
      "step": 21565
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4961133003234863,
      "learning_rate": 4.959777618330084e-06,
      "loss": 0.3698,
      "step": 21566
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5486747026443481,
      "learning_rate": 4.958898052832127e-06,
      "loss": 0.4777,
      "step": 21567
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.521235466003418,
      "learning_rate": 4.9580185396184546e-06,
      "loss": 0.4024,
      "step": 21568
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5606985092163086,
      "learning_rate": 4.957139078698191e-06,
      "loss": 0.4508,
      "step": 21569
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5231822729110718,
      "learning_rate": 4.956259670080461e-06,
      "loss": 0.4026,
      "step": 21570
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5779207944869995,
      "learning_rate": 4.955380313774386e-06,
      "loss": 0.4269,
      "step": 21571
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5551657676696777,
      "learning_rate": 4.954501009789084e-06,
      "loss": 0.4586,
      "step": 21572
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.593762755393982,
      "learning_rate": 4.953621758133672e-06,
      "loss": 0.4264,
      "step": 21573
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5328994989395142,
      "learning_rate": 4.95274255881727e-06,
      "loss": 0.3664,
      "step": 21574
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5433417558670044,
      "learning_rate": 4.951863411849005e-06,
      "loss": 0.427,
      "step": 21575
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.474148154258728,
      "learning_rate": 4.9509843172379835e-06,
      "loss": 0.3524,
      "step": 21576
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.686898112297058,
      "learning_rate": 4.950105274993332e-06,
      "loss": 0.9023,
      "step": 21577
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.523478627204895,
      "learning_rate": 4.949226285124161e-06,
      "loss": 0.9722,
      "step": 21578
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5321736335754395,
      "learning_rate": 4.948347347639594e-06,
      "loss": 0.4392,
      "step": 21579
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4506077766418457,
      "learning_rate": 4.947468462548738e-06,
      "loss": 0.3812,
      "step": 21580
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7145296335220337,
      "learning_rate": 4.946589629860714e-06,
      "loss": 0.4765,
      "step": 21581
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5115315914154053,
      "learning_rate": 4.9457108495846406e-06,
      "loss": 0.387,
      "step": 21582
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9756394624710083,
      "learning_rate": 4.944832121729627e-06,
      "loss": 0.4532,
      "step": 21583
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5480488538742065,
      "learning_rate": 4.943953446304786e-06,
      "loss": 0.4028,
      "step": 21584
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6692689657211304,
      "learning_rate": 4.943074823319231e-06,
      "loss": 0.4392,
      "step": 21585
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.536478877067566,
      "learning_rate": 4.942196252782077e-06,
      "loss": 0.3647,
      "step": 21586
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8626614809036255,
      "learning_rate": 4.941317734702443e-06,
      "loss": 0.4536,
      "step": 21587
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5037403106689453,
      "learning_rate": 4.940439269089426e-06,
      "loss": 0.3987,
      "step": 21588
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.514415979385376,
      "learning_rate": 4.939560855952145e-06,
      "loss": 0.4321,
      "step": 21589
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5064246654510498,
      "learning_rate": 4.938682495299708e-06,
      "loss": 0.3876,
      "step": 21590
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6318098306655884,
      "learning_rate": 4.937804187141233e-06,
      "loss": 0.4221,
      "step": 21591
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5624028444290161,
      "learning_rate": 4.936925931485822e-06,
      "loss": 0.4009,
      "step": 21592
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5494184494018555,
      "learning_rate": 4.936047728342583e-06,
      "loss": 0.4806,
      "step": 21593
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5569263696670532,
      "learning_rate": 4.935169577720626e-06,
      "loss": 0.4002,
      "step": 21594
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5219581127166748,
      "learning_rate": 4.934291479629063e-06,
      "loss": 0.4431,
      "step": 21595
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.598371148109436,
      "learning_rate": 4.933413434076993e-06,
      "loss": 0.4454,
      "step": 21596
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6150763034820557,
      "learning_rate": 4.932535441073533e-06,
      "loss": 0.4862,
      "step": 21597
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.538787603378296,
      "learning_rate": 4.93165750062778e-06,
      "loss": 0.3991,
      "step": 21598
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5138803720474243,
      "learning_rate": 4.930779612748847e-06,
      "loss": 0.4323,
      "step": 21599
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5488797426223755,
      "learning_rate": 4.9299017774458316e-06,
      "loss": 0.3993,
      "step": 21600
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5433340072631836,
      "learning_rate": 4.929023994727842e-06,
      "loss": 0.4143,
      "step": 21601
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.578042984008789,
      "learning_rate": 4.9281462646039866e-06,
      "loss": 0.4199,
      "step": 21602
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6000186204910278,
      "learning_rate": 4.927268587083365e-06,
      "loss": 0.432,
      "step": 21603
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5378642082214355,
      "learning_rate": 4.926390962175075e-06,
      "loss": 0.3916,
      "step": 21604
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6981273889541626,
      "learning_rate": 4.925513389888225e-06,
      "loss": 0.9221,
      "step": 21605
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3777965307235718,
      "learning_rate": 4.924635870231915e-06,
      "loss": 1.0097,
      "step": 21606
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6020519733428955,
      "learning_rate": 4.923758403215251e-06,
      "loss": 0.4638,
      "step": 21607
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.582789659500122,
      "learning_rate": 4.922880988847329e-06,
      "loss": 0.3938,
      "step": 21608
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6444289684295654,
      "learning_rate": 4.9220036271372475e-06,
      "loss": 0.4533,
      "step": 21609
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.548505425453186,
      "learning_rate": 4.921126318094107e-06,
      "loss": 0.3795,
      "step": 21610
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5828771591186523,
      "learning_rate": 4.9202490617270125e-06,
      "loss": 0.4624,
      "step": 21611
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5368332862854004,
      "learning_rate": 4.9193718580450554e-06,
      "loss": 0.405,
      "step": 21612
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7836147546768188,
      "learning_rate": 4.91849470705734e-06,
      "loss": 1.009,
      "step": 21613
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5102591514587402,
      "learning_rate": 4.9176176087729545e-06,
      "loss": 1.0158,
      "step": 21614
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6023787260055542,
      "learning_rate": 4.916740563201007e-06,
      "loss": 0.3974,
      "step": 21615
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5929232835769653,
      "learning_rate": 4.915863570350584e-06,
      "loss": 0.4381,
      "step": 21616
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5749393701553345,
      "learning_rate": 4.9149866302307855e-06,
      "loss": 0.4623,
      "step": 21617
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5840338468551636,
      "learning_rate": 4.9141097428507125e-06,
      "loss": 0.3916,
      "step": 21618
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5495024919509888,
      "learning_rate": 4.913232908219452e-06,
      "loss": 0.4158,
      "step": 21619
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.64847993850708,
      "learning_rate": 4.9123561263460986e-06,
      "loss": 0.4235,
      "step": 21620
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.566757321357727,
      "learning_rate": 4.911479397239746e-06,
      "loss": 0.4155,
      "step": 21621
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4612889289855957,
      "learning_rate": 4.9106027209094895e-06,
      "loss": 0.3814,
      "step": 21622
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6230233907699585,
      "learning_rate": 4.909726097364429e-06,
      "loss": 0.447,
      "step": 21623
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5396854877471924,
      "learning_rate": 4.9088495266136404e-06,
      "loss": 0.3785,
      "step": 21624
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5417357683181763,
      "learning_rate": 4.907973008666225e-06,
      "loss": 0.423,
      "step": 21625
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4876364469528198,
      "learning_rate": 4.90709654353127e-06,
      "loss": 0.3974,
      "step": 21626
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5689412355422974,
      "learning_rate": 4.906220131217873e-06,
      "loss": 0.4424,
      "step": 21627
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4823887348175049,
      "learning_rate": 4.905343771735118e-06,
      "loss": 0.3671,
      "step": 21628
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6133562326431274,
      "learning_rate": 4.9044674650920905e-06,
      "loss": 0.438,
      "step": 21629
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.478963017463684,
      "learning_rate": 4.903591211297886e-06,
      "loss": 0.3721,
      "step": 21630
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.585453987121582,
      "learning_rate": 4.902715010361591e-06,
      "loss": 0.4394,
      "step": 21631
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6002743244171143,
      "learning_rate": 4.901838862292291e-06,
      "loss": 0.3936,
      "step": 21632
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5343066453933716,
      "learning_rate": 4.900962767099077e-06,
      "loss": 0.4792,
      "step": 21633
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5983821153640747,
      "learning_rate": 4.900086724791029e-06,
      "loss": 0.4028,
      "step": 21634
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5644409656524658,
      "learning_rate": 4.89921073537724e-06,
      "loss": 0.4193,
      "step": 21635
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.538975715637207,
      "learning_rate": 4.8983347988667896e-06,
      "loss": 0.3971,
      "step": 21636
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.976876139640808,
      "learning_rate": 4.897458915268765e-06,
      "loss": 0.4366,
      "step": 21637
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.458411693572998,
      "learning_rate": 4.896583084592254e-06,
      "loss": 0.3919,
      "step": 21638
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5245094299316406,
      "learning_rate": 4.895707306846338e-06,
      "loss": 0.4032,
      "step": 21639
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5912095308303833,
      "learning_rate": 4.894831582040095e-06,
      "loss": 0.4024,
      "step": 21640
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5560358762741089,
      "learning_rate": 4.893955910182613e-06,
      "loss": 0.4672,
      "step": 21641
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4796807765960693,
      "learning_rate": 4.893080291282972e-06,
      "loss": 0.384,
      "step": 21642
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5291271209716797,
      "learning_rate": 4.892204725350259e-06,
      "loss": 0.4301,
      "step": 21643
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5042346715927124,
      "learning_rate": 4.891329212393551e-06,
      "loss": 0.356,
      "step": 21644
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8193869590759277,
      "learning_rate": 4.890453752421924e-06,
      "loss": 0.4482,
      "step": 21645
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5750916004180908,
      "learning_rate": 4.889578345444462e-06,
      "loss": 0.4045,
      "step": 21646
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5391021966934204,
      "learning_rate": 4.88870299147025e-06,
      "loss": 0.4421,
      "step": 21647
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5474733114242554,
      "learning_rate": 4.887827690508356e-06,
      "loss": 0.3696,
      "step": 21648
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5228291749954224,
      "learning_rate": 4.886952442567868e-06,
      "loss": 0.425,
      "step": 21649
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.621429204940796,
      "learning_rate": 4.886077247657856e-06,
      "loss": 0.391,
      "step": 21650
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5721869468688965,
      "learning_rate": 4.885202105787403e-06,
      "loss": 0.4579,
      "step": 21651
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.517281174659729,
      "learning_rate": 4.88432701696558e-06,
      "loss": 0.386,
      "step": 21652
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5882331132888794,
      "learning_rate": 4.883451981201466e-06,
      "loss": 0.4598,
      "step": 21653
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5349111557006836,
      "learning_rate": 4.882576998504141e-06,
      "loss": 0.3982,
      "step": 21654
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5817439556121826,
      "learning_rate": 4.881702068882676e-06,
      "loss": 0.4368,
      "step": 21655
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5195519924163818,
      "learning_rate": 4.88082719234614e-06,
      "loss": 0.3853,
      "step": 21656
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5329104661941528,
      "learning_rate": 4.879952368903613e-06,
      "loss": 0.4283,
      "step": 21657
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.540662407875061,
      "learning_rate": 4.879077598564167e-06,
      "loss": 0.3948,
      "step": 21658
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6623083353042603,
      "learning_rate": 4.8782028813368785e-06,
      "loss": 0.5244,
      "step": 21659
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.566342830657959,
      "learning_rate": 4.8773282172308165e-06,
      "loss": 0.3685,
      "step": 21660
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6117546558380127,
      "learning_rate": 4.8764536062550495e-06,
      "loss": 0.4629,
      "step": 21661
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4884321689605713,
      "learning_rate": 4.875579048418652e-06,
      "loss": 0.3754,
      "step": 21662
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.627190351486206,
      "learning_rate": 4.8747045437306965e-06,
      "loss": 0.4217,
      "step": 21663
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.488655686378479,
      "learning_rate": 4.873830092200251e-06,
      "loss": 0.3538,
      "step": 21664
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5343855619430542,
      "learning_rate": 4.8729556938363805e-06,
      "loss": 0.3934,
      "step": 21665
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.490256667137146,
      "learning_rate": 4.872081348648158e-06,
      "loss": 0.3976,
      "step": 21666
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5297731161117554,
      "learning_rate": 4.871207056644657e-06,
      "loss": 0.3997,
      "step": 21667
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.492308259010315,
      "learning_rate": 4.870332817834935e-06,
      "loss": 0.3907,
      "step": 21668
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5941485166549683,
      "learning_rate": 4.869458632228068e-06,
      "loss": 0.4372,
      "step": 21669
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5132704973220825,
      "learning_rate": 4.868584499833116e-06,
      "loss": 0.3887,
      "step": 21670
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6472352743148804,
      "learning_rate": 4.867710420659152e-06,
      "loss": 0.4559,
      "step": 21671
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5494341850280762,
      "learning_rate": 4.866836394715234e-06,
      "loss": 0.3807,
      "step": 21672
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7501212358474731,
      "learning_rate": 4.865962422010432e-06,
      "loss": 0.9753,
      "step": 21673
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5138241052627563,
      "learning_rate": 4.8650885025538116e-06,
      "loss": 0.9723,
      "step": 21674
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5943182706832886,
      "learning_rate": 4.864214636354435e-06,
      "loss": 0.4539,
      "step": 21675
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4926373958587646,
      "learning_rate": 4.863340823421362e-06,
      "loss": 0.3794,
      "step": 21676
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.566826581954956,
      "learning_rate": 4.862467063763659e-06,
      "loss": 0.439,
      "step": 21677
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5460833311080933,
      "learning_rate": 4.861593357390386e-06,
      "loss": 0.4219,
      "step": 21678
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.666877031326294,
      "learning_rate": 4.860719704310612e-06,
      "loss": 0.4508,
      "step": 21679
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5214394330978394,
      "learning_rate": 4.859846104533392e-06,
      "loss": 0.3954,
      "step": 21680
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4735100269317627,
      "learning_rate": 4.858972558067784e-06,
      "loss": 0.4151,
      "step": 21681
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4733597040176392,
      "learning_rate": 4.858099064922852e-06,
      "loss": 0.3776,
      "step": 21682
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.595122218132019,
      "learning_rate": 4.8572256251076544e-06,
      "loss": 0.4568,
      "step": 21683
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4757347106933594,
      "learning_rate": 4.856352238631254e-06,
      "loss": 0.388,
      "step": 21684
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.609487771987915,
      "learning_rate": 4.855478905502708e-06,
      "loss": 0.462,
      "step": 21685
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4868661165237427,
      "learning_rate": 4.854605625731066e-06,
      "loss": 0.3968,
      "step": 21686
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6194027662277222,
      "learning_rate": 4.853732399325393e-06,
      "loss": 0.4458,
      "step": 21687
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5591708421707153,
      "learning_rate": 4.852859226294747e-06,
      "loss": 0.3681,
      "step": 21688
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5892630815505981,
      "learning_rate": 4.851986106648179e-06,
      "loss": 0.4255,
      "step": 21689
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5234096050262451,
      "learning_rate": 4.85111304039475e-06,
      "loss": 0.371,
      "step": 21690
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6461700201034546,
      "learning_rate": 4.850240027543509e-06,
      "loss": 0.436,
      "step": 21691
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4401540756225586,
      "learning_rate": 4.849367068103518e-06,
      "loss": 0.3715,
      "step": 21692
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6404951810836792,
      "learning_rate": 4.848494162083822e-06,
      "loss": 0.4543,
      "step": 21693
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5620590448379517,
      "learning_rate": 4.84762130949348e-06,
      "loss": 0.3813,
      "step": 21694
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5471134185791016,
      "learning_rate": 4.846748510341548e-06,
      "loss": 0.4365,
      "step": 21695
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6178184747695923,
      "learning_rate": 4.845875764637073e-06,
      "loss": 0.3696,
      "step": 21696
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6086251735687256,
      "learning_rate": 4.845003072389106e-06,
      "loss": 0.4519,
      "step": 21697
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5779908895492554,
      "learning_rate": 4.8441304336067e-06,
      "loss": 0.3875,
      "step": 21698
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5406700372695923,
      "learning_rate": 4.8432578482989055e-06,
      "loss": 0.4014,
      "step": 21699
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5020287036895752,
      "learning_rate": 4.8423853164747815e-06,
      "loss": 0.3861,
      "step": 21700
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7487690448760986,
      "learning_rate": 4.841512838143361e-06,
      "loss": 0.4089,
      "step": 21701
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.476076602935791,
      "learning_rate": 4.840640413313701e-06,
      "loss": 0.356,
      "step": 21702
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6401101350784302,
      "learning_rate": 4.8397680419948514e-06,
      "loss": 0.4518,
      "step": 21703
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5096927881240845,
      "learning_rate": 4.838895724195861e-06,
      "loss": 0.3734,
      "step": 21704
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.540101170539856,
      "learning_rate": 4.838023459925775e-06,
      "loss": 0.4393,
      "step": 21705
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.580605149269104,
      "learning_rate": 4.837151249193638e-06,
      "loss": 0.4063,
      "step": 21706
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5830656290054321,
      "learning_rate": 4.836279092008498e-06,
      "loss": 0.4469,
      "step": 21707
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5278695821762085,
      "learning_rate": 4.8354069883794045e-06,
      "loss": 0.3821,
      "step": 21708
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.60358726978302,
      "learning_rate": 4.834534938315394e-06,
      "loss": 0.4199,
      "step": 21709
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5478761196136475,
      "learning_rate": 4.833662941825521e-06,
      "loss": 0.4021,
      "step": 21710
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5170514583587646,
      "learning_rate": 4.832790998918822e-06,
      "loss": 0.4369,
      "step": 21711
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.630663275718689,
      "learning_rate": 4.831919109604345e-06,
      "loss": 0.3936,
      "step": 21712
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5786222219467163,
      "learning_rate": 4.8310472738911285e-06,
      "loss": 0.4128,
      "step": 21713
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.62654447555542,
      "learning_rate": 4.830175491788217e-06,
      "loss": 0.4411,
      "step": 21714
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6211515665054321,
      "learning_rate": 4.829303763304656e-06,
      "loss": 0.4359,
      "step": 21715
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5435137748718262,
      "learning_rate": 4.828432088449484e-06,
      "loss": 0.3889,
      "step": 21716
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6623177528381348,
      "learning_rate": 4.827560467231736e-06,
      "loss": 0.44,
      "step": 21717
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6003284454345703,
      "learning_rate": 4.826688899660459e-06,
      "loss": 0.3735,
      "step": 21718
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5828441381454468,
      "learning_rate": 4.82581738574469e-06,
      "loss": 0.4284,
      "step": 21719
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.542015790939331,
      "learning_rate": 4.824945925493472e-06,
      "loss": 0.3864,
      "step": 21720
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5444554090499878,
      "learning_rate": 4.82407451891584e-06,
      "loss": 0.4618,
      "step": 21721
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.615911841392517,
      "learning_rate": 4.823203166020828e-06,
      "loss": 0.3869,
      "step": 21722
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5666972398757935,
      "learning_rate": 4.822331866817478e-06,
      "loss": 0.4856,
      "step": 21723
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4325275421142578,
      "learning_rate": 4.82146062131483e-06,
      "loss": 0.3589,
      "step": 21724
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6407268047332764,
      "learning_rate": 4.820589429521911e-06,
      "loss": 0.4211,
      "step": 21725
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.49131178855896,
      "learning_rate": 4.819718291447767e-06,
      "loss": 0.3987,
      "step": 21726
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6051206588745117,
      "learning_rate": 4.818847207101424e-06,
      "loss": 0.4516,
      "step": 21727
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5202399492263794,
      "learning_rate": 4.817976176491924e-06,
      "loss": 0.3913,
      "step": 21728
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6948151588439941,
      "learning_rate": 4.817105199628295e-06,
      "loss": 0.4414,
      "step": 21729
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6944524049758911,
      "learning_rate": 4.816234276519572e-06,
      "loss": 0.4121,
      "step": 21730
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.537902593612671,
      "learning_rate": 4.815363407174792e-06,
      "loss": 0.4208,
      "step": 21731
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6366647481918335,
      "learning_rate": 4.814492591602984e-06,
      "loss": 0.3866,
      "step": 21732
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6408238410949707,
      "learning_rate": 4.813621829813177e-06,
      "loss": 0.4334,
      "step": 21733
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4938992261886597,
      "learning_rate": 4.812751121814405e-06,
      "loss": 0.3771,
      "step": 21734
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.480949878692627,
      "learning_rate": 4.811880467615698e-06,
      "loss": 0.3851,
      "step": 21735
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6659910678863525,
      "learning_rate": 4.811009867226092e-06,
      "loss": 0.4324,
      "step": 21736
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.500006079673767,
      "learning_rate": 4.81013932065461e-06,
      "loss": 0.4163,
      "step": 21737
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5328820943832397,
      "learning_rate": 4.809268827910279e-06,
      "loss": 0.3837,
      "step": 21738
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.256953716278076,
      "learning_rate": 4.80839838900213e-06,
      "loss": 0.4035,
      "step": 21739
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4681452512741089,
      "learning_rate": 4.807528003939196e-06,
      "loss": 0.3705,
      "step": 21740
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7621630430221558,
      "learning_rate": 4.806657672730499e-06,
      "loss": 0.4279,
      "step": 21741
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4758951663970947,
      "learning_rate": 4.805787395385063e-06,
      "loss": 0.379,
      "step": 21742
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5668354034423828,
      "learning_rate": 4.804917171911918e-06,
      "loss": 0.4988,
      "step": 21743
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5658730268478394,
      "learning_rate": 4.804047002320094e-06,
      "loss": 0.4171,
      "step": 21744
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4729632139205933,
      "learning_rate": 4.803176886618605e-06,
      "loss": 0.4326,
      "step": 21745
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5197809934616089,
      "learning_rate": 4.802306824816488e-06,
      "loss": 0.4105,
      "step": 21746
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6244990825653076,
      "learning_rate": 4.801436816922757e-06,
      "loss": 0.475,
      "step": 21747
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6033191680908203,
      "learning_rate": 4.800566862946442e-06,
      "loss": 0.4126,
      "step": 21748
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5711928606033325,
      "learning_rate": 4.799696962896559e-06,
      "loss": 0.4186,
      "step": 21749
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4920271635055542,
      "learning_rate": 4.798827116782134e-06,
      "loss": 0.382,
      "step": 21750
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6796611547470093,
      "learning_rate": 4.797957324612193e-06,
      "loss": 0.9464,
      "step": 21751
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4889825582504272,
      "learning_rate": 4.797087586395753e-06,
      "loss": 1.0543,
      "step": 21752
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6344492435455322,
      "learning_rate": 4.796217902141831e-06,
      "loss": 0.4357,
      "step": 21753
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6729435920715332,
      "learning_rate": 4.795348271859451e-06,
      "loss": 0.3789,
      "step": 21754
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5587921142578125,
      "learning_rate": 4.794478695557631e-06,
      "loss": 0.4447,
      "step": 21755
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5968756675720215,
      "learning_rate": 4.793609173245396e-06,
      "loss": 0.4079,
      "step": 21756
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5554476976394653,
      "learning_rate": 4.792739704931759e-06,
      "loss": 0.418,
      "step": 21757
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6549192667007446,
      "learning_rate": 4.791870290625734e-06,
      "loss": 0.3838,
      "step": 21758
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6569346189498901,
      "learning_rate": 4.7910009303363425e-06,
      "loss": 0.4337,
      "step": 21759
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.505102276802063,
      "learning_rate": 4.790131624072604e-06,
      "loss": 0.3779,
      "step": 21760
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5966895818710327,
      "learning_rate": 4.789262371843528e-06,
      "loss": 0.4604,
      "step": 21761
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4732385873794556,
      "learning_rate": 4.788393173658138e-06,
      "loss": 0.3872,
      "step": 21762
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6561532020568848,
      "learning_rate": 4.787524029525441e-06,
      "loss": 1.0048,
      "step": 21763
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4577192068099976,
      "learning_rate": 4.786654939454458e-06,
      "loss": 1.0359,
      "step": 21764
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8414556980133057,
      "learning_rate": 4.785785903454196e-06,
      "loss": 0.4325,
      "step": 21765
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.529144525527954,
      "learning_rate": 4.7849169215336735e-06,
      "loss": 0.3754,
      "step": 21766
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5779324769973755,
      "learning_rate": 4.784047993701905e-06,
      "loss": 0.4004,
      "step": 21767
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.654444932937622,
      "learning_rate": 4.783179119967898e-06,
      "loss": 0.4289,
      "step": 21768
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5980457067489624,
      "learning_rate": 4.7823103003406636e-06,
      "loss": 0.3899,
      "step": 21769
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6665900945663452,
      "learning_rate": 4.7814415348292145e-06,
      "loss": 0.4188,
      "step": 21770
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5571691989898682,
      "learning_rate": 4.780572823442561e-06,
      "loss": 0.4629,
      "step": 21771
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4439839124679565,
      "learning_rate": 4.779704166189718e-06,
      "loss": 0.3747,
      "step": 21772
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4926514625549316,
      "learning_rate": 4.77883556307969e-06,
      "loss": 0.4074,
      "step": 21773
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5357236862182617,
      "learning_rate": 4.777967014121483e-06,
      "loss": 0.4087,
      "step": 21774
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5511765480041504,
      "learning_rate": 4.7770985193241085e-06,
      "loss": 0.4316,
      "step": 21775
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5426160097122192,
      "learning_rate": 4.776230078696578e-06,
      "loss": 0.3896,
      "step": 21776
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6074286699295044,
      "learning_rate": 4.775361692247894e-06,
      "loss": 0.4175,
      "step": 21777
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6290838718414307,
      "learning_rate": 4.774493359987061e-06,
      "loss": 0.3781,
      "step": 21778
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6051280498504639,
      "learning_rate": 4.773625081923088e-06,
      "loss": 0.4117,
      "step": 21779
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5125900506973267,
      "learning_rate": 4.772756858064983e-06,
      "loss": 0.3794,
      "step": 21780
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5630689859390259,
      "learning_rate": 4.771888688421745e-06,
      "loss": 0.4541,
      "step": 21781
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5788066387176514,
      "learning_rate": 4.771020573002385e-06,
      "loss": 0.4021,
      "step": 21782
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.535105586051941,
      "learning_rate": 4.7701525118159e-06,
      "loss": 0.428,
      "step": 21783
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5319628715515137,
      "learning_rate": 4.7692845048712995e-06,
      "loss": 0.4225,
      "step": 21784
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6398216485977173,
      "learning_rate": 4.7684165521775795e-06,
      "loss": 0.4485,
      "step": 21785
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.512134075164795,
      "learning_rate": 4.7675486537437465e-06,
      "loss": 0.3947,
      "step": 21786
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.524278998374939,
      "learning_rate": 4.7666808095788045e-06,
      "loss": 0.4368,
      "step": 21787
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6156251430511475,
      "learning_rate": 4.76581301969175e-06,
      "loss": 0.407,
      "step": 21788
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5716224908828735,
      "learning_rate": 4.7649452840915815e-06,
      "loss": 0.3973,
      "step": 21789
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5017977952957153,
      "learning_rate": 4.764077602787301e-06,
      "loss": 0.3732,
      "step": 21790
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6980557441711426,
      "learning_rate": 4.763209975787908e-06,
      "loss": 0.4996,
      "step": 21791
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5127191543579102,
      "learning_rate": 4.762342403102407e-06,
      "loss": 0.3902,
      "step": 21792
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6978815793991089,
      "learning_rate": 4.761474884739789e-06,
      "loss": 0.4593,
      "step": 21793
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4926838874816895,
      "learning_rate": 4.760607420709051e-06,
      "loss": 0.3857,
      "step": 21794
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5218781232833862,
      "learning_rate": 4.759740011019192e-06,
      "loss": 0.4443,
      "step": 21795
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5660552978515625,
      "learning_rate": 4.758872655679212e-06,
      "loss": 0.3751,
      "step": 21796
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5709236860275269,
      "learning_rate": 4.7580053546981e-06,
      "loss": 0.4442,
      "step": 21797
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5503487586975098,
      "learning_rate": 4.757138108084858e-06,
      "loss": 0.3858,
      "step": 21798
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5287882089614868,
      "learning_rate": 4.756270915848475e-06,
      "loss": 0.4199,
      "step": 21799
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5754629373550415,
      "learning_rate": 4.75540377799795e-06,
      "loss": 0.4038,
      "step": 21800
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5647847652435303,
      "learning_rate": 4.754536694542271e-06,
      "loss": 0.395,
      "step": 21801
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5414401292800903,
      "learning_rate": 4.753669665490436e-06,
      "loss": 0.3981,
      "step": 21802
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4992406368255615,
      "learning_rate": 4.752802690851438e-06,
      "loss": 0.4066,
      "step": 21803
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5488166809082031,
      "learning_rate": 4.751935770634267e-06,
      "loss": 0.397,
      "step": 21804
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5198994874954224,
      "learning_rate": 4.7510689048479105e-06,
      "loss": 0.4133,
      "step": 21805
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4462908506393433,
      "learning_rate": 4.750202093501363e-06,
      "loss": 0.3866,
      "step": 21806
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6937891244888306,
      "learning_rate": 4.749335336603614e-06,
      "loss": 0.4797,
      "step": 21807
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.56452476978302,
      "learning_rate": 4.7484686341636585e-06,
      "loss": 0.3942,
      "step": 21808
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5714845657348633,
      "learning_rate": 4.747601986190481e-06,
      "loss": 0.4372,
      "step": 21809
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5704602003097534,
      "learning_rate": 4.746735392693066e-06,
      "loss": 0.4155,
      "step": 21810
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6645872592926025,
      "learning_rate": 4.745868853680406e-06,
      "loss": 0.4172,
      "step": 21811
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4840868711471558,
      "learning_rate": 4.74500236916149e-06,
      "loss": 0.3771,
      "step": 21812
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5012543201446533,
      "learning_rate": 4.744135939145299e-06,
      "loss": 0.4296,
      "step": 21813
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.574755072593689,
      "learning_rate": 4.743269563640828e-06,
      "loss": 0.4479,
      "step": 21814
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6540035009384155,
      "learning_rate": 4.742403242657052e-06,
      "loss": 0.4107,
      "step": 21815
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4839773178100586,
      "learning_rate": 4.741536976202966e-06,
      "loss": 0.3674,
      "step": 21816
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5686919689178467,
      "learning_rate": 4.740670764287547e-06,
      "loss": 0.4179,
      "step": 21817
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5449423789978027,
      "learning_rate": 4.739804606919787e-06,
      "loss": 0.4132,
      "step": 21818
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5467336177825928,
      "learning_rate": 4.738938504108659e-06,
      "loss": 0.4369,
      "step": 21819
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8563017845153809,
      "learning_rate": 4.738072455863156e-06,
      "loss": 0.4126,
      "step": 21820
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5525435209274292,
      "learning_rate": 4.737206462192253e-06,
      "loss": 0.4085,
      "step": 21821
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8124017715454102,
      "learning_rate": 4.736340523104933e-06,
      "loss": 0.4116,
      "step": 21822
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.570939064025879,
      "learning_rate": 4.735474638610184e-06,
      "loss": 0.4074,
      "step": 21823
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4857513904571533,
      "learning_rate": 4.7346088087169805e-06,
      "loss": 0.3972,
      "step": 21824
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.555645227432251,
      "learning_rate": 4.733743033434301e-06,
      "loss": 0.4024,
      "step": 21825
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6151654720306396,
      "learning_rate": 4.732877312771126e-06,
      "loss": 0.4047,
      "step": 21826
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4814791679382324,
      "learning_rate": 4.7320116467364366e-06,
      "loss": 0.4261,
      "step": 21827
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4921371936798096,
      "learning_rate": 4.7311460353392135e-06,
      "loss": 0.3877,
      "step": 21828
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8122196197509766,
      "learning_rate": 4.730280478588431e-06,
      "loss": 0.9916,
      "step": 21829
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4333075284957886,
      "learning_rate": 4.729414976493064e-06,
      "loss": 0.9541,
      "step": 21830
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.4859611988067627,
      "learning_rate": 4.728549529062092e-06,
      "loss": 1.0044,
      "step": 21831
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5011159181594849,
      "learning_rate": 4.727684136304494e-06,
      "loss": 0.9604,
      "step": 21832
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5972802639007568,
      "learning_rate": 4.726818798229238e-06,
      "loss": 0.4374,
      "step": 21833
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6289165019989014,
      "learning_rate": 4.725953514845308e-06,
      "loss": 0.4215,
      "step": 21834
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5427439212799072,
      "learning_rate": 4.72508828616167e-06,
      "loss": 0.4516,
      "step": 21835
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6714341640472412,
      "learning_rate": 4.724223112187304e-06,
      "loss": 0.4448,
      "step": 21836
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5814930200576782,
      "learning_rate": 4.723357992931178e-06,
      "loss": 0.4321,
      "step": 21837
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6043649911880493,
      "learning_rate": 4.722492928402267e-06,
      "loss": 0.4064,
      "step": 21838
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6344102621078491,
      "learning_rate": 4.721627918609548e-06,
      "loss": 0.4722,
      "step": 21839
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.783496618270874,
      "learning_rate": 4.720762963561987e-06,
      "loss": 0.347,
      "step": 21840
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5365368127822876,
      "learning_rate": 4.719898063268553e-06,
      "loss": 0.4463,
      "step": 21841
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6081316471099854,
      "learning_rate": 4.7190332177382194e-06,
      "loss": 0.4059,
      "step": 21842
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5372146368026733,
      "learning_rate": 4.718168426979956e-06,
      "loss": 0.4186,
      "step": 21843
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5464853048324585,
      "learning_rate": 4.717303691002735e-06,
      "loss": 0.3787,
      "step": 21844
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5364468097686768,
      "learning_rate": 4.716439009815522e-06,
      "loss": 0.4292,
      "step": 21845
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5077462196350098,
      "learning_rate": 4.7155743834272825e-06,
      "loss": 0.38,
      "step": 21846
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.520904779434204,
      "learning_rate": 4.714709811846987e-06,
      "loss": 0.4406,
      "step": 21847
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5049972534179688,
      "learning_rate": 4.713845295083606e-06,
      "loss": 0.3835,
      "step": 21848
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5592825412750244,
      "learning_rate": 4.712980833146098e-06,
      "loss": 0.4244,
      "step": 21849
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.586450457572937,
      "learning_rate": 4.712116426043437e-06,
      "loss": 0.4201,
      "step": 21850
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5637283325195312,
      "learning_rate": 4.711252073784581e-06,
      "loss": 0.4085,
      "step": 21851
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5369303226470947,
      "learning_rate": 4.710387776378501e-06,
      "loss": 0.3681,
      "step": 21852
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.616113305091858,
      "learning_rate": 4.7095235338341565e-06,
      "loss": 0.4542,
      "step": 21853
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5249543190002441,
      "learning_rate": 4.708659346160511e-06,
      "loss": 0.392,
      "step": 21854
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5200663805007935,
      "learning_rate": 4.7077952133665336e-06,
      "loss": 0.4045,
      "step": 21855
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5097393989562988,
      "learning_rate": 4.706931135461182e-06,
      "loss": 0.4015,
      "step": 21856
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5141736268997192,
      "learning_rate": 4.706067112453417e-06,
      "loss": 0.4641,
      "step": 21857
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5145595073699951,
      "learning_rate": 4.7052031443521995e-06,
      "loss": 0.3896,
      "step": 21858
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5768852233886719,
      "learning_rate": 4.704339231166496e-06,
      "loss": 0.4674,
      "step": 21859
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4917898178100586,
      "learning_rate": 4.703475372905263e-06,
      "loss": 0.363,
      "step": 21860
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6259582042694092,
      "learning_rate": 4.702611569577457e-06,
      "loss": 0.4495,
      "step": 21861
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5116119384765625,
      "learning_rate": 4.701747821192038e-06,
      "loss": 0.3764,
      "step": 21862
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5270506143569946,
      "learning_rate": 4.700884127757969e-06,
      "loss": 0.4511,
      "step": 21863
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5059794187545776,
      "learning_rate": 4.7000204892842075e-06,
      "loss": 0.3876,
      "step": 21864
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8209978342056274,
      "learning_rate": 4.69915690577971e-06,
      "loss": 0.9756,
      "step": 21865
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5896223783493042,
      "learning_rate": 4.698293377253427e-06,
      "loss": 1.0852,
      "step": 21866
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5547845363616943,
      "learning_rate": 4.697429903714319e-06,
      "loss": 0.4185,
      "step": 21867
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5769444704055786,
      "learning_rate": 4.696566485171347e-06,
      "loss": 0.3909,
      "step": 21868
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5593634843826294,
      "learning_rate": 4.695703121633457e-06,
      "loss": 0.4232,
      "step": 21869
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4603238105773926,
      "learning_rate": 4.694839813109611e-06,
      "loss": 0.3798,
      "step": 21870
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5361454486846924,
      "learning_rate": 4.693976559608756e-06,
      "loss": 0.4117,
      "step": 21871
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5985779762268066,
      "learning_rate": 4.6931133611398524e-06,
      "loss": 0.4062,
      "step": 21872
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4954626560211182,
      "learning_rate": 4.692250217711847e-06,
      "loss": 0.3961,
      "step": 21873
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6712132692337036,
      "learning_rate": 4.691387129333692e-06,
      "loss": 0.4009,
      "step": 21874
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4872771501541138,
      "learning_rate": 4.6905240960143474e-06,
      "loss": 0.4123,
      "step": 21875
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4813199043273926,
      "learning_rate": 4.689661117762757e-06,
      "loss": 0.3857,
      "step": 21876
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5353721380233765,
      "learning_rate": 4.688798194587869e-06,
      "loss": 0.4146,
      "step": 21877
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5030401945114136,
      "learning_rate": 4.687935326498637e-06,
      "loss": 0.372,
      "step": 21878
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.571034550666809,
      "learning_rate": 4.687072513504009e-06,
      "loss": 0.4204,
      "step": 21879
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4698927402496338,
      "learning_rate": 4.686209755612939e-06,
      "loss": 0.3774,
      "step": 21880
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6123836040496826,
      "learning_rate": 4.685347052834372e-06,
      "loss": 0.4673,
      "step": 21881
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6095012426376343,
      "learning_rate": 4.684484405177251e-06,
      "loss": 0.4387,
      "step": 21882
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5874693393707275,
      "learning_rate": 4.683621812650525e-06,
      "loss": 0.4327,
      "step": 21883
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.467909812927246,
      "learning_rate": 4.682759275263147e-06,
      "loss": 0.3775,
      "step": 21884
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5676544904708862,
      "learning_rate": 4.681896793024055e-06,
      "loss": 0.4052,
      "step": 21885
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5550593137741089,
      "learning_rate": 4.681034365942199e-06,
      "loss": 0.405,
      "step": 21886
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.562188744544983,
      "learning_rate": 4.68017199402652e-06,
      "loss": 0.4407,
      "step": 21887
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5100020170211792,
      "learning_rate": 4.679309677285968e-06,
      "loss": 0.3871,
      "step": 21888
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5421277284622192,
      "learning_rate": 4.678447415729478e-06,
      "loss": 0.4287,
      "step": 21889
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6268683671951294,
      "learning_rate": 4.677585209365999e-06,
      "loss": 0.3941,
      "step": 21890
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5817643404006958,
      "learning_rate": 4.6767230582044755e-06,
      "loss": 0.4832,
      "step": 21891
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5569883584976196,
      "learning_rate": 4.6758609622538465e-06,
      "loss": 0.3804,
      "step": 21892
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6679521799087524,
      "learning_rate": 4.674998921523049e-06,
      "loss": 0.5093,
      "step": 21893
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5473989248275757,
      "learning_rate": 4.674136936021028e-06,
      "loss": 0.4186,
      "step": 21894
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5560877323150635,
      "learning_rate": 4.673275005756727e-06,
      "loss": 0.4144,
      "step": 21895
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5542813539505005,
      "learning_rate": 4.6724131307390815e-06,
      "loss": 0.3942,
      "step": 21896
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5729714632034302,
      "learning_rate": 4.671551310977028e-06,
      "loss": 0.4374,
      "step": 21897
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5241048336029053,
      "learning_rate": 4.670689546479509e-06,
      "loss": 0.3872,
      "step": 21898
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5743738412857056,
      "learning_rate": 4.669827837255461e-06,
      "loss": 0.473,
      "step": 21899
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.582345724105835,
      "learning_rate": 4.668966183313824e-06,
      "loss": 0.3867,
      "step": 21900
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.664443016052246,
      "learning_rate": 4.668104584663533e-06,
      "loss": 0.4402,
      "step": 21901
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.459146499633789,
      "learning_rate": 4.6672430413135205e-06,
      "loss": 0.3678,
      "step": 21902
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5490708351135254,
      "learning_rate": 4.666381553272725e-06,
      "loss": 0.4492,
      "step": 21903
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5447425842285156,
      "learning_rate": 4.665520120550086e-06,
      "loss": 0.4121,
      "step": 21904
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.581224799156189,
      "learning_rate": 4.664658743154529e-06,
      "loss": 0.4054,
      "step": 21905
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4588308334350586,
      "learning_rate": 4.663797421094996e-06,
      "loss": 0.3491,
      "step": 21906
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6028833389282227,
      "learning_rate": 4.662936154380413e-06,
      "loss": 0.4336,
      "step": 21907
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8679025173187256,
      "learning_rate": 4.662074943019721e-06,
      "loss": 0.4119,
      "step": 21908
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5290508270263672,
      "learning_rate": 4.661213787021843e-06,
      "loss": 0.4551,
      "step": 21909
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.538039207458496,
      "learning_rate": 4.660352686395716e-06,
      "loss": 0.4127,
      "step": 21910
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5903141498565674,
      "learning_rate": 4.6594916411502735e-06,
      "loss": 0.4922,
      "step": 21911
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4809738397598267,
      "learning_rate": 4.658630651294442e-06,
      "loss": 0.3689,
      "step": 21912
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5766828060150146,
      "learning_rate": 4.657769716837149e-06,
      "loss": 0.4477,
      "step": 21913
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.494154930114746,
      "learning_rate": 4.656908837787327e-06,
      "loss": 0.3829,
      "step": 21914
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5466241836547852,
      "learning_rate": 4.656048014153904e-06,
      "loss": 0.4191,
      "step": 21915
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4632863998413086,
      "learning_rate": 4.655187245945812e-06,
      "loss": 0.3865,
      "step": 21916
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6365994215011597,
      "learning_rate": 4.654326533171976e-06,
      "loss": 0.4343,
      "step": 21917
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5093634128570557,
      "learning_rate": 4.653465875841318e-06,
      "loss": 0.3682,
      "step": 21918
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4578404426574707,
      "learning_rate": 4.6526052739627666e-06,
      "loss": 0.3958,
      "step": 21919
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7555516958236694,
      "learning_rate": 4.6517447275452554e-06,
      "loss": 0.3791,
      "step": 21920
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5639749765396118,
      "learning_rate": 4.650884236597699e-06,
      "loss": 0.393,
      "step": 21921
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.464112639427185,
      "learning_rate": 4.65002380112903e-06,
      "loss": 0.3578,
      "step": 21922
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5684754848480225,
      "learning_rate": 4.649163421148166e-06,
      "loss": 0.45,
      "step": 21923
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5985478162765503,
      "learning_rate": 4.648303096664034e-06,
      "loss": 0.4082,
      "step": 21924
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4634860754013062,
      "learning_rate": 4.64744282768556e-06,
      "loss": 0.408,
      "step": 21925
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5478872060775757,
      "learning_rate": 4.6465826142216596e-06,
      "loss": 0.3496,
      "step": 21926
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5440654754638672,
      "learning_rate": 4.645722456281262e-06,
      "loss": 0.403,
      "step": 21927
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.490090012550354,
      "learning_rate": 4.6448623538732794e-06,
      "loss": 0.3655,
      "step": 21928
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5432801246643066,
      "learning_rate": 4.644002307006643e-06,
      "loss": 0.3956,
      "step": 21929
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5725953578948975,
      "learning_rate": 4.643142315690262e-06,
      "loss": 0.4222,
      "step": 21930
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6048022508621216,
      "learning_rate": 4.642282379933063e-06,
      "loss": 0.4521,
      "step": 21931
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4903895854949951,
      "learning_rate": 4.641422499743965e-06,
      "loss": 0.3827,
      "step": 21932
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5661530494689941,
      "learning_rate": 4.640562675131886e-06,
      "loss": 0.4336,
      "step": 21933
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4638190269470215,
      "learning_rate": 4.639702906105738e-06,
      "loss": 0.3782,
      "step": 21934
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5568406581878662,
      "learning_rate": 4.638843192674442e-06,
      "loss": 0.4678,
      "step": 21935
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5328119993209839,
      "learning_rate": 4.6379835348469195e-06,
      "loss": 0.4118,
      "step": 21936
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5598773956298828,
      "learning_rate": 4.6371239326320814e-06,
      "loss": 0.4164,
      "step": 21937
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.583181619644165,
      "learning_rate": 4.63626438603884e-06,
      "loss": 0.4423,
      "step": 21938
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4817248582839966,
      "learning_rate": 4.635404895076115e-06,
      "loss": 0.4108,
      "step": 21939
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4892722368240356,
      "learning_rate": 4.634545459752818e-06,
      "loss": 0.3848,
      "step": 21940
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.036215305328369,
      "learning_rate": 4.633686080077869e-06,
      "loss": 0.419,
      "step": 21941
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5400360822677612,
      "learning_rate": 4.632826756060177e-06,
      "loss": 0.4303,
      "step": 21942
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.642685055732727,
      "learning_rate": 4.63196748770865e-06,
      "loss": 0.5134,
      "step": 21943
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4893018007278442,
      "learning_rate": 4.631108275032203e-06,
      "loss": 0.3668,
      "step": 21944
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6537084579467773,
      "learning_rate": 4.630249118039754e-06,
      "loss": 0.4558,
      "step": 21945
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4969216585159302,
      "learning_rate": 4.629390016740203e-06,
      "loss": 0.3712,
      "step": 21946
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4716073274612427,
      "learning_rate": 4.6285309711424706e-06,
      "loss": 0.4164,
      "step": 21947
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5347249507904053,
      "learning_rate": 4.627671981255457e-06,
      "loss": 0.3832,
      "step": 21948
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7476850748062134,
      "learning_rate": 4.626813047088079e-06,
      "loss": 0.4485,
      "step": 21949
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4501316547393799,
      "learning_rate": 4.625954168649239e-06,
      "loss": 0.3639,
      "step": 21950
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5190805196762085,
      "learning_rate": 4.625095345947848e-06,
      "loss": 0.4372,
      "step": 21951
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5113816261291504,
      "learning_rate": 4.624236578992817e-06,
      "loss": 0.3788,
      "step": 21952
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6001235246658325,
      "learning_rate": 4.623377867793047e-06,
      "loss": 0.4218,
      "step": 21953
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5199652910232544,
      "learning_rate": 4.6225192123574445e-06,
      "loss": 0.3928,
      "step": 21954
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5612558126449585,
      "learning_rate": 4.621660612694916e-06,
      "loss": 0.4243,
      "step": 21955
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5726031064987183,
      "learning_rate": 4.620802068814367e-06,
      "loss": 0.3837,
      "step": 21956
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8892982006072998,
      "learning_rate": 4.6199435807247075e-06,
      "loss": 0.9072,
      "step": 21957
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4686410427093506,
      "learning_rate": 4.619085148434836e-06,
      "loss": 1.0275,
      "step": 21958
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.516053557395935,
      "learning_rate": 4.61822677195365e-06,
      "loss": 0.4,
      "step": 21959
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.42281174659729,
      "learning_rate": 4.6173684512900606e-06,
      "loss": 0.3874,
      "step": 21960
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8109856843948364,
      "learning_rate": 4.6165101864529704e-06,
      "loss": 0.4633,
      "step": 21961
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5184651613235474,
      "learning_rate": 4.615651977451274e-06,
      "loss": 0.4025,
      "step": 21962
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6554843187332153,
      "learning_rate": 4.614793824293881e-06,
      "loss": 0.4362,
      "step": 21963
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.678777813911438,
      "learning_rate": 4.6139357269896845e-06,
      "loss": 0.4491,
      "step": 21964
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8282331228256226,
      "learning_rate": 4.61307768554759e-06,
      "loss": 0.4227,
      "step": 21965
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5706008672714233,
      "learning_rate": 4.61221969997649e-06,
      "loss": 0.3926,
      "step": 21966
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4688066244125366,
      "learning_rate": 4.611361770285288e-06,
      "loss": 0.4287,
      "step": 21967
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4921879768371582,
      "learning_rate": 4.610503896482885e-06,
      "loss": 0.3845,
      "step": 21968
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6036667823791504,
      "learning_rate": 4.609646078578174e-06,
      "loss": 0.4191,
      "step": 21969
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5621365308761597,
      "learning_rate": 4.608788316580049e-06,
      "loss": 0.3878,
      "step": 21970
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5534411668777466,
      "learning_rate": 4.607930610497412e-06,
      "loss": 0.4153,
      "step": 21971
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7021673917770386,
      "learning_rate": 4.607072960339155e-06,
      "loss": 0.3745,
      "step": 21972
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4896444082260132,
      "learning_rate": 4.606215366114184e-06,
      "loss": 0.4082,
      "step": 21973
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8701108694076538,
      "learning_rate": 4.605357827831376e-06,
      "loss": 0.3803,
      "step": 21974
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.606757640838623,
      "learning_rate": 4.604500345499634e-06,
      "loss": 0.4216,
      "step": 21975
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5679084062576294,
      "learning_rate": 4.603642919127852e-06,
      "loss": 0.3628,
      "step": 21976
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5595982074737549,
      "learning_rate": 4.602785548724925e-06,
      "loss": 0.4132,
      "step": 21977
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5714452266693115,
      "learning_rate": 4.601928234299743e-06,
      "loss": 0.3628,
      "step": 21978
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5271161794662476,
      "learning_rate": 4.601070975861194e-06,
      "loss": 0.4534,
      "step": 21979
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4949780702590942,
      "learning_rate": 4.600213773418172e-06,
      "loss": 0.3961,
      "step": 21980
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6277376413345337,
      "learning_rate": 4.599356626979571e-06,
      "loss": 0.4531,
      "step": 21981
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5786378383636475,
      "learning_rate": 4.598499536554274e-06,
      "loss": 0.3851,
      "step": 21982
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8582656383514404,
      "learning_rate": 4.5976425021511785e-06,
      "loss": 0.491,
      "step": 21983
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5321296453475952,
      "learning_rate": 4.596785523779165e-06,
      "loss": 0.3789,
      "step": 21984
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5405274629592896,
      "learning_rate": 4.5959286014471294e-06,
      "loss": 0.4481,
      "step": 21985
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6071127653121948,
      "learning_rate": 4.595071735163953e-06,
      "loss": 0.4154,
      "step": 21986
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4963716268539429,
      "learning_rate": 4.594214924938525e-06,
      "loss": 0.4457,
      "step": 21987
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6357553005218506,
      "learning_rate": 4.593358170779736e-06,
      "loss": 0.3769,
      "step": 21988
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6547132730484009,
      "learning_rate": 4.592501472696468e-06,
      "loss": 0.4386,
      "step": 21989
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.619899868965149,
      "learning_rate": 4.591644830697603e-06,
      "loss": 0.4033,
      "step": 21990
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6124476194381714,
      "learning_rate": 4.59078824479203e-06,
      "loss": 0.4104,
      "step": 21991
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5261586904525757,
      "learning_rate": 4.5899317149886335e-06,
      "loss": 0.3823,
      "step": 21992
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.572341799736023,
      "learning_rate": 4.589075241296299e-06,
      "loss": 0.4343,
      "step": 21993
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5470603704452515,
      "learning_rate": 4.588218823723908e-06,
      "loss": 0.3865,
      "step": 21994
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.54662024974823,
      "learning_rate": 4.587362462280337e-06,
      "loss": 0.4278,
      "step": 21995
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5241687297821045,
      "learning_rate": 4.5865061569744715e-06,
      "loss": 0.3924,
      "step": 21996
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.539499044418335,
      "learning_rate": 4.585649907815199e-06,
      "loss": 0.4811,
      "step": 21997
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5620009899139404,
      "learning_rate": 4.584793714811391e-06,
      "loss": 0.3822,
      "step": 21998
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5897856950759888,
      "learning_rate": 4.583937577971935e-06,
      "loss": 0.475,
      "step": 21999
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5436925888061523,
      "learning_rate": 4.583081497305702e-06,
      "loss": 0.4104,
      "step": 22000
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5635546445846558,
      "learning_rate": 4.582225472821582e-06,
      "loss": 0.4536,
      "step": 22001
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5849941968917847,
      "learning_rate": 4.581369504528442e-06,
      "loss": 0.3998,
      "step": 22002
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5434143543243408,
      "learning_rate": 4.580513592435166e-06,
      "loss": 0.4546,
      "step": 22003
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5447275638580322,
      "learning_rate": 4.579657736550634e-06,
      "loss": 0.402,
      "step": 22004
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.580222487449646,
      "learning_rate": 4.5788019368837174e-06,
      "loss": 0.4598,
      "step": 22005
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4791806936264038,
      "learning_rate": 4.577946193443292e-06,
      "loss": 0.3737,
      "step": 22006
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6115292310714722,
      "learning_rate": 4.577090506238233e-06,
      "loss": 0.4311,
      "step": 22007
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5562132596969604,
      "learning_rate": 4.5762348752774184e-06,
      "loss": 0.4029,
      "step": 22008
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5661741495132446,
      "learning_rate": 4.575379300569729e-06,
      "loss": 0.4443,
      "step": 22009
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.456040859222412,
      "learning_rate": 4.574523782124022e-06,
      "loss": 0.3786,
      "step": 22010
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6133607625961304,
      "learning_rate": 4.573668319949179e-06,
      "loss": 0.4871,
      "step": 22011
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5200750827789307,
      "learning_rate": 4.5728129140540734e-06,
      "loss": 0.3824,
      "step": 22012
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5861876010894775,
      "learning_rate": 4.57195756444758e-06,
      "loss": 0.4473,
      "step": 22013
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5000017881393433,
      "learning_rate": 4.571102271138567e-06,
      "loss": 0.403,
      "step": 22014
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.1568620204925537,
      "learning_rate": 4.5702470341359e-06,
      "loss": 0.4468,
      "step": 22015
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7344882488250732,
      "learning_rate": 4.569391853448454e-06,
      "loss": 0.431,
      "step": 22016
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5897767543792725,
      "learning_rate": 4.568536729085102e-06,
      "loss": 0.4777,
      "step": 22017
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4788036346435547,
      "learning_rate": 4.567681661054707e-06,
      "loss": 0.3938,
      "step": 22018
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.569990873336792,
      "learning_rate": 4.566826649366143e-06,
      "loss": 0.424,
      "step": 22019
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5242247581481934,
      "learning_rate": 4.565971694028271e-06,
      "loss": 0.4331,
      "step": 22020
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.602052927017212,
      "learning_rate": 4.565116795049965e-06,
      "loss": 0.4145,
      "step": 22021
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4808131456375122,
      "learning_rate": 4.564261952440086e-06,
      "loss": 0.3706,
      "step": 22022
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7514774799346924,
      "learning_rate": 4.563407166207502e-06,
      "loss": 0.4281,
      "step": 22023
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4562586545944214,
      "learning_rate": 4.5625524363610825e-06,
      "loss": 0.3817,
      "step": 22024
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5620042085647583,
      "learning_rate": 4.5616977629096905e-06,
      "loss": 0.3992,
      "step": 22025
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6053837537765503,
      "learning_rate": 4.560843145862185e-06,
      "loss": 0.4094,
      "step": 22026
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5126136541366577,
      "learning_rate": 4.5599885852274325e-06,
      "loss": 0.4166,
      "step": 22027
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5308510065078735,
      "learning_rate": 4.559134081014299e-06,
      "loss": 0.3834,
      "step": 22028
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5822181701660156,
      "learning_rate": 4.5582796332316484e-06,
      "loss": 0.4237,
      "step": 22029
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4337944984436035,
      "learning_rate": 4.55742524188834e-06,
      "loss": 0.3619,
      "step": 22030
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6147265434265137,
      "learning_rate": 4.556570906993231e-06,
      "loss": 0.4736,
      "step": 22031
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.544419288635254,
      "learning_rate": 4.555716628555186e-06,
      "loss": 0.3895,
      "step": 22032
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.548280119895935,
      "learning_rate": 4.554862406583069e-06,
      "loss": 0.4059,
      "step": 22033
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4386180639266968,
      "learning_rate": 4.554008241085733e-06,
      "loss": 0.3554,
      "step": 22034
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5553843975067139,
      "learning_rate": 4.553154132072044e-06,
      "loss": 0.4654,
      "step": 22035
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5551563501358032,
      "learning_rate": 4.552300079550852e-06,
      "loss": 0.367,
      "step": 22036
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5526750087738037,
      "learning_rate": 4.551446083531023e-06,
      "loss": 0.4608,
      "step": 22037
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.569713830947876,
      "learning_rate": 4.550592144021408e-06,
      "loss": 0.3835,
      "step": 22038
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.532366394996643,
      "learning_rate": 4.549738261030866e-06,
      "loss": 0.4148,
      "step": 22039
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5853556394577026,
      "learning_rate": 4.548884434568258e-06,
      "loss": 0.3899,
      "step": 22040
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6013782024383545,
      "learning_rate": 4.5480306646424344e-06,
      "loss": 0.4724,
      "step": 22041
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.522644281387329,
      "learning_rate": 4.5471769512622475e-06,
      "loss": 0.3861,
      "step": 22042
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7792555093765259,
      "learning_rate": 4.546323294436556e-06,
      "loss": 0.4446,
      "step": 22043
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5683003664016724,
      "learning_rate": 4.545469694174211e-06,
      "loss": 0.3866,
      "step": 22044
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5479739904403687,
      "learning_rate": 4.5446161504840726e-06,
      "loss": 0.448,
      "step": 22045
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5334861278533936,
      "learning_rate": 4.543762663374988e-06,
      "loss": 0.3927,
      "step": 22046
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5599604845046997,
      "learning_rate": 4.5429092328558054e-06,
      "loss": 0.4536,
      "step": 22047
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4761079549789429,
      "learning_rate": 4.542055858935382e-06,
      "loss": 0.3877,
      "step": 22048
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5272915363311768,
      "learning_rate": 4.541202541622569e-06,
      "loss": 0.4246,
      "step": 22049
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4894788265228271,
      "learning_rate": 4.540349280926215e-06,
      "loss": 0.3806,
      "step": 22050
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5178301334381104,
      "learning_rate": 4.539496076855165e-06,
      "loss": 0.3813,
      "step": 22051
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.508432388305664,
      "learning_rate": 4.5386429294182744e-06,
      "loss": 0.3786,
      "step": 22052
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5026835203170776,
      "learning_rate": 4.537789838624393e-06,
      "loss": 0.4319,
      "step": 22053
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.478949785232544,
      "learning_rate": 4.5369368044823605e-06,
      "loss": 0.4002,
      "step": 22054
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5394619703292847,
      "learning_rate": 4.536083827001033e-06,
      "loss": 0.4463,
      "step": 22055
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.467109203338623,
      "learning_rate": 4.535230906189251e-06,
      "loss": 0.3674,
      "step": 22056
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5544440746307373,
      "learning_rate": 4.534378042055866e-06,
      "loss": 0.4111,
      "step": 22057
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5158652067184448,
      "learning_rate": 4.533525234609717e-06,
      "loss": 0.3947,
      "step": 22058
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5137689113616943,
      "learning_rate": 4.532672483859654e-06,
      "loss": 0.4151,
      "step": 22059
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.506102442741394,
      "learning_rate": 4.531819789814522e-06,
      "loss": 0.4099,
      "step": 22060
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4835902452468872,
      "learning_rate": 4.5309671524831634e-06,
      "loss": 0.4424,
      "step": 22061
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5092885494232178,
      "learning_rate": 4.530114571874416e-06,
      "loss": 0.3837,
      "step": 22062
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5991730690002441,
      "learning_rate": 4.52926204799713e-06,
      "loss": 0.4538,
      "step": 22063
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5057072639465332,
      "learning_rate": 4.528409580860143e-06,
      "loss": 0.4017,
      "step": 22064
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.567732810974121,
      "learning_rate": 4.527557170472302e-06,
      "loss": 0.4218,
      "step": 22065
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5271944999694824,
      "learning_rate": 4.5267048168424445e-06,
      "loss": 0.3766,
      "step": 22066
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5343838930130005,
      "learning_rate": 4.525852519979406e-06,
      "loss": 0.3904,
      "step": 22067
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5461459159851074,
      "learning_rate": 4.5250002798920304e-06,
      "loss": 0.399,
      "step": 22068
    },
    {
      "epoch": 0.69,
      "grad_norm": 3.603268623352051,
      "learning_rate": 4.52414809658916e-06,
      "loss": 0.4747,
      "step": 22069
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5655134916305542,
      "learning_rate": 4.523295970079628e-06,
      "loss": 0.3727,
      "step": 22070
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6055526733398438,
      "learning_rate": 4.5224439003722774e-06,
      "loss": 0.443,
      "step": 22071
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5359848737716675,
      "learning_rate": 4.521591887475939e-06,
      "loss": 0.3736,
      "step": 22072
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5468416213989258,
      "learning_rate": 4.520739931399457e-06,
      "loss": 0.4229,
      "step": 22073
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6120758056640625,
      "learning_rate": 4.519888032151659e-06,
      "loss": 0.3675,
      "step": 22074
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6330960988998413,
      "learning_rate": 4.519036189741386e-06,
      "loss": 0.4359,
      "step": 22075
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4920127391815186,
      "learning_rate": 4.518184404177477e-06,
      "loss": 0.3826,
      "step": 22076
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5209671258926392,
      "learning_rate": 4.51733267546876e-06,
      "loss": 0.4383,
      "step": 22077
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4810154438018799,
      "learning_rate": 4.516481003624067e-06,
      "loss": 0.388,
      "step": 22078
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5419574975967407,
      "learning_rate": 4.515629388652235e-06,
      "loss": 0.4215,
      "step": 22079
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5318481922149658,
      "learning_rate": 4.514777830562096e-06,
      "loss": 0.4192,
      "step": 22080
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5922859907150269,
      "learning_rate": 4.513926329362485e-06,
      "loss": 0.4255,
      "step": 22081
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5595382452011108,
      "learning_rate": 4.513074885062233e-06,
      "loss": 0.425,
      "step": 22082
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6173540353775024,
      "learning_rate": 4.512223497670163e-06,
      "loss": 0.4741,
      "step": 22083
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5354195833206177,
      "learning_rate": 4.511372167195111e-06,
      "loss": 0.3899,
      "step": 22084
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5340490341186523,
      "learning_rate": 4.51052089364591e-06,
      "loss": 0.4513,
      "step": 22085
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5284043550491333,
      "learning_rate": 4.509669677031386e-06,
      "loss": 0.3814,
      "step": 22086
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4889825582504272,
      "learning_rate": 4.508818517360363e-06,
      "loss": 0.4282,
      "step": 22087
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5788533687591553,
      "learning_rate": 4.507967414641672e-06,
      "loss": 0.4148,
      "step": 22088
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.673396348953247,
      "learning_rate": 4.507116368884145e-06,
      "loss": 0.4474,
      "step": 22089
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5358906984329224,
      "learning_rate": 4.506265380096603e-06,
      "loss": 0.4144,
      "step": 22090
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6285532712936401,
      "learning_rate": 4.505414448287876e-06,
      "loss": 0.4085,
      "step": 22091
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4864366054534912,
      "learning_rate": 4.504563573466784e-06,
      "loss": 0.3587,
      "step": 22092
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5664986371994019,
      "learning_rate": 4.503712755642159e-06,
      "loss": 0.4223,
      "step": 22093
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.529828429222107,
      "learning_rate": 4.502861994822818e-06,
      "loss": 0.4021,
      "step": 22094
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5802886486053467,
      "learning_rate": 4.5020112910175885e-06,
      "loss": 0.486,
      "step": 22095
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5998904705047607,
      "learning_rate": 4.5011606442352974e-06,
      "loss": 0.4219,
      "step": 22096
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6221508979797363,
      "learning_rate": 4.500310054484763e-06,
      "loss": 0.4264,
      "step": 22097
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6433863639831543,
      "learning_rate": 4.499459521774805e-06,
      "loss": 0.4218,
      "step": 22098
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6575474739074707,
      "learning_rate": 4.498609046114247e-06,
      "loss": 0.9709,
      "step": 22099
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.486441731452942,
      "learning_rate": 4.4977586275119115e-06,
      "loss": 1.0055,
      "step": 22100
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.539040207862854,
      "learning_rate": 4.4969082659766196e-06,
      "loss": 0.4048,
      "step": 22101
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.47288179397583,
      "learning_rate": 4.496057961517189e-06,
      "loss": 0.3849,
      "step": 22102
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5851331949234009,
      "learning_rate": 4.495207714142437e-06,
      "loss": 0.4332,
      "step": 22103
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5168542861938477,
      "learning_rate": 4.494357523861182e-06,
      "loss": 0.4193,
      "step": 22104
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5376176834106445,
      "learning_rate": 4.493507390682247e-06,
      "loss": 0.408,
      "step": 22105
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5468283891677856,
      "learning_rate": 4.492657314614443e-06,
      "loss": 0.3998,
      "step": 22106
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5184285640716553,
      "learning_rate": 4.491807295666592e-06,
      "loss": 0.4166,
      "step": 22107
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6233530044555664,
      "learning_rate": 4.490957333847503e-06,
      "loss": 0.384,
      "step": 22108
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5567032098770142,
      "learning_rate": 4.490107429166001e-06,
      "loss": 0.414,
      "step": 22109
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5054600238800049,
      "learning_rate": 4.489257581630891e-06,
      "loss": 0.3743,
      "step": 22110
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5804786682128906,
      "learning_rate": 4.488407791250993e-06,
      "loss": 0.4158,
      "step": 22111
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6543409824371338,
      "learning_rate": 4.487558058035123e-06,
      "loss": 0.3842,
      "step": 22112
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5419971942901611,
      "learning_rate": 4.486708381992091e-06,
      "loss": 0.3963,
      "step": 22113
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5631476640701294,
      "learning_rate": 4.485858763130705e-06,
      "loss": 0.3848,
      "step": 22114
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5663162469863892,
      "learning_rate": 4.485009201459781e-06,
      "loss": 0.3994,
      "step": 22115
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5334664583206177,
      "learning_rate": 4.484159696988132e-06,
      "loss": 0.3882,
      "step": 22116
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6355825662612915,
      "learning_rate": 4.48331024972457e-06,
      "loss": 0.4285,
      "step": 22117
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.513288974761963,
      "learning_rate": 4.482460859677902e-06,
      "loss": 0.3748,
      "step": 22118
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5856616497039795,
      "learning_rate": 4.481611526856934e-06,
      "loss": 0.4592,
      "step": 22119
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4899770021438599,
      "learning_rate": 4.4807622512704796e-06,
      "loss": 0.368,
      "step": 22120
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.813976764678955,
      "learning_rate": 4.4799130329273485e-06,
      "loss": 0.5028,
      "step": 22121
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5209509134292603,
      "learning_rate": 4.479063871836344e-06,
      "loss": 0.383,
      "step": 22122
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5150216817855835,
      "learning_rate": 4.478214768006279e-06,
      "loss": 0.4166,
      "step": 22123
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6121134757995605,
      "learning_rate": 4.477365721445953e-06,
      "loss": 0.4188,
      "step": 22124
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5705077648162842,
      "learning_rate": 4.47651673216418e-06,
      "loss": 0.4416,
      "step": 22125
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4901041984558105,
      "learning_rate": 4.475667800169756e-06,
      "loss": 0.3834,
      "step": 22126
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5137983560562134,
      "learning_rate": 4.474818925471496e-06,
      "loss": 0.432,
      "step": 22127
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5378438234329224,
      "learning_rate": 4.473970108078194e-06,
      "loss": 0.4022,
      "step": 22128
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.514074444770813,
      "learning_rate": 4.473121347998662e-06,
      "loss": 0.4218,
      "step": 22129
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5228891372680664,
      "learning_rate": 4.472272645241698e-06,
      "loss": 0.3958,
      "step": 22130
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6850688457489014,
      "learning_rate": 4.471423999816103e-06,
      "loss": 0.4894,
      "step": 22131
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4920068979263306,
      "learning_rate": 4.470575411730688e-06,
      "loss": 0.3736,
      "step": 22132
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5628021955490112,
      "learning_rate": 4.469726880994246e-06,
      "loss": 0.4508,
      "step": 22133
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5296781063079834,
      "learning_rate": 4.468878407615577e-06,
      "loss": 0.3892,
      "step": 22134
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7769012451171875,
      "learning_rate": 4.468029991603483e-06,
      "loss": 0.4291,
      "step": 22135
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.429208755493164,
      "learning_rate": 4.467181632966765e-06,
      "loss": 0.3699,
      "step": 22136
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5738329887390137,
      "learning_rate": 4.466333331714223e-06,
      "loss": 0.4792,
      "step": 22137
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6072722673416138,
      "learning_rate": 4.465485087854655e-06,
      "loss": 0.3758,
      "step": 22138
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5934042930603027,
      "learning_rate": 4.464636901396851e-06,
      "loss": 0.4592,
      "step": 22139
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5080147981643677,
      "learning_rate": 4.463788772349616e-06,
      "loss": 0.392,
      "step": 22140
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5611857175827026,
      "learning_rate": 4.462940700721747e-06,
      "loss": 0.4624,
      "step": 22141
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7539600133895874,
      "learning_rate": 4.462092686522034e-06,
      "loss": 0.4025,
      "step": 22142
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4435049295425415,
      "learning_rate": 4.4612447297592775e-06,
      "loss": 0.3927,
      "step": 22143
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7319248914718628,
      "learning_rate": 4.460396830442268e-06,
      "loss": 0.4042,
      "step": 22144
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6715868711471558,
      "learning_rate": 4.459548988579805e-06,
      "loss": 0.4593,
      "step": 22145
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6060220003128052,
      "learning_rate": 4.4587012041806746e-06,
      "loss": 0.3606,
      "step": 22146
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6101988554000854,
      "learning_rate": 4.457853477253674e-06,
      "loss": 0.4553,
      "step": 22147
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5167796611785889,
      "learning_rate": 4.457005807807599e-06,
      "loss": 0.3869,
      "step": 22148
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5416929721832275,
      "learning_rate": 4.456158195851237e-06,
      "loss": 0.4218,
      "step": 22149
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4635525941848755,
      "learning_rate": 4.455310641393376e-06,
      "loss": 0.3885,
      "step": 22150
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5412083864212036,
      "learning_rate": 4.45446314444281e-06,
      "loss": 0.4113,
      "step": 22151
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5433872938156128,
      "learning_rate": 4.453615705008329e-06,
      "loss": 0.3907,
      "step": 22152
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5009031295776367,
      "learning_rate": 4.4527683230987254e-06,
      "loss": 0.4094,
      "step": 22153
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5426456928253174,
      "learning_rate": 4.451920998722785e-06,
      "loss": 0.3732,
      "step": 22154
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6081302165985107,
      "learning_rate": 4.451073731889292e-06,
      "loss": 0.437,
      "step": 22155
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6114110946655273,
      "learning_rate": 4.4502265226070385e-06,
      "loss": 0.4216,
      "step": 22156
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5452947616577148,
      "learning_rate": 4.4493793708848145e-06,
      "loss": 0.4369,
      "step": 22157
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5689915418624878,
      "learning_rate": 4.4485322767313965e-06,
      "loss": 0.3827,
      "step": 22158
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5466607809066772,
      "learning_rate": 4.447685240155582e-06,
      "loss": 0.4127,
      "step": 22159
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5250627994537354,
      "learning_rate": 4.446838261166146e-06,
      "loss": 0.375,
      "step": 22160
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5586979389190674,
      "learning_rate": 4.445991339771879e-06,
      "loss": 0.4688,
      "step": 22161
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.513597011566162,
      "learning_rate": 4.445144475981561e-06,
      "loss": 0.3845,
      "step": 22162
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5909550189971924,
      "learning_rate": 4.444297669803981e-06,
      "loss": 0.4764,
      "step": 22163
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5307210683822632,
      "learning_rate": 4.443450921247914e-06,
      "loss": 0.3626,
      "step": 22164
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.568084478378296,
      "learning_rate": 4.442604230322146e-06,
      "loss": 0.4779,
      "step": 22165
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.4663538932800293,
      "learning_rate": 4.441757597035464e-06,
      "loss": 0.4292,
      "step": 22166
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.614314317703247,
      "learning_rate": 4.440911021396639e-06,
      "loss": 0.512,
      "step": 22167
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5290489196777344,
      "learning_rate": 4.4400645034144606e-06,
      "loss": 0.3878,
      "step": 22168
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5532795190811157,
      "learning_rate": 4.4392180430976994e-06,
      "loss": 0.4193,
      "step": 22169
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.56777822971344,
      "learning_rate": 4.438371640455143e-06,
      "loss": 0.4283,
      "step": 22170
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6176596879959106,
      "learning_rate": 4.437525295495564e-06,
      "loss": 0.4465,
      "step": 22171
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6205644607543945,
      "learning_rate": 4.436679008227741e-06,
      "loss": 0.4272,
      "step": 22172
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.801777720451355,
      "learning_rate": 4.435832778660457e-06,
      "loss": 0.4299,
      "step": 22173
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5709658861160278,
      "learning_rate": 4.434986606802485e-06,
      "loss": 0.3996,
      "step": 22174
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5752443075180054,
      "learning_rate": 4.434140492662596e-06,
      "loss": 0.4203,
      "step": 22175
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4703069925308228,
      "learning_rate": 4.4332944362495724e-06,
      "loss": 0.3665,
      "step": 22176
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6074609756469727,
      "learning_rate": 4.4324484375721855e-06,
      "loss": 0.4331,
      "step": 22177
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4929078817367554,
      "learning_rate": 4.431602496639216e-06,
      "loss": 0.3786,
      "step": 22178
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5002923011779785,
      "learning_rate": 4.4307566134594316e-06,
      "loss": 0.4151,
      "step": 22179
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5439079999923706,
      "learning_rate": 4.429910788041604e-06,
      "loss": 0.4153,
      "step": 22180
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.570696234703064,
      "learning_rate": 4.429065020394508e-06,
      "loss": 0.5253,
      "step": 22181
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.569397211074829,
      "learning_rate": 4.42821931052692e-06,
      "loss": 0.3939,
      "step": 22182
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.505685806274414,
      "learning_rate": 4.4273736584476045e-06,
      "loss": 0.4221,
      "step": 22183
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5441763401031494,
      "learning_rate": 4.4265280641653375e-06,
      "loss": 0.3877,
      "step": 22184
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5509390830993652,
      "learning_rate": 4.425682527688885e-06,
      "loss": 0.5012,
      "step": 22185
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5433990955352783,
      "learning_rate": 4.424837049027021e-06,
      "loss": 0.382,
      "step": 22186
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.546095609664917,
      "learning_rate": 4.423991628188509e-06,
      "loss": 0.4397,
      "step": 22187
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5485413074493408,
      "learning_rate": 4.4231462651821185e-06,
      "loss": 0.4191,
      "step": 22188
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6047947406768799,
      "learning_rate": 4.422300960016625e-06,
      "loss": 0.4285,
      "step": 22189
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5607613325119019,
      "learning_rate": 4.421455712700789e-06,
      "loss": 0.3884,
      "step": 22190
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5872136354446411,
      "learning_rate": 4.420610523243373e-06,
      "loss": 0.4186,
      "step": 22191
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5371671915054321,
      "learning_rate": 4.419765391653149e-06,
      "loss": 0.3873,
      "step": 22192
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5622855424880981,
      "learning_rate": 4.418920317938881e-06,
      "loss": 0.4373,
      "step": 22193
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.519549012184143,
      "learning_rate": 4.418075302109337e-06,
      "loss": 0.3825,
      "step": 22194
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5972998142242432,
      "learning_rate": 4.417230344173279e-06,
      "loss": 0.4608,
      "step": 22195
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5063060522079468,
      "learning_rate": 4.416385444139465e-06,
      "loss": 0.3858,
      "step": 22196
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6707189083099365,
      "learning_rate": 4.415540602016663e-06,
      "loss": 0.4453,
      "step": 22197
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.538204312324524,
      "learning_rate": 4.414695817813639e-06,
      "loss": 0.3655,
      "step": 22198
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.575939655303955,
      "learning_rate": 4.4138510915391466e-06,
      "loss": 0.4158,
      "step": 22199
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6049480438232422,
      "learning_rate": 4.413006423201954e-06,
      "loss": 0.3808,
      "step": 22200
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5378081798553467,
      "learning_rate": 4.412161812810817e-06,
      "loss": 0.4688,
      "step": 22201
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5659949779510498,
      "learning_rate": 4.4113172603744995e-06,
      "loss": 0.4189,
      "step": 22202
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5742816925048828,
      "learning_rate": 4.410472765901755e-06,
      "loss": 0.392,
      "step": 22203
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6568830013275146,
      "learning_rate": 4.409628329401351e-06,
      "loss": 0.3774,
      "step": 22204
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5247713327407837,
      "learning_rate": 4.408783950882036e-06,
      "loss": 0.3883,
      "step": 22205
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5435905456542969,
      "learning_rate": 4.407939630352576e-06,
      "loss": 0.4088,
      "step": 22206
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5543736219406128,
      "learning_rate": 4.407095367821722e-06,
      "loss": 0.3938,
      "step": 22207
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4613595008850098,
      "learning_rate": 4.406251163298231e-06,
      "loss": 0.3734,
      "step": 22208
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6416704654693604,
      "learning_rate": 4.405407016790865e-06,
      "loss": 0.4352,
      "step": 22209
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5852097272872925,
      "learning_rate": 4.404562928308374e-06,
      "loss": 0.4136,
      "step": 22210
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5387169122695923,
      "learning_rate": 4.403718897859508e-06,
      "loss": 0.4687,
      "step": 22211
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4890090227127075,
      "learning_rate": 4.4028749254530265e-06,
      "loss": 0.3954,
      "step": 22212
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6129504442214966,
      "learning_rate": 4.4020310110976835e-06,
      "loss": 0.4387,
      "step": 22213
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4599919319152832,
      "learning_rate": 4.4011871548022335e-06,
      "loss": 0.393,
      "step": 22214
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5603889226913452,
      "learning_rate": 4.400343356575426e-06,
      "loss": 0.4315,
      "step": 22215
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.624950885772705,
      "learning_rate": 4.399499616426007e-06,
      "loss": 0.4036,
      "step": 22216
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5641872882843018,
      "learning_rate": 4.398655934362735e-06,
      "loss": 0.4126,
      "step": 22217
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5116355419158936,
      "learning_rate": 4.39781231039436e-06,
      "loss": 0.3816,
      "step": 22218
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6006066799163818,
      "learning_rate": 4.396968744529626e-06,
      "loss": 0.4648,
      "step": 22219
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5120011568069458,
      "learning_rate": 4.3961252367772914e-06,
      "loss": 0.3685,
      "step": 22220
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.614104151725769,
      "learning_rate": 4.395281787146093e-06,
      "loss": 0.4204,
      "step": 22221
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5285968780517578,
      "learning_rate": 4.39443839564479e-06,
      "loss": 0.3679,
      "step": 22222
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.083706855773926,
      "learning_rate": 4.39359506228212e-06,
      "loss": 0.4054,
      "step": 22223
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4970464706420898,
      "learning_rate": 4.3927517870668345e-06,
      "loss": 0.4122,
      "step": 22224
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5023047924041748,
      "learning_rate": 4.391908570007683e-06,
      "loss": 0.4232,
      "step": 22225
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.550302267074585,
      "learning_rate": 4.391065411113409e-06,
      "loss": 0.3935,
      "step": 22226
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7743185758590698,
      "learning_rate": 4.390222310392751e-06,
      "loss": 0.4707,
      "step": 22227
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5340813398361206,
      "learning_rate": 4.389379267854458e-06,
      "loss": 0.3672,
      "step": 22228
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.566900372505188,
      "learning_rate": 4.3885362835072745e-06,
      "loss": 0.4366,
      "step": 22229
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5423873662948608,
      "learning_rate": 4.387693357359946e-06,
      "loss": 0.3676,
      "step": 22230
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5304869413375854,
      "learning_rate": 4.3868504894212114e-06,
      "loss": 0.4212,
      "step": 22231
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.532461166381836,
      "learning_rate": 4.38600767969981e-06,
      "loss": 0.3872,
      "step": 22232
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6494845151901245,
      "learning_rate": 4.385164928204486e-06,
      "loss": 0.4125,
      "step": 22233
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4988312721252441,
      "learning_rate": 4.384322234943984e-06,
      "loss": 0.4094,
      "step": 22234
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5468833446502686,
      "learning_rate": 4.383479599927036e-06,
      "loss": 0.3974,
      "step": 22235
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5162526369094849,
      "learning_rate": 4.38263702316239e-06,
      "loss": 0.3762,
      "step": 22236
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6996774673461914,
      "learning_rate": 4.381794504658779e-06,
      "loss": 0.4369,
      "step": 22237
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5831117630004883,
      "learning_rate": 4.380952044424944e-06,
      "loss": 0.3723,
      "step": 22238
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5374250411987305,
      "learning_rate": 4.380109642469619e-06,
      "loss": 0.3919,
      "step": 22239
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5446650981903076,
      "learning_rate": 4.379267298801544e-06,
      "loss": 0.4015,
      "step": 22240
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6560823917388916,
      "learning_rate": 4.378425013429458e-06,
      "loss": 0.4649,
      "step": 22241
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5754754543304443,
      "learning_rate": 4.377582786362094e-06,
      "loss": 0.4255,
      "step": 22242
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.653896450996399,
      "learning_rate": 4.376740617608184e-06,
      "loss": 0.453,
      "step": 22243
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4990718364715576,
      "learning_rate": 4.375898507176466e-06,
      "loss": 0.3821,
      "step": 22244
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6166200637817383,
      "learning_rate": 4.375056455075677e-06,
      "loss": 0.4178,
      "step": 22245
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5487316846847534,
      "learning_rate": 4.374214461314548e-06,
      "loss": 0.4022,
      "step": 22246
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.677207589149475,
      "learning_rate": 4.373372525901807e-06,
      "loss": 0.9481,
      "step": 22247
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4441251754760742,
      "learning_rate": 4.37253064884619e-06,
      "loss": 0.9975,
      "step": 22248
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.684221863746643,
      "learning_rate": 4.371688830156429e-06,
      "loss": 0.4865,
      "step": 22249
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6105095148086548,
      "learning_rate": 4.370847069841259e-06,
      "loss": 0.3539,
      "step": 22250
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.032466411590576,
      "learning_rate": 4.370005367909405e-06,
      "loss": 1.017,
      "step": 22251
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4824000597000122,
      "learning_rate": 4.369163724369596e-06,
      "loss": 1.0258,
      "step": 22252
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.557618498802185,
      "learning_rate": 4.368322139230563e-06,
      "loss": 0.4436,
      "step": 22253
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0848922729492188,
      "learning_rate": 4.367480612501039e-06,
      "loss": 0.3838,
      "step": 22254
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7375704050064087,
      "learning_rate": 4.366639144189743e-06,
      "loss": 1.0032,
      "step": 22255
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.329906702041626,
      "learning_rate": 4.365797734305412e-06,
      "loss": 0.9324,
      "step": 22256
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5168840885162354,
      "learning_rate": 4.3649563828567655e-06,
      "loss": 0.4475,
      "step": 22257
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.484483242034912,
      "learning_rate": 4.364115089852534e-06,
      "loss": 0.3839,
      "step": 22258
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5810571908950806,
      "learning_rate": 4.363273855301439e-06,
      "loss": 0.434,
      "step": 22259
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.485595464706421,
      "learning_rate": 4.362432679212206e-06,
      "loss": 0.365,
      "step": 22260
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.568306803703308,
      "learning_rate": 4.361591561593565e-06,
      "loss": 0.4382,
      "step": 22261
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6208350658416748,
      "learning_rate": 4.3607505024542376e-06,
      "loss": 0.4133,
      "step": 22262
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4979327917099,
      "learning_rate": 4.359909501802939e-06,
      "loss": 0.4196,
      "step": 22263
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4500794410705566,
      "learning_rate": 4.3590685596484e-06,
      "loss": 0.361,
      "step": 22264
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5662932395935059,
      "learning_rate": 4.358227675999339e-06,
      "loss": 0.4877,
      "step": 22265
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4094934463500977,
      "learning_rate": 4.357386850864482e-06,
      "loss": 0.3662,
      "step": 22266
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.633852243423462,
      "learning_rate": 4.3565460842525475e-06,
      "loss": 0.4644,
      "step": 22267
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5354474782943726,
      "learning_rate": 4.35570537617225e-06,
      "loss": 0.3935,
      "step": 22268
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6038048267364502,
      "learning_rate": 4.354864726632314e-06,
      "loss": 0.4246,
      "step": 22269
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4867371320724487,
      "learning_rate": 4.354024135641462e-06,
      "loss": 0.3901,
      "step": 22270
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5672075748443604,
      "learning_rate": 4.353183603208403e-06,
      "loss": 0.4374,
      "step": 22271
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5505249500274658,
      "learning_rate": 4.352343129341864e-06,
      "loss": 0.3955,
      "step": 22272
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.548509120941162,
      "learning_rate": 4.3515027140505535e-06,
      "loss": 0.4073,
      "step": 22273
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5773954391479492,
      "learning_rate": 4.350662357343197e-06,
      "loss": 0.4069,
      "step": 22274
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6047391891479492,
      "learning_rate": 4.349822059228502e-06,
      "loss": 0.4657,
      "step": 22275
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4997755289077759,
      "learning_rate": 4.348981819715187e-06,
      "loss": 0.3589,
      "step": 22276
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4927005767822266,
      "learning_rate": 4.3481416388119704e-06,
      "loss": 0.4248,
      "step": 22277
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5666881799697876,
      "learning_rate": 4.347301516527563e-06,
      "loss": 0.4213,
      "step": 22278
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6023712158203125,
      "learning_rate": 4.346461452870674e-06,
      "loss": 0.5116,
      "step": 22279
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4952818155288696,
      "learning_rate": 4.34562144785002e-06,
      "loss": 0.3907,
      "step": 22280
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5373822450637817,
      "learning_rate": 4.344781501474318e-06,
      "loss": 0.4709,
      "step": 22281
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4922423362731934,
      "learning_rate": 4.343941613752275e-06,
      "loss": 0.3644,
      "step": 22282
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5226686000823975,
      "learning_rate": 4.343101784692597e-06,
      "loss": 0.4521,
      "step": 22283
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.510477900505066,
      "learning_rate": 4.3422620143040005e-06,
      "loss": 0.3733,
      "step": 22284
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5796406269073486,
      "learning_rate": 4.341422302595194e-06,
      "loss": 0.417,
      "step": 22285
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5465548038482666,
      "learning_rate": 4.340582649574889e-06,
      "loss": 0.4023,
      "step": 22286
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5707253217697144,
      "learning_rate": 4.339743055251794e-06,
      "loss": 0.4197,
      "step": 22287
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5303961038589478,
      "learning_rate": 4.33890351963461e-06,
      "loss": 0.3942,
      "step": 22288
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5365551710128784,
      "learning_rate": 4.338064042732049e-06,
      "loss": 0.4349,
      "step": 22289
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.539146900177002,
      "learning_rate": 4.337224624552821e-06,
      "loss": 0.3821,
      "step": 22290
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.693816900253296,
      "learning_rate": 4.336385265105626e-06,
      "loss": 1.0318,
      "step": 22291
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5429069995880127,
      "learning_rate": 4.335545964399176e-06,
      "loss": 0.996,
      "step": 22292
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.621687889099121,
      "learning_rate": 4.334706722442168e-06,
      "loss": 0.4441,
      "step": 22293
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5456432104110718,
      "learning_rate": 4.333867539243314e-06,
      "loss": 0.3956,
      "step": 22294
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5971367359161377,
      "learning_rate": 4.333028414811311e-06,
      "loss": 0.4403,
      "step": 22295
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5330283641815186,
      "learning_rate": 4.3321893491548664e-06,
      "loss": 0.3783,
      "step": 22296
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4806896448135376,
      "learning_rate": 4.331350342282684e-06,
      "loss": 0.4032,
      "step": 22297
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6551634073257446,
      "learning_rate": 4.330511394203462e-06,
      "loss": 0.3717,
      "step": 22298
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5293309688568115,
      "learning_rate": 4.3296725049259015e-06,
      "loss": 0.4508,
      "step": 22299
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5060924291610718,
      "learning_rate": 4.328833674458703e-06,
      "loss": 0.3862,
      "step": 22300
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5398238897323608,
      "learning_rate": 4.3279949028105684e-06,
      "loss": 0.4401,
      "step": 22301
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.55703866481781,
      "learning_rate": 4.3271561899902e-06,
      "loss": 0.3831,
      "step": 22302
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.605918049812317,
      "learning_rate": 4.326317536006295e-06,
      "loss": 0.4273,
      "step": 22303
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5082340240478516,
      "learning_rate": 4.325478940867544e-06,
      "loss": 0.3808,
      "step": 22304
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5175025463104248,
      "learning_rate": 4.324640404582651e-06,
      "loss": 0.4627,
      "step": 22305
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.519620418548584,
      "learning_rate": 4.323801927160316e-06,
      "loss": 0.3814,
      "step": 22306
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5685635805130005,
      "learning_rate": 4.322963508609228e-06,
      "loss": 0.4213,
      "step": 22307
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5119835138320923,
      "learning_rate": 4.322125148938091e-06,
      "loss": 0.3783,
      "step": 22308
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5502737760543823,
      "learning_rate": 4.321286848155592e-06,
      "loss": 0.4561,
      "step": 22309
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5184834003448486,
      "learning_rate": 4.320448606270431e-06,
      "loss": 0.3929,
      "step": 22310
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5522887706756592,
      "learning_rate": 4.3196104232912975e-06,
      "loss": 0.4475,
      "step": 22311
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5376578569412231,
      "learning_rate": 4.3187722992268876e-06,
      "loss": 0.3854,
      "step": 22312
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6062352657318115,
      "learning_rate": 4.317934234085897e-06,
      "loss": 0.471,
      "step": 22313
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6140371561050415,
      "learning_rate": 4.317096227877014e-06,
      "loss": 0.4422,
      "step": 22314
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5673400163650513,
      "learning_rate": 4.316258280608927e-06,
      "loss": 0.4608,
      "step": 22315
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5027358531951904,
      "learning_rate": 4.315420392290331e-06,
      "loss": 0.3618,
      "step": 22316
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.593215823173523,
      "learning_rate": 4.314582562929915e-06,
      "loss": 0.4259,
      "step": 22317
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5201780796051025,
      "learning_rate": 4.313744792536377e-06,
      "loss": 0.3814,
      "step": 22318
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.560369610786438,
      "learning_rate": 4.312907081118392e-06,
      "loss": 0.4746,
      "step": 22319
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4619580507278442,
      "learning_rate": 4.312069428684654e-06,
      "loss": 0.3842,
      "step": 22320
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5700558423995972,
      "learning_rate": 4.31123183524385e-06,
      "loss": 0.4229,
      "step": 22321
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5698838233947754,
      "learning_rate": 4.310394300804673e-06,
      "loss": 0.4121,
      "step": 22322
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5575649738311768,
      "learning_rate": 4.309556825375805e-06,
      "loss": 0.457,
      "step": 22323
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.487236738204956,
      "learning_rate": 4.30871940896593e-06,
      "loss": 0.3875,
      "step": 22324
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5807658433914185,
      "learning_rate": 4.3078820515837345e-06,
      "loss": 0.4371,
      "step": 22325
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5070815086364746,
      "learning_rate": 4.307044753237909e-06,
      "loss": 0.3765,
      "step": 22326
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5865715742111206,
      "learning_rate": 4.3062075139371284e-06,
      "loss": 0.4772,
      "step": 22327
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5794028043746948,
      "learning_rate": 4.305370333690085e-06,
      "loss": 0.3827,
      "step": 22328
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6133214235305786,
      "learning_rate": 4.3045332125054515e-06,
      "loss": 0.4086,
      "step": 22329
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8749573230743408,
      "learning_rate": 4.303696150391922e-06,
      "loss": 0.4662,
      "step": 22330
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5933332443237305,
      "learning_rate": 4.302859147358168e-06,
      "loss": 0.421,
      "step": 22331
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5227819681167603,
      "learning_rate": 4.302022203412874e-06,
      "loss": 0.3812,
      "step": 22332
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6617265939712524,
      "learning_rate": 4.301185318564727e-06,
      "loss": 0.4899,
      "step": 22333
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5038976669311523,
      "learning_rate": 4.300348492822399e-06,
      "loss": 0.3814,
      "step": 22334
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5728809833526611,
      "learning_rate": 4.29951172619457e-06,
      "loss": 0.4244,
      "step": 22335
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.600976824760437,
      "learning_rate": 4.2986750186899175e-06,
      "loss": 0.3967,
      "step": 22336
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6163606643676758,
      "learning_rate": 4.297838370317123e-06,
      "loss": 0.4336,
      "step": 22337
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5118590593338013,
      "learning_rate": 4.297001781084866e-06,
      "loss": 0.3941,
      "step": 22338
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0505716800689697,
      "learning_rate": 4.296165251001822e-06,
      "loss": 0.9675,
      "step": 22339
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4768954515457153,
      "learning_rate": 4.29532878007666e-06,
      "loss": 0.9767,
      "step": 22340
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6257174015045166,
      "learning_rate": 4.2944923683180604e-06,
      "loss": 0.4301,
      "step": 22341
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5295273065567017,
      "learning_rate": 4.293656015734703e-06,
      "loss": 0.3715,
      "step": 22342
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6715658903121948,
      "learning_rate": 4.292819722335253e-06,
      "loss": 0.4643,
      "step": 22343
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5291063785552979,
      "learning_rate": 4.291983488128393e-06,
      "loss": 0.386,
      "step": 22344
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5833449363708496,
      "learning_rate": 4.291147313122788e-06,
      "loss": 0.4378,
      "step": 22345
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5479042530059814,
      "learning_rate": 4.290311197327118e-06,
      "loss": 0.3963,
      "step": 22346
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7185688018798828,
      "learning_rate": 4.289475140750048e-06,
      "loss": 0.4521,
      "step": 22347
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5204094648361206,
      "learning_rate": 4.288639143400252e-06,
      "loss": 0.3771,
      "step": 22348
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5799129009246826,
      "learning_rate": 4.287803205286405e-06,
      "loss": 0.4427,
      "step": 22349
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4950380325317383,
      "learning_rate": 4.286967326417172e-06,
      "loss": 0.3776,
      "step": 22350
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4774588346481323,
      "learning_rate": 4.2861315068012196e-06,
      "loss": 0.4816,
      "step": 22351
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4901196956634521,
      "learning_rate": 4.285295746447222e-06,
      "loss": 0.3863,
      "step": 22352
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5098328590393066,
      "learning_rate": 4.2844600453638456e-06,
      "loss": 0.4498,
      "step": 22353
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.496854305267334,
      "learning_rate": 4.283624403559764e-06,
      "loss": 0.3852,
      "step": 22354
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5006521940231323,
      "learning_rate": 4.282788821043632e-06,
      "loss": 0.4272,
      "step": 22355
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5063856840133667,
      "learning_rate": 4.281953297824122e-06,
      "loss": 0.3604,
      "step": 22356
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6483230590820312,
      "learning_rate": 4.2811178339099e-06,
      "loss": 0.4451,
      "step": 22357
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5950034856796265,
      "learning_rate": 4.280282429309635e-06,
      "loss": 0.3965,
      "step": 22358
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5571324825286865,
      "learning_rate": 4.279447084031987e-06,
      "loss": 0.4129,
      "step": 22359
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5989540815353394,
      "learning_rate": 4.278611798085617e-06,
      "loss": 0.392,
      "step": 22360
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.577768325805664,
      "learning_rate": 4.277776571479192e-06,
      "loss": 0.4414,
      "step": 22361
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5136287212371826,
      "learning_rate": 4.276941404221377e-06,
      "loss": 0.3804,
      "step": 22362
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4979125261306763,
      "learning_rate": 4.276106296320828e-06,
      "loss": 0.4222,
      "step": 22363
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5679237842559814,
      "learning_rate": 4.275271247786213e-06,
      "loss": 0.3806,
      "step": 22364
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6229946613311768,
      "learning_rate": 4.274436258626187e-06,
      "loss": 0.4255,
      "step": 22365
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.498620629310608,
      "learning_rate": 4.273601328849415e-06,
      "loss": 0.368,
      "step": 22366
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5692566633224487,
      "learning_rate": 4.272766458464551e-06,
      "loss": 0.4543,
      "step": 22367
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4984482526779175,
      "learning_rate": 4.2719316474802565e-06,
      "loss": 0.3784,
      "step": 22368
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5427074432373047,
      "learning_rate": 4.271096895905194e-06,
      "loss": 0.4343,
      "step": 22369
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4973050355911255,
      "learning_rate": 4.270262203748019e-06,
      "loss": 0.3912,
      "step": 22370
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9098331928253174,
      "learning_rate": 4.269427571017384e-06,
      "loss": 0.503,
      "step": 22371
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6466519832611084,
      "learning_rate": 4.268592997721947e-06,
      "loss": 0.4033,
      "step": 22372
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4968903064727783,
      "learning_rate": 4.2677584838703665e-06,
      "loss": 0.4388,
      "step": 22373
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5346777439117432,
      "learning_rate": 4.2669240294713e-06,
      "loss": 0.3929,
      "step": 22374
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.533105492591858,
      "learning_rate": 4.266089634533399e-06,
      "loss": 0.4215,
      "step": 22375
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5390195846557617,
      "learning_rate": 4.265255299065315e-06,
      "loss": 0.3944,
      "step": 22376
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.909970998764038,
      "learning_rate": 4.2644210230757035e-06,
      "loss": 1.0228,
      "step": 22377
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5036466121673584,
      "learning_rate": 4.26358680657322e-06,
      "loss": 1.011,
      "step": 22378
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7246419191360474,
      "learning_rate": 4.2627526495665115e-06,
      "loss": 0.4242,
      "step": 22379
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4965895414352417,
      "learning_rate": 4.261918552064236e-06,
      "loss": 0.3883,
      "step": 22380
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4933884143829346,
      "learning_rate": 4.261084514075037e-06,
      "loss": 0.4066,
      "step": 22381
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5446193218231201,
      "learning_rate": 4.260250535607572e-06,
      "loss": 0.421,
      "step": 22382
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5315048694610596,
      "learning_rate": 4.2594166166704846e-06,
      "loss": 0.4266,
      "step": 22383
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.471366047859192,
      "learning_rate": 4.2585827572724255e-06,
      "loss": 0.3595,
      "step": 22384
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5586357116699219,
      "learning_rate": 4.257748957422049e-06,
      "loss": 0.4845,
      "step": 22385
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6800432205200195,
      "learning_rate": 4.256915217127997e-06,
      "loss": 0.3952,
      "step": 22386
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5712438821792603,
      "learning_rate": 4.256081536398914e-06,
      "loss": 0.4587,
      "step": 22387
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5327657461166382,
      "learning_rate": 4.2552479152434515e-06,
      "loss": 0.3948,
      "step": 22388
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5352959632873535,
      "learning_rate": 4.254414353670253e-06,
      "loss": 0.4395,
      "step": 22389
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5310717821121216,
      "learning_rate": 4.253580851687971e-06,
      "loss": 0.4281,
      "step": 22390
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5564963817596436,
      "learning_rate": 4.252747409305243e-06,
      "loss": 0.4493,
      "step": 22391
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5707422494888306,
      "learning_rate": 4.251914026530712e-06,
      "loss": 0.3735,
      "step": 22392
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4451353549957275,
      "learning_rate": 4.251080703373023e-06,
      "loss": 0.3942,
      "step": 22393
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4917845726013184,
      "learning_rate": 4.250247439840824e-06,
      "loss": 0.3844,
      "step": 22394
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6622329950332642,
      "learning_rate": 4.249414235942755e-06,
      "loss": 0.4558,
      "step": 22395
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5663994550704956,
      "learning_rate": 4.248581091687451e-06,
      "loss": 0.3863,
      "step": 22396
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5229300260543823,
      "learning_rate": 4.2477480070835575e-06,
      "loss": 0.4266,
      "step": 22397
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5075517892837524,
      "learning_rate": 4.24691498213972e-06,
      "loss": 0.3737,
      "step": 22398
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5470887422561646,
      "learning_rate": 4.24608201686457e-06,
      "loss": 0.3985,
      "step": 22399
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4965287446975708,
      "learning_rate": 4.245249111266755e-06,
      "loss": 0.3892,
      "step": 22400
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5257023572921753,
      "learning_rate": 4.244416265354904e-06,
      "loss": 0.4139,
      "step": 22401
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5544421672821045,
      "learning_rate": 4.243583479137664e-06,
      "loss": 0.3734,
      "step": 22402
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5412529706954956,
      "learning_rate": 4.242750752623666e-06,
      "loss": 0.4513,
      "step": 22403
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4614323377609253,
      "learning_rate": 4.241918085821547e-06,
      "loss": 0.3799,
      "step": 22404
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5526865720748901,
      "learning_rate": 4.241085478739949e-06,
      "loss": 0.424,
      "step": 22405
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5559580326080322,
      "learning_rate": 4.240252931387501e-06,
      "loss": 0.4013,
      "step": 22406
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5795702934265137,
      "learning_rate": 4.239420443772844e-06,
      "loss": 0.4852,
      "step": 22407
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4849989414215088,
      "learning_rate": 4.238588015904603e-06,
      "loss": 0.379,
      "step": 22408
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7614120244979858,
      "learning_rate": 4.237755647791419e-06,
      "loss": 1.0158,
      "step": 22409
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.461294412612915,
      "learning_rate": 4.236923339441928e-06,
      "loss": 0.9694,
      "step": 22410
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.554074764251709,
      "learning_rate": 4.236091090864756e-06,
      "loss": 0.4223,
      "step": 22411
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.512503743171692,
      "learning_rate": 4.235258902068532e-06,
      "loss": 0.4143,
      "step": 22412
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5220750570297241,
      "learning_rate": 4.234426773061893e-06,
      "loss": 0.393,
      "step": 22413
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.515990972518921,
      "learning_rate": 4.233594703853467e-06,
      "loss": 0.3848,
      "step": 22414
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6289355754852295,
      "learning_rate": 4.232762694451889e-06,
      "loss": 0.4651,
      "step": 22415
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.503630518913269,
      "learning_rate": 4.231930744865784e-06,
      "loss": 0.3625,
      "step": 22416
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5342947244644165,
      "learning_rate": 4.2310988551037766e-06,
      "loss": 0.4693,
      "step": 22417
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5099846124649048,
      "learning_rate": 4.230267025174501e-06,
      "loss": 0.3862,
      "step": 22418
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7000840902328491,
      "learning_rate": 4.229435255086584e-06,
      "loss": 0.448,
      "step": 22419
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5864369869232178,
      "learning_rate": 4.228603544848648e-06,
      "loss": 0.4277,
      "step": 22420
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5761537551879883,
      "learning_rate": 4.227771894469326e-06,
      "loss": 0.4179,
      "step": 22421
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.546325922012329,
      "learning_rate": 4.2269403039572365e-06,
      "loss": 0.406,
      "step": 22422
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6138161420822144,
      "learning_rate": 4.226108773321011e-06,
      "loss": 0.443,
      "step": 22423
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.499375581741333,
      "learning_rate": 4.225277302569266e-06,
      "loss": 0.3841,
      "step": 22424
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.51408052444458,
      "learning_rate": 4.2244458917106304e-06,
      "loss": 0.3928,
      "step": 22425
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.640117883682251,
      "learning_rate": 4.22361454075373e-06,
      "loss": 0.4456,
      "step": 22426
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5201514959335327,
      "learning_rate": 4.222783249707184e-06,
      "loss": 0.455,
      "step": 22427
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5528606176376343,
      "learning_rate": 4.221952018579609e-06,
      "loss": 0.4107,
      "step": 22428
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.579622745513916,
      "learning_rate": 4.221120847379633e-06,
      "loss": 0.4274,
      "step": 22429
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5320913791656494,
      "learning_rate": 4.220289736115873e-06,
      "loss": 0.4088,
      "step": 22430
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8458563089370728,
      "learning_rate": 4.219458684796955e-06,
      "loss": 0.9794,
      "step": 22431
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6558003425598145,
      "learning_rate": 4.218627693431495e-06,
      "loss": 1.017,
      "step": 22432
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4904314279556274,
      "learning_rate": 4.217796762028106e-06,
      "loss": 0.3961,
      "step": 22433
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4553191661834717,
      "learning_rate": 4.216965890595411e-06,
      "loss": 0.3738,
      "step": 22434
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.549307107925415,
      "learning_rate": 4.2161350791420305e-06,
      "loss": 0.4088,
      "step": 22435
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6288865804672241,
      "learning_rate": 4.21530432767658e-06,
      "loss": 0.3844,
      "step": 22436
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6290860176086426,
      "learning_rate": 4.214473636207668e-06,
      "loss": 0.4596,
      "step": 22437
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5313801765441895,
      "learning_rate": 4.213643004743917e-06,
      "loss": 0.3795,
      "step": 22438
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.613124132156372,
      "learning_rate": 4.2128124332939435e-06,
      "loss": 0.4457,
      "step": 22439
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.495348572731018,
      "learning_rate": 4.211981921866356e-06,
      "loss": 0.386,
      "step": 22440
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5069888830184937,
      "learning_rate": 4.2111514704697754e-06,
      "loss": 0.4543,
      "step": 22441
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.0467965602874756,
      "learning_rate": 4.210321079112808e-06,
      "loss": 0.3949,
      "step": 22442
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7550764083862305,
      "learning_rate": 4.209490747804071e-06,
      "loss": 0.9307,
      "step": 22443
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4530638456344604,
      "learning_rate": 4.2086604765521725e-06,
      "loss": 1.0058,
      "step": 22444
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7922029495239258,
      "learning_rate": 4.207830265365724e-06,
      "loss": 0.3997,
      "step": 22445
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.532516598701477,
      "learning_rate": 4.207000114253343e-06,
      "loss": 0.365,
      "step": 22446
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5462851524353027,
      "learning_rate": 4.2061700232236335e-06,
      "loss": 0.4407,
      "step": 22447
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.480867624282837,
      "learning_rate": 4.205339992285202e-06,
      "loss": 0.3685,
      "step": 22448
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6426212787628174,
      "learning_rate": 4.204510021446662e-06,
      "loss": 0.9571,
      "step": 22449
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4551644325256348,
      "learning_rate": 4.203680110716619e-06,
      "loss": 1.0191,
      "step": 22450
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.526921272277832,
      "learning_rate": 4.202850260103687e-06,
      "loss": 0.4416,
      "step": 22451
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.481913447380066,
      "learning_rate": 4.2020204696164665e-06,
      "loss": 0.3835,
      "step": 22452
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5480040311813354,
      "learning_rate": 4.201190739263563e-06,
      "loss": 0.4123,
      "step": 22453
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5034457445144653,
      "learning_rate": 4.200361069053583e-06,
      "loss": 0.3826,
      "step": 22454
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6151906251907349,
      "learning_rate": 4.199531458995137e-06,
      "loss": 0.4729,
      "step": 22455
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5184005498886108,
      "learning_rate": 4.198701909096823e-06,
      "loss": 0.381,
      "step": 22456
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5705829858779907,
      "learning_rate": 4.197872419367248e-06,
      "loss": 0.4383,
      "step": 22457
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4710921049118042,
      "learning_rate": 4.197042989815012e-06,
      "loss": 0.3739,
      "step": 22458
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8615001440048218,
      "learning_rate": 4.196213620448724e-06,
      "loss": 0.4037,
      "step": 22459
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.551169514656067,
      "learning_rate": 4.195384311276976e-06,
      "loss": 0.3905,
      "step": 22460
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.551689624786377,
      "learning_rate": 4.194555062308375e-06,
      "loss": 0.4283,
      "step": 22461
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5216401815414429,
      "learning_rate": 4.193725873551526e-06,
      "loss": 0.3973,
      "step": 22462
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6148995161056519,
      "learning_rate": 4.1928967450150235e-06,
      "loss": 0.468,
      "step": 22463
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.599652886390686,
      "learning_rate": 4.192067676707464e-06,
      "loss": 0.4064,
      "step": 22464
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6773024797439575,
      "learning_rate": 4.1912386686374515e-06,
      "loss": 0.5064,
      "step": 22465
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5000149011611938,
      "learning_rate": 4.190409720813582e-06,
      "loss": 0.3813,
      "step": 22466
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5438088178634644,
      "learning_rate": 4.189580833244456e-06,
      "loss": 0.4627,
      "step": 22467
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4803175926208496,
      "learning_rate": 4.18875200593867e-06,
      "loss": 0.3892,
      "step": 22468
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5169386863708496,
      "learning_rate": 4.187923238904814e-06,
      "loss": 0.4136,
      "step": 22469
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4773402214050293,
      "learning_rate": 4.187094532151488e-06,
      "loss": 0.3773,
      "step": 22470
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.556362509727478,
      "learning_rate": 4.186265885687291e-06,
      "loss": 0.416,
      "step": 22471
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4815136194229126,
      "learning_rate": 4.185437299520812e-06,
      "loss": 0.36,
      "step": 22472
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5779473781585693,
      "learning_rate": 4.1846087736606434e-06,
      "loss": 0.4467,
      "step": 22473
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6042084693908691,
      "learning_rate": 4.183780308115382e-06,
      "loss": 0.4218,
      "step": 22474
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5696804523468018,
      "learning_rate": 4.182951902893621e-06,
      "loss": 0.442,
      "step": 22475
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5004034042358398,
      "learning_rate": 4.182123558003949e-06,
      "loss": 0.4011,
      "step": 22476
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7450571060180664,
      "learning_rate": 4.181295273454962e-06,
      "loss": 0.4963,
      "step": 22477
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5041402578353882,
      "learning_rate": 4.1804670492552435e-06,
      "loss": 0.3932,
      "step": 22478
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5497881174087524,
      "learning_rate": 4.179638885413392e-06,
      "loss": 0.4305,
      "step": 22479
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5806145668029785,
      "learning_rate": 4.178810781937987e-06,
      "loss": 0.4048,
      "step": 22480
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.646618127822876,
      "learning_rate": 4.1779827388376244e-06,
      "loss": 0.4249,
      "step": 22481
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5729560852050781,
      "learning_rate": 4.177154756120895e-06,
      "loss": 0.3897,
      "step": 22482
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5353477001190186,
      "learning_rate": 4.17632683379638e-06,
      "loss": 0.4282,
      "step": 22483
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5842992067337036,
      "learning_rate": 4.1754989718726666e-06,
      "loss": 0.4133,
      "step": 22484
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6051857471466064,
      "learning_rate": 4.174671170358342e-06,
      "loss": 0.4276,
      "step": 22485
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5200669765472412,
      "learning_rate": 4.173843429261992e-06,
      "loss": 0.3937,
      "step": 22486
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5356372594833374,
      "learning_rate": 4.173015748592206e-06,
      "loss": 0.5024,
      "step": 22487
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5735801458358765,
      "learning_rate": 4.1721881283575646e-06,
      "loss": 0.3894,
      "step": 22488
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5111626386642456,
      "learning_rate": 4.171360568566648e-06,
      "loss": 0.3996,
      "step": 22489
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5500140190124512,
      "learning_rate": 4.170533069228042e-06,
      "loss": 0.3993,
      "step": 22490
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6560068130493164,
      "learning_rate": 4.169705630350335e-06,
      "loss": 0.4162,
      "step": 22491
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5150558948516846,
      "learning_rate": 4.168878251942099e-06,
      "loss": 0.3649,
      "step": 22492
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4934492111206055,
      "learning_rate": 4.168050934011924e-06,
      "loss": 0.4122,
      "step": 22493
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.579158902168274,
      "learning_rate": 4.167223676568382e-06,
      "loss": 0.4111,
      "step": 22494
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5599377155303955,
      "learning_rate": 4.166396479620062e-06,
      "loss": 0.445,
      "step": 22495
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.54575777053833,
      "learning_rate": 4.165569343175535e-06,
      "loss": 0.3927,
      "step": 22496
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5822948217391968,
      "learning_rate": 4.164742267243385e-06,
      "loss": 0.4078,
      "step": 22497
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5747376680374146,
      "learning_rate": 4.16391525183219e-06,
      "loss": 0.3801,
      "step": 22498
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4964455366134644,
      "learning_rate": 4.163088296950528e-06,
      "loss": 0.412,
      "step": 22499
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5382592678070068,
      "learning_rate": 4.16226140260697e-06,
      "loss": 0.378,
      "step": 22500
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6493250131607056,
      "learning_rate": 4.161434568810095e-06,
      "loss": 0.4417,
      "step": 22501
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6266703605651855,
      "learning_rate": 4.160607795568482e-06,
      "loss": 0.3953,
      "step": 22502
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6228731870651245,
      "learning_rate": 4.159781082890706e-06,
      "loss": 0.4461,
      "step": 22503
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.529585838317871,
      "learning_rate": 4.158954430785339e-06,
      "loss": 0.3873,
      "step": 22504
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5178050994873047,
      "learning_rate": 4.158127839260952e-06,
      "loss": 0.4365,
      "step": 22505
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5199815034866333,
      "learning_rate": 4.157301308326121e-06,
      "loss": 0.382,
      "step": 22506
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5957447290420532,
      "learning_rate": 4.156474837989422e-06,
      "loss": 0.3807,
      "step": 22507
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4877206087112427,
      "learning_rate": 4.155648428259419e-06,
      "loss": 0.3717,
      "step": 22508
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6361103057861328,
      "learning_rate": 4.154822079144692e-06,
      "loss": 0.4824,
      "step": 22509
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.499692440032959,
      "learning_rate": 4.153995790653802e-06,
      "loss": 0.3893,
      "step": 22510
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5698697566986084,
      "learning_rate": 4.153169562795328e-06,
      "loss": 0.4729,
      "step": 22511
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5643547773361206,
      "learning_rate": 4.152343395577832e-06,
      "loss": 0.3842,
      "step": 22512
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5837304592132568,
      "learning_rate": 4.1515172890098894e-06,
      "loss": 0.4231,
      "step": 22513
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.514678955078125,
      "learning_rate": 4.150691243100061e-06,
      "loss": 0.3744,
      "step": 22514
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5366955995559692,
      "learning_rate": 4.149865257856921e-06,
      "loss": 0.4922,
      "step": 22515
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5270824432373047,
      "learning_rate": 4.14903933328903e-06,
      "loss": 0.3885,
      "step": 22516
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4656479358673096,
      "learning_rate": 4.148213469404957e-06,
      "loss": 0.4018,
      "step": 22517
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.537084937095642,
      "learning_rate": 4.147387666213272e-06,
      "loss": 0.3804,
      "step": 22518
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.539048194885254,
      "learning_rate": 4.1465619237225355e-06,
      "loss": 0.4403,
      "step": 22519
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5274955034255981,
      "learning_rate": 4.145736241941308e-06,
      "loss": 0.3968,
      "step": 22520
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6364405155181885,
      "learning_rate": 4.144910620878158e-06,
      "loss": 0.4674,
      "step": 22521
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5526570081710815,
      "learning_rate": 4.144085060541646e-06,
      "loss": 0.3662,
      "step": 22522
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5055181980133057,
      "learning_rate": 4.143259560940342e-06,
      "loss": 0.4159,
      "step": 22523
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.464209794998169,
      "learning_rate": 4.142434122082799e-06,
      "loss": 0.3761,
      "step": 22524
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5355870723724365,
      "learning_rate": 4.141608743977579e-06,
      "loss": 0.4955,
      "step": 22525
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.567123293876648,
      "learning_rate": 4.1407834266332436e-06,
      "loss": 0.3737,
      "step": 22526
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.555823564529419,
      "learning_rate": 4.139958170058358e-06,
      "loss": 0.4554,
      "step": 22527
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4871219396591187,
      "learning_rate": 4.139132974261473e-06,
      "loss": 0.3857,
      "step": 22528
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5529221296310425,
      "learning_rate": 4.138307839251153e-06,
      "loss": 0.4242,
      "step": 22529
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5660825967788696,
      "learning_rate": 4.137482765035953e-06,
      "loss": 0.4048,
      "step": 22530
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.461449146270752,
      "learning_rate": 4.136657751624433e-06,
      "loss": 0.3999,
      "step": 22531
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4658502340316772,
      "learning_rate": 4.135832799025145e-06,
      "loss": 0.3677,
      "step": 22532
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4856117963790894,
      "learning_rate": 4.135007907246649e-06,
      "loss": 0.4006,
      "step": 22533
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4546951055526733,
      "learning_rate": 4.134183076297502e-06,
      "loss": 0.3731,
      "step": 22534
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6790130138397217,
      "learning_rate": 4.133358306186257e-06,
      "loss": 0.4611,
      "step": 22535
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.555499792098999,
      "learning_rate": 4.132533596921464e-06,
      "loss": 0.3984,
      "step": 22536
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5466676950454712,
      "learning_rate": 4.13170894851168e-06,
      "loss": 0.4387,
      "step": 22537
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6468290090560913,
      "learning_rate": 4.130884360965458e-06,
      "loss": 0.4167,
      "step": 22538
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5535928010940552,
      "learning_rate": 4.130059834291354e-06,
      "loss": 0.4509,
      "step": 22539
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.1742331981658936,
      "learning_rate": 4.129235368497917e-06,
      "loss": 0.3972,
      "step": 22540
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.546794056892395,
      "learning_rate": 4.128410963593692e-06,
      "loss": 0.4362,
      "step": 22541
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.483908772468567,
      "learning_rate": 4.127586619587235e-06,
      "loss": 0.38,
      "step": 22542
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.511475920677185,
      "learning_rate": 4.126762336487099e-06,
      "loss": 0.4609,
      "step": 22543
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4901165962219238,
      "learning_rate": 4.1259381143018255e-06,
      "loss": 0.3784,
      "step": 22544
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5231760740280151,
      "learning_rate": 4.125113953039971e-06,
      "loss": 0.4262,
      "step": 22545
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4860830307006836,
      "learning_rate": 4.124289852710075e-06,
      "loss": 0.3876,
      "step": 22546
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5837129354476929,
      "learning_rate": 4.1234658133206926e-06,
      "loss": 0.4121,
      "step": 22547
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5624233484268188,
      "learning_rate": 4.122641834880363e-06,
      "loss": 0.3698,
      "step": 22548
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.607113242149353,
      "learning_rate": 4.121817917397637e-06,
      "loss": 0.3946,
      "step": 22549
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5315241813659668,
      "learning_rate": 4.120994060881062e-06,
      "loss": 0.3728,
      "step": 22550
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6749643087387085,
      "learning_rate": 4.12017026533918e-06,
      "loss": 0.4272,
      "step": 22551
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5205460786819458,
      "learning_rate": 4.1193465307805295e-06,
      "loss": 0.3898,
      "step": 22552
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6489813327789307,
      "learning_rate": 4.1185228572136605e-06,
      "loss": 1.0456,
      "step": 22553
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4267672300338745,
      "learning_rate": 4.117699244647118e-06,
      "loss": 1.005,
      "step": 22554
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5972719192504883,
      "learning_rate": 4.116875693089439e-06,
      "loss": 0.403,
      "step": 22555
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.588775396347046,
      "learning_rate": 4.116052202549165e-06,
      "loss": 0.3744,
      "step": 22556
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.533771276473999,
      "learning_rate": 4.115228773034837e-06,
      "loss": 0.4242,
      "step": 22557
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.525727391242981,
      "learning_rate": 4.1144054045549985e-06,
      "loss": 0.3511,
      "step": 22558
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.537344217300415,
      "learning_rate": 4.1135820971181895e-06,
      "loss": 0.4359,
      "step": 22559
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5462349653244019,
      "learning_rate": 4.112758850732949e-06,
      "loss": 0.4018,
      "step": 22560
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5535091161727905,
      "learning_rate": 4.111935665407808e-06,
      "loss": 0.4224,
      "step": 22561
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.571081280708313,
      "learning_rate": 4.11111254115131e-06,
      "loss": 0.4197,
      "step": 22562
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.459932804107666,
      "learning_rate": 4.110289477971996e-06,
      "loss": 0.4405,
      "step": 22563
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4925270080566406,
      "learning_rate": 4.109466475878394e-06,
      "loss": 0.3686,
      "step": 22564
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.538869023323059,
      "learning_rate": 4.108643534879048e-06,
      "loss": 0.3981,
      "step": 22565
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4754046201705933,
      "learning_rate": 4.107820654982485e-06,
      "loss": 0.3705,
      "step": 22566
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5056989192962646,
      "learning_rate": 4.1069978361972486e-06,
      "loss": 0.4208,
      "step": 22567
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5078133344650269,
      "learning_rate": 4.106175078531864e-06,
      "loss": 0.3674,
      "step": 22568
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6070446968078613,
      "learning_rate": 4.105352381994868e-06,
      "loss": 0.3822,
      "step": 22569
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.605207085609436,
      "learning_rate": 4.104529746594799e-06,
      "loss": 0.4167,
      "step": 22570
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5435703992843628,
      "learning_rate": 4.103707172340182e-06,
      "loss": 0.4187,
      "step": 22571
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5805591344833374,
      "learning_rate": 4.102884659239547e-06,
      "loss": 0.4,
      "step": 22572
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.645944595336914,
      "learning_rate": 4.102062207301429e-06,
      "loss": 0.4607,
      "step": 22573
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.523972749710083,
      "learning_rate": 4.101239816534357e-06,
      "loss": 0.38,
      "step": 22574
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6581895351409912,
      "learning_rate": 4.100417486946864e-06,
      "loss": 0.4357,
      "step": 22575
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5036816596984863,
      "learning_rate": 4.099595218547475e-06,
      "loss": 0.3906,
      "step": 22576
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5055148601531982,
      "learning_rate": 4.098773011344715e-06,
      "loss": 0.4389,
      "step": 22577
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5496296882629395,
      "learning_rate": 4.097950865347117e-06,
      "loss": 0.3856,
      "step": 22578
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5809019804000854,
      "learning_rate": 4.097128780563209e-06,
      "loss": 0.3953,
      "step": 22579
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5346976518630981,
      "learning_rate": 4.096306757001511e-06,
      "loss": 0.3925,
      "step": 22580
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6386475563049316,
      "learning_rate": 4.095484794670557e-06,
      "loss": 0.4239,
      "step": 22581
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5056387186050415,
      "learning_rate": 4.094662893578862e-06,
      "loss": 0.3618,
      "step": 22582
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4938924312591553,
      "learning_rate": 4.093841053734961e-06,
      "loss": 0.451,
      "step": 22583
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4791821241378784,
      "learning_rate": 4.093019275147369e-06,
      "loss": 0.3705,
      "step": 22584
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.536260962486267,
      "learning_rate": 4.092197557824612e-06,
      "loss": 0.456,
      "step": 22585
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5952162742614746,
      "learning_rate": 4.091375901775219e-06,
      "loss": 0.3864,
      "step": 22586
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8173868656158447,
      "learning_rate": 4.0905543070077036e-06,
      "loss": 0.4874,
      "step": 22587
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5633021593093872,
      "learning_rate": 4.089732773530588e-06,
      "loss": 0.4071,
      "step": 22588
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6120126247406006,
      "learning_rate": 4.088911301352394e-06,
      "loss": 0.432,
      "step": 22589
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.512087106704712,
      "learning_rate": 4.088089890481646e-06,
      "loss": 0.384,
      "step": 22590
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5581814050674438,
      "learning_rate": 4.087268540926859e-06,
      "loss": 0.4475,
      "step": 22591
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.48357093334198,
      "learning_rate": 4.0864472526965485e-06,
      "loss": 0.3644,
      "step": 22592
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7171664237976074,
      "learning_rate": 4.085626025799237e-06,
      "loss": 0.4319,
      "step": 22593
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.657277226448059,
      "learning_rate": 4.0848048602434395e-06,
      "loss": 0.3901,
      "step": 22594
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6335644721984863,
      "learning_rate": 4.0839837560376796e-06,
      "loss": 0.4628,
      "step": 22595
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5113908052444458,
      "learning_rate": 4.083162713190467e-06,
      "loss": 0.3703,
      "step": 22596
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5558515787124634,
      "learning_rate": 4.0823417317103165e-06,
      "loss": 0.46,
      "step": 22597
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.749303936958313,
      "learning_rate": 4.081520811605744e-06,
      "loss": 0.4455,
      "step": 22598
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.656603455543518,
      "learning_rate": 4.0806999528852684e-06,
      "loss": 0.5186,
      "step": 22599
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4707424640655518,
      "learning_rate": 4.079879155557397e-06,
      "loss": 0.3745,
      "step": 22600
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5659239292144775,
      "learning_rate": 4.079058419630648e-06,
      "loss": 0.4467,
      "step": 22601
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4460028409957886,
      "learning_rate": 4.078237745113529e-06,
      "loss": 0.4015,
      "step": 22602
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5535553693771362,
      "learning_rate": 4.077417132014557e-06,
      "loss": 0.4569,
      "step": 22603
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5164313316345215,
      "learning_rate": 4.076596580342236e-06,
      "loss": 0.3974,
      "step": 22604
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.663102388381958,
      "learning_rate": 4.07577609010508e-06,
      "loss": 0.4404,
      "step": 22605
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5049303770065308,
      "learning_rate": 4.074955661311605e-06,
      "loss": 0.3932,
      "step": 22606
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5197875499725342,
      "learning_rate": 4.074135293970312e-06,
      "loss": 0.418,
      "step": 22607
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6201313734054565,
      "learning_rate": 4.0733149880897095e-06,
      "loss": 0.4265,
      "step": 22608
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5519659519195557,
      "learning_rate": 4.072494743678308e-06,
      "loss": 0.4389,
      "step": 22609
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4991142749786377,
      "learning_rate": 4.071674560744614e-06,
      "loss": 0.3842,
      "step": 22610
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.523764967918396,
      "learning_rate": 4.070854439297138e-06,
      "loss": 0.4248,
      "step": 22611
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5408908128738403,
      "learning_rate": 4.070034379344384e-06,
      "loss": 0.3946,
      "step": 22612
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4143210649490356,
      "learning_rate": 4.069214380894851e-06,
      "loss": 0.3694,
      "step": 22613
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.577392578125,
      "learning_rate": 4.06839444395705e-06,
      "loss": 0.4032,
      "step": 22614
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5156586170196533,
      "learning_rate": 4.067574568539486e-06,
      "loss": 0.428,
      "step": 22615
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4864211082458496,
      "learning_rate": 4.066754754650658e-06,
      "loss": 0.3499,
      "step": 22616
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5681736469268799,
      "learning_rate": 4.065935002299073e-06,
      "loss": 0.4248,
      "step": 22617
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4665881395339966,
      "learning_rate": 4.0651153114932286e-06,
      "loss": 0.3657,
      "step": 22618
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5330439805984497,
      "learning_rate": 4.064295682241631e-06,
      "loss": 0.4313,
      "step": 22619
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.529068946838379,
      "learning_rate": 4.063476114552777e-06,
      "loss": 0.4119,
      "step": 22620
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5514377355575562,
      "learning_rate": 4.0626566084351685e-06,
      "loss": 0.4405,
      "step": 22621
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5357584953308105,
      "learning_rate": 4.061837163897308e-06,
      "loss": 0.402,
      "step": 22622
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6426851749420166,
      "learning_rate": 4.061017780947691e-06,
      "loss": 0.4247,
      "step": 22623
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6145819425582886,
      "learning_rate": 4.060198459594814e-06,
      "loss": 0.4217,
      "step": 22624
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.644879698753357,
      "learning_rate": 4.059379199847178e-06,
      "loss": 0.4233,
      "step": 22625
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5764251947402954,
      "learning_rate": 4.058560001713278e-06,
      "loss": 0.4108,
      "step": 22626
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5265804529190063,
      "learning_rate": 4.057740865201618e-06,
      "loss": 0.431,
      "step": 22627
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5406595468521118,
      "learning_rate": 4.056921790320681e-06,
      "loss": 0.3923,
      "step": 22628
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4904813766479492,
      "learning_rate": 4.05610277707897e-06,
      "loss": 0.3762,
      "step": 22629
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5663142204284668,
      "learning_rate": 4.055283825484977e-06,
      "loss": 0.3958,
      "step": 22630
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7214215993881226,
      "learning_rate": 4.0544649355472e-06,
      "loss": 0.9468,
      "step": 22631
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4209380149841309,
      "learning_rate": 4.053646107274129e-06,
      "loss": 0.9681,
      "step": 22632
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5442897081375122,
      "learning_rate": 4.052827340674253e-06,
      "loss": 0.4878,
      "step": 22633
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4980217218399048,
      "learning_rate": 4.052008635756067e-06,
      "loss": 0.3919,
      "step": 22634
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5271668434143066,
      "learning_rate": 4.051189992528067e-06,
      "loss": 0.4526,
      "step": 22635
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5394442081451416,
      "learning_rate": 4.050371410998737e-06,
      "loss": 0.3746,
      "step": 22636
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5281790494918823,
      "learning_rate": 4.049552891176571e-06,
      "loss": 0.4512,
      "step": 22637
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6827627420425415,
      "learning_rate": 4.048734433070054e-06,
      "loss": 0.3978,
      "step": 22638
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6102451086044312,
      "learning_rate": 4.0479160366876815e-06,
      "loss": 0.437,
      "step": 22639
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5511693954467773,
      "learning_rate": 4.047097702037933e-06,
      "loss": 0.3776,
      "step": 22640
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5382477045059204,
      "learning_rate": 4.046279429129302e-06,
      "loss": 0.4378,
      "step": 22641
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4736770391464233,
      "learning_rate": 4.045461217970275e-06,
      "loss": 0.3677,
      "step": 22642
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5663511753082275,
      "learning_rate": 4.044643068569339e-06,
      "loss": 0.4201,
      "step": 22643
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5183172225952148,
      "learning_rate": 4.043824980934974e-06,
      "loss": 0.3952,
      "step": 22644
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5709694623947144,
      "learning_rate": 4.043006955075667e-06,
      "loss": 0.4217,
      "step": 22645
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6175795793533325,
      "learning_rate": 4.042188990999905e-06,
      "loss": 0.3814,
      "step": 22646
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.446850061416626,
      "learning_rate": 4.041371088716172e-06,
      "loss": 0.3768,
      "step": 22647
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5398412942886353,
      "learning_rate": 4.04055324823295e-06,
      "loss": 0.3954,
      "step": 22648
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.509708046913147,
      "learning_rate": 4.039735469558718e-06,
      "loss": 0.4057,
      "step": 22649
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5867648124694824,
      "learning_rate": 4.038917752701959e-06,
      "loss": 0.3943,
      "step": 22650
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5744549036026,
      "learning_rate": 4.038100097671155e-06,
      "loss": 0.4236,
      "step": 22651
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4575384855270386,
      "learning_rate": 4.03728250447479e-06,
      "loss": 0.3842,
      "step": 22652
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4738787412643433,
      "learning_rate": 4.036464973121341e-06,
      "loss": 0.4258,
      "step": 22653
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5577880144119263,
      "learning_rate": 4.035647503619282e-06,
      "loss": 0.3881,
      "step": 22654
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.552148461341858,
      "learning_rate": 4.034830095977097e-06,
      "loss": 0.4399,
      "step": 22655
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5328221321105957,
      "learning_rate": 4.034012750203266e-06,
      "loss": 0.3818,
      "step": 22656
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5111029148101807,
      "learning_rate": 4.033195466306259e-06,
      "loss": 0.4309,
      "step": 22657
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5893365144729614,
      "learning_rate": 4.03237824429456e-06,
      "loss": 0.3996,
      "step": 22658
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.550756573677063,
      "learning_rate": 4.031561084176637e-06,
      "loss": 0.4489,
      "step": 22659
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.558877944946289,
      "learning_rate": 4.030743985960974e-06,
      "loss": 0.3779,
      "step": 22660
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.571542739868164,
      "learning_rate": 4.029926949656038e-06,
      "loss": 0.4307,
      "step": 22661
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.482570767402649,
      "learning_rate": 4.0291099752703055e-06,
      "loss": 0.3683,
      "step": 22662
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.520155906677246,
      "learning_rate": 4.028293062812254e-06,
      "loss": 0.4579,
      "step": 22663
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5615090131759644,
      "learning_rate": 4.027476212290353e-06,
      "loss": 0.3845,
      "step": 22664
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6055352687835693,
      "learning_rate": 4.026659423713071e-06,
      "loss": 0.4383,
      "step": 22665
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.578973650932312,
      "learning_rate": 4.025842697088882e-06,
      "loss": 0.4131,
      "step": 22666
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.582365870475769,
      "learning_rate": 4.025026032426262e-06,
      "loss": 0.448,
      "step": 22667
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5662922859191895,
      "learning_rate": 4.024209429733674e-06,
      "loss": 0.3853,
      "step": 22668
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5287226438522339,
      "learning_rate": 4.023392889019589e-06,
      "loss": 0.4328,
      "step": 22669
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5270229578018188,
      "learning_rate": 4.022576410292475e-06,
      "loss": 0.4193,
      "step": 22670
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.518101692199707,
      "learning_rate": 4.021759993560803e-06,
      "loss": 0.4258,
      "step": 22671
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5287184715270996,
      "learning_rate": 4.020943638833043e-06,
      "loss": 0.3982,
      "step": 22672
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5180526971817017,
      "learning_rate": 4.020127346117659e-06,
      "loss": 0.4276,
      "step": 22673
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5452457666397095,
      "learning_rate": 4.019311115423112e-06,
      "loss": 0.4013,
      "step": 22674
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5745515823364258,
      "learning_rate": 4.018494946757874e-06,
      "loss": 0.4518,
      "step": 22675
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5234285593032837,
      "learning_rate": 4.017678840130411e-06,
      "loss": 0.3895,
      "step": 22676
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5018501281738281,
      "learning_rate": 4.01686279554918e-06,
      "loss": 0.4523,
      "step": 22677
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4861524105072021,
      "learning_rate": 4.016046813022655e-06,
      "loss": 0.3809,
      "step": 22678
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5329701900482178,
      "learning_rate": 4.015230892559289e-06,
      "loss": 0.416,
      "step": 22679
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5261149406433105,
      "learning_rate": 4.014415034167552e-06,
      "loss": 0.3895,
      "step": 22680
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5833170413970947,
      "learning_rate": 4.0135992378558985e-06,
      "loss": 0.4496,
      "step": 22681
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5616689920425415,
      "learning_rate": 4.012783503632794e-06,
      "loss": 0.3794,
      "step": 22682
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.021627426147461,
      "learning_rate": 4.0119678315067025e-06,
      "loss": 0.4154,
      "step": 22683
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6389273405075073,
      "learning_rate": 4.011152221486078e-06,
      "loss": 0.3821,
      "step": 22684
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6201215982437134,
      "learning_rate": 4.0103366735793814e-06,
      "loss": 0.4245,
      "step": 22685
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.520808219909668,
      "learning_rate": 4.009521187795069e-06,
      "loss": 0.3838,
      "step": 22686
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6350642442703247,
      "learning_rate": 4.008705764141603e-06,
      "loss": 0.4504,
      "step": 22687
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5566779375076294,
      "learning_rate": 4.007890402627441e-06,
      "loss": 0.3679,
      "step": 22688
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4688490629196167,
      "learning_rate": 4.007075103261038e-06,
      "loss": 0.4241,
      "step": 22689
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6129242181777954,
      "learning_rate": 4.006259866050845e-06,
      "loss": 0.4128,
      "step": 22690
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4817872047424316,
      "learning_rate": 4.005444691005322e-06,
      "loss": 0.4364,
      "step": 22691
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5880348682403564,
      "learning_rate": 4.004629578132927e-06,
      "loss": 0.3978,
      "step": 22692
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.572446346282959,
      "learning_rate": 4.003814527442107e-06,
      "loss": 0.4047,
      "step": 22693
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.512316107749939,
      "learning_rate": 4.002999538941322e-06,
      "loss": 0.4206,
      "step": 22694
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6420954465866089,
      "learning_rate": 4.002184612639019e-06,
      "loss": 0.4433,
      "step": 22695
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5499132871627808,
      "learning_rate": 4.001369748543654e-06,
      "loss": 0.4056,
      "step": 22696
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4884614944458008,
      "learning_rate": 4.000554946663675e-06,
      "loss": 0.4005,
      "step": 22697
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5695196390151978,
      "learning_rate": 3.999740207007534e-06,
      "loss": 0.3689,
      "step": 22698
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6183903217315674,
      "learning_rate": 3.998925529583686e-06,
      "loss": 0.422,
      "step": 22699
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5079994201660156,
      "learning_rate": 3.998110914400577e-06,
      "loss": 0.3685,
      "step": 22700
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8005812168121338,
      "learning_rate": 3.997296361466651e-06,
      "loss": 0.9393,
      "step": 22701
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.411499261856079,
      "learning_rate": 3.996481870790361e-06,
      "loss": 0.9761,
      "step": 22702
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6180890798568726,
      "learning_rate": 3.9956674423801534e-06,
      "loss": 0.4113,
      "step": 22703
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5563514232635498,
      "learning_rate": 3.9948530762444835e-06,
      "loss": 0.3817,
      "step": 22704
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5921605825424194,
      "learning_rate": 3.994038772391783e-06,
      "loss": 0.4749,
      "step": 22705
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.530085563659668,
      "learning_rate": 3.993224530830504e-06,
      "loss": 0.3869,
      "step": 22706
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5365163087844849,
      "learning_rate": 3.992410351569092e-06,
      "loss": 0.4113,
      "step": 22707
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4892610311508179,
      "learning_rate": 3.991596234615995e-06,
      "loss": 0.3826,
      "step": 22708
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6045535802841187,
      "learning_rate": 3.990782179979653e-06,
      "loss": 0.4399,
      "step": 22709
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4917875528335571,
      "learning_rate": 3.9899681876685045e-06,
      "loss": 0.368,
      "step": 22710
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6973321437835693,
      "learning_rate": 3.989154257690998e-06,
      "loss": 0.428,
      "step": 22711
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5431888103485107,
      "learning_rate": 3.988340390055576e-06,
      "loss": 0.363,
      "step": 22712
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8508714437484741,
      "learning_rate": 3.987526584770675e-06,
      "loss": 0.4681,
      "step": 22713
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5217145681381226,
      "learning_rate": 3.98671284184474e-06,
      "loss": 0.4064,
      "step": 22714
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5809392929077148,
      "learning_rate": 3.985899161286205e-06,
      "loss": 0.4272,
      "step": 22715
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5249584913253784,
      "learning_rate": 3.9850855431035175e-06,
      "loss": 0.378,
      "step": 22716
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5590473413467407,
      "learning_rate": 3.9842719873051075e-06,
      "loss": 0.4229,
      "step": 22717
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5177642107009888,
      "learning_rate": 3.983458493899416e-06,
      "loss": 0.387,
      "step": 22718
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6969966888427734,
      "learning_rate": 3.982645062894884e-06,
      "loss": 0.9246,
      "step": 22719
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4315540790557861,
      "learning_rate": 3.981831694299946e-06,
      "loss": 0.979,
      "step": 22720
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5957483053207397,
      "learning_rate": 3.981018388123032e-06,
      "loss": 0.4627,
      "step": 22721
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5564017295837402,
      "learning_rate": 3.980205144372583e-06,
      "loss": 0.3958,
      "step": 22722
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.549807071685791,
      "learning_rate": 3.979391963057033e-06,
      "loss": 0.4465,
      "step": 22723
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5182074308395386,
      "learning_rate": 3.9785788441848194e-06,
      "loss": 0.3925,
      "step": 22724
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.1334054470062256,
      "learning_rate": 3.977765787764372e-06,
      "loss": 0.4474,
      "step": 22725
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4535859823226929,
      "learning_rate": 3.976952793804118e-06,
      "loss": 0.3991,
      "step": 22726
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6689341068267822,
      "learning_rate": 3.976139862312497e-06,
      "loss": 0.4405,
      "step": 22727
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.553808331489563,
      "learning_rate": 3.975326993297941e-06,
      "loss": 0.4155,
      "step": 22728
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4893321990966797,
      "learning_rate": 3.974514186768875e-06,
      "loss": 0.4285,
      "step": 22729
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.506346344947815,
      "learning_rate": 3.973701442733735e-06,
      "loss": 0.3818,
      "step": 22730
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5342010259628296,
      "learning_rate": 3.972888761200945e-06,
      "loss": 0.4195,
      "step": 22731
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5402312278747559,
      "learning_rate": 3.972076142178939e-06,
      "loss": 0.3741,
      "step": 22732
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7172380685806274,
      "learning_rate": 3.97126358567614e-06,
      "loss": 0.4891,
      "step": 22733
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5015257596969604,
      "learning_rate": 3.970451091700977e-06,
      "loss": 0.3556,
      "step": 22734
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6975347995758057,
      "learning_rate": 3.9696386602618815e-06,
      "loss": 0.9699,
      "step": 22735
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4721853733062744,
      "learning_rate": 3.968826291367277e-06,
      "loss": 1.0442,
      "step": 22736
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7047619819641113,
      "learning_rate": 3.968013985025584e-06,
      "loss": 1.0553,
      "step": 22737
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.453419804573059,
      "learning_rate": 3.9672017412452315e-06,
      "loss": 1.0586,
      "step": 22738
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5485846996307373,
      "learning_rate": 3.966389560034644e-06,
      "loss": 0.4467,
      "step": 22739
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5624104738235474,
      "learning_rate": 3.965577441402252e-06,
      "loss": 0.3696,
      "step": 22740
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7689228057861328,
      "learning_rate": 3.964765385356465e-06,
      "loss": 0.4369,
      "step": 22741
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4697409868240356,
      "learning_rate": 3.963953391905711e-06,
      "loss": 0.3806,
      "step": 22742
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5429712533950806,
      "learning_rate": 3.963141461058413e-06,
      "loss": 0.4249,
      "step": 22743
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6402153968811035,
      "learning_rate": 3.962329592822994e-06,
      "loss": 0.4209,
      "step": 22744
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6401501893997192,
      "learning_rate": 3.961517787207872e-06,
      "loss": 0.4319,
      "step": 22745
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5551682710647583,
      "learning_rate": 3.960706044221465e-06,
      "loss": 0.362,
      "step": 22746
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5889580249786377,
      "learning_rate": 3.959894363872192e-06,
      "loss": 0.4643,
      "step": 22747
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5351768732070923,
      "learning_rate": 3.959082746168476e-06,
      "loss": 0.3634,
      "step": 22748
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6594189405441284,
      "learning_rate": 3.9582711911187295e-06,
      "loss": 0.4358,
      "step": 22749
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5426121950149536,
      "learning_rate": 3.957459698731375e-06,
      "loss": 0.3939,
      "step": 22750
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6399277448654175,
      "learning_rate": 3.956648269014823e-06,
      "loss": 0.4476,
      "step": 22751
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.525940179824829,
      "learning_rate": 3.955836901977496e-06,
      "loss": 0.3759,
      "step": 22752
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.519465684890747,
      "learning_rate": 3.9550255976278e-06,
      "loss": 0.3968,
      "step": 22753
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5915590524673462,
      "learning_rate": 3.954214355974157e-06,
      "loss": 0.4033,
      "step": 22754
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5637201070785522,
      "learning_rate": 3.953403177024982e-06,
      "loss": 0.4393,
      "step": 22755
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5306401252746582,
      "learning_rate": 3.952592060788685e-06,
      "loss": 0.4032,
      "step": 22756
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7371970415115356,
      "learning_rate": 3.951781007273674e-06,
      "loss": 0.3767,
      "step": 22757
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5274642705917358,
      "learning_rate": 3.950970016488367e-06,
      "loss": 0.3825,
      "step": 22758
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5601755380630493,
      "learning_rate": 3.950159088441173e-06,
      "loss": 0.4845,
      "step": 22759
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5353593826293945,
      "learning_rate": 3.949348223140506e-06,
      "loss": 0.3869,
      "step": 22760
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6855473518371582,
      "learning_rate": 3.948537420594776e-06,
      "loss": 0.4565,
      "step": 22761
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.484858751296997,
      "learning_rate": 3.947726680812384e-06,
      "loss": 0.3682,
      "step": 22762
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5838700532913208,
      "learning_rate": 3.9469160038017455e-06,
      "loss": 0.407,
      "step": 22763
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.499634027481079,
      "learning_rate": 3.94610538957127e-06,
      "loss": 0.3692,
      "step": 22764
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4823296070098877,
      "learning_rate": 3.945294838129359e-06,
      "loss": 0.4173,
      "step": 22765
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5239602327346802,
      "learning_rate": 3.944484349484427e-06,
      "loss": 0.3795,
      "step": 22766
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4786086082458496,
      "learning_rate": 3.94367392364487e-06,
      "loss": 0.4169,
      "step": 22767
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4775530099868774,
      "learning_rate": 3.942863560619105e-06,
      "loss": 0.3724,
      "step": 22768
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6057668924331665,
      "learning_rate": 3.942053260415526e-06,
      "loss": 0.4327,
      "step": 22769
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.613595724105835,
      "learning_rate": 3.941243023042542e-06,
      "loss": 0.4102,
      "step": 22770
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4827897548675537,
      "learning_rate": 3.9404328485085596e-06,
      "loss": 0.4272,
      "step": 22771
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5332363843917847,
      "learning_rate": 3.939622736821978e-06,
      "loss": 0.3785,
      "step": 22772
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7768117189407349,
      "learning_rate": 3.938812687991197e-06,
      "loss": 0.4646,
      "step": 22773
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.511069655418396,
      "learning_rate": 3.93800270202462e-06,
      "loss": 0.385,
      "step": 22774
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5823358297348022,
      "learning_rate": 3.937192778930649e-06,
      "loss": 0.4382,
      "step": 22775
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4875516891479492,
      "learning_rate": 3.936382918717687e-06,
      "loss": 0.3703,
      "step": 22776
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.52259361743927,
      "learning_rate": 3.935573121394131e-06,
      "loss": 0.4527,
      "step": 22777
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.559181809425354,
      "learning_rate": 3.934763386968375e-06,
      "loss": 0.3969,
      "step": 22778
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.587483525276184,
      "learning_rate": 3.9339537154488214e-06,
      "loss": 0.4311,
      "step": 22779
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5481377840042114,
      "learning_rate": 3.933144106843871e-06,
      "loss": 0.3813,
      "step": 22780
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5185750722885132,
      "learning_rate": 3.932334561161919e-06,
      "loss": 0.4257,
      "step": 22781
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5808544158935547,
      "learning_rate": 3.9315250784113566e-06,
      "loss": 0.3972,
      "step": 22782
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6287671327590942,
      "learning_rate": 3.930715658600582e-06,
      "loss": 0.4485,
      "step": 22783
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.549015760421753,
      "learning_rate": 3.929906301737995e-06,
      "loss": 0.3974,
      "step": 22784
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5148499011993408,
      "learning_rate": 3.929097007831984e-06,
      "loss": 0.4454,
      "step": 22785
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5502216815948486,
      "learning_rate": 3.928287776890948e-06,
      "loss": 0.3977,
      "step": 22786
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6004670858383179,
      "learning_rate": 3.927478608923273e-06,
      "loss": 0.3908,
      "step": 22787
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5867173671722412,
      "learning_rate": 3.926669503937358e-06,
      "loss": 0.4012,
      "step": 22788
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5707725286483765,
      "learning_rate": 3.925860461941589e-06,
      "loss": 0.4186,
      "step": 22789
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5001695156097412,
      "learning_rate": 3.925051482944361e-06,
      "loss": 0.366,
      "step": 22790
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4854927062988281,
      "learning_rate": 3.924242566954067e-06,
      "loss": 0.3897,
      "step": 22791
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6758334636688232,
      "learning_rate": 3.9234337139790915e-06,
      "loss": 0.4048,
      "step": 22792
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6131978034973145,
      "learning_rate": 3.922624924027822e-06,
      "loss": 0.446,
      "step": 22793
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5042864084243774,
      "learning_rate": 3.921816197108652e-06,
      "loss": 0.3613,
      "step": 22794
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5301426649093628,
      "learning_rate": 3.921007533229967e-06,
      "loss": 0.5003,
      "step": 22795
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5315890312194824,
      "learning_rate": 3.920198932400156e-06,
      "loss": 0.3729,
      "step": 22796
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5584474802017212,
      "learning_rate": 3.919390394627607e-06,
      "loss": 0.415,
      "step": 22797
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5533826351165771,
      "learning_rate": 3.918581919920699e-06,
      "loss": 0.3391,
      "step": 22798
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8941638469696045,
      "learning_rate": 3.91777350828782e-06,
      "loss": 0.4541,
      "step": 22799
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4999144077301025,
      "learning_rate": 3.916965159737359e-06,
      "loss": 0.388,
      "step": 22800
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5788058042526245,
      "learning_rate": 3.916156874277695e-06,
      "loss": 0.4242,
      "step": 22801
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5556697845458984,
      "learning_rate": 3.915348651917215e-06,
      "loss": 0.3887,
      "step": 22802
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5028332471847534,
      "learning_rate": 3.914540492664296e-06,
      "loss": 0.452,
      "step": 22803
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5019114017486572,
      "learning_rate": 3.913732396527327e-06,
      "loss": 0.3796,
      "step": 22804
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7660475969314575,
      "learning_rate": 3.912924363514681e-06,
      "loss": 0.4626,
      "step": 22805
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5576951503753662,
      "learning_rate": 3.912116393634744e-06,
      "loss": 0.4055,
      "step": 22806
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.549593448638916,
      "learning_rate": 3.911308486895898e-06,
      "loss": 0.4307,
      "step": 22807
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4920439720153809,
      "learning_rate": 3.910500643306519e-06,
      "loss": 0.3807,
      "step": 22808
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.549048662185669,
      "learning_rate": 3.909692862874983e-06,
      "loss": 0.4241,
      "step": 22809
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5083465576171875,
      "learning_rate": 3.908885145609671e-06,
      "loss": 0.3816,
      "step": 22810
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5213226079940796,
      "learning_rate": 3.90807749151896e-06,
      "loss": 0.4667,
      "step": 22811
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5154973268508911,
      "learning_rate": 3.907269900611231e-06,
      "loss": 0.376,
      "step": 22812
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4996663331985474,
      "learning_rate": 3.906462372894856e-06,
      "loss": 0.4267,
      "step": 22813
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6580324172973633,
      "learning_rate": 3.905654908378206e-06,
      "loss": 0.4162,
      "step": 22814
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5312490463256836,
      "learning_rate": 3.90484750706966e-06,
      "loss": 0.4549,
      "step": 22815
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.65058171749115,
      "learning_rate": 3.904040168977596e-06,
      "loss": 0.3975,
      "step": 22816
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5672261714935303,
      "learning_rate": 3.903232894110379e-06,
      "loss": 0.391,
      "step": 22817
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.501283884048462,
      "learning_rate": 3.90242568247639e-06,
      "loss": 0.3739,
      "step": 22818
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.574669599533081,
      "learning_rate": 3.901618534083994e-06,
      "loss": 0.4485,
      "step": 22819
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5086578130722046,
      "learning_rate": 3.900811448941569e-06,
      "loss": 0.3882,
      "step": 22820
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5508346557617188,
      "learning_rate": 3.900004427057478e-06,
      "loss": 0.4436,
      "step": 22821
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6569370031356812,
      "learning_rate": 3.8991974684401e-06,
      "loss": 0.3962,
      "step": 22822
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4910380840301514,
      "learning_rate": 3.898390573097795e-06,
      "loss": 0.3921,
      "step": 22823
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4998444318771362,
      "learning_rate": 3.897583741038943e-06,
      "loss": 0.3785,
      "step": 22824
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5589288473129272,
      "learning_rate": 3.8967769722719006e-06,
      "loss": 0.4604,
      "step": 22825
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.489585280418396,
      "learning_rate": 3.89597026680504e-06,
      "loss": 0.3669,
      "step": 22826
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5384591817855835,
      "learning_rate": 3.895163624646733e-06,
      "loss": 0.431,
      "step": 22827
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5104511976242065,
      "learning_rate": 3.894357045805341e-06,
      "loss": 0.3985,
      "step": 22828
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.847599983215332,
      "learning_rate": 3.893550530289226e-06,
      "loss": 0.4856,
      "step": 22829
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.489267110824585,
      "learning_rate": 3.892744078106758e-06,
      "loss": 0.38,
      "step": 22830
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6676807403564453,
      "learning_rate": 3.8919376892662985e-06,
      "loss": 0.4418,
      "step": 22831
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4669256210327148,
      "learning_rate": 3.891131363776217e-06,
      "loss": 0.3779,
      "step": 22832
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6036741733551025,
      "learning_rate": 3.890325101644872e-06,
      "loss": 0.469,
      "step": 22833
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5757533311843872,
      "learning_rate": 3.889518902880621e-06,
      "loss": 0.3529,
      "step": 22834
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7995867729187012,
      "learning_rate": 3.8887127674918305e-06,
      "loss": 0.4306,
      "step": 22835
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5750298500061035,
      "learning_rate": 3.887906695486865e-06,
      "loss": 0.3771,
      "step": 22836
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6907622814178467,
      "learning_rate": 3.887100686874077e-06,
      "loss": 0.4124,
      "step": 22837
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4864003658294678,
      "learning_rate": 3.886294741661832e-06,
      "loss": 0.4052,
      "step": 22838
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5084606409072876,
      "learning_rate": 3.885488859858485e-06,
      "loss": 0.377,
      "step": 22839
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5130138397216797,
      "learning_rate": 3.884683041472399e-06,
      "loss": 0.392,
      "step": 22840
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.610012412071228,
      "learning_rate": 3.883877286511926e-06,
      "loss": 0.4125,
      "step": 22841
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5159978866577148,
      "learning_rate": 3.883071594985426e-06,
      "loss": 0.387,
      "step": 22842
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.537121295928955,
      "learning_rate": 3.882265966901257e-06,
      "loss": 0.4412,
      "step": 22843
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4983243942260742,
      "learning_rate": 3.8814604022677735e-06,
      "loss": 0.3694,
      "step": 22844
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6204818487167358,
      "learning_rate": 3.880654901093326e-06,
      "loss": 0.4696,
      "step": 22845
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5390316247940063,
      "learning_rate": 3.8798494633862725e-06,
      "loss": 0.3704,
      "step": 22846
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5560251474380493,
      "learning_rate": 3.879044089154968e-06,
      "loss": 0.4252,
      "step": 22847
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5392661094665527,
      "learning_rate": 3.878238778407766e-06,
      "loss": 0.3598,
      "step": 22848
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.589290738105774,
      "learning_rate": 3.877433531153017e-06,
      "loss": 0.416,
      "step": 22849
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.547885775566101,
      "learning_rate": 3.876628347399069e-06,
      "loss": 0.4449,
      "step": 22850
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5629371404647827,
      "learning_rate": 3.8758232271542785e-06,
      "loss": 0.4489,
      "step": 22851
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5668845176696777,
      "learning_rate": 3.875018170426996e-06,
      "loss": 0.4064,
      "step": 22852
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5068767070770264,
      "learning_rate": 3.874213177225567e-06,
      "loss": 0.4227,
      "step": 22853
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4780714511871338,
      "learning_rate": 3.873408247558347e-06,
      "loss": 0.3862,
      "step": 22854
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5340771675109863,
      "learning_rate": 3.872603381433676e-06,
      "loss": 0.4272,
      "step": 22855
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5974416732788086,
      "learning_rate": 3.87179857885991e-06,
      "loss": 0.4039,
      "step": 22856
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6514538526535034,
      "learning_rate": 3.870993839845388e-06,
      "loss": 0.4337,
      "step": 22857
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5380910634994507,
      "learning_rate": 3.8701891643984645e-06,
      "loss": 0.3884,
      "step": 22858
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7202763557434082,
      "learning_rate": 3.8693845525274775e-06,
      "loss": 0.4102,
      "step": 22859
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5406465530395508,
      "learning_rate": 3.8685800042407805e-06,
      "loss": 0.4307,
      "step": 22860
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5439276695251465,
      "learning_rate": 3.867775519546709e-06,
      "loss": 0.4214,
      "step": 22861
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.580721139907837,
      "learning_rate": 3.866971098453613e-06,
      "loss": 0.4181,
      "step": 22862
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6412678956985474,
      "learning_rate": 3.866166740969835e-06,
      "loss": 0.4406,
      "step": 22863
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5602107048034668,
      "learning_rate": 3.8653624471037185e-06,
      "loss": 0.3978,
      "step": 22864
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6478773355484009,
      "learning_rate": 3.864558216863599e-06,
      "loss": 0.4631,
      "step": 22865
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5225608348846436,
      "learning_rate": 3.863754050257822e-06,
      "loss": 0.3881,
      "step": 22866
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.510061502456665,
      "learning_rate": 3.8629499472947276e-06,
      "loss": 0.4143,
      "step": 22867
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5941991806030273,
      "learning_rate": 3.862145907982659e-06,
      "loss": 0.3991,
      "step": 22868
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5689946413040161,
      "learning_rate": 3.861341932329953e-06,
      "loss": 0.3863,
      "step": 22869
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4678250551223755,
      "learning_rate": 3.860538020344943e-06,
      "loss": 0.3626,
      "step": 22870
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.512546420097351,
      "learning_rate": 3.859734172035971e-06,
      "loss": 0.403,
      "step": 22871
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.580675721168518,
      "learning_rate": 3.858930387411379e-06,
      "loss": 0.3672,
      "step": 22872
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.530339241027832,
      "learning_rate": 3.8581266664794955e-06,
      "loss": 0.4038,
      "step": 22873
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4819897413253784,
      "learning_rate": 3.857323009248664e-06,
      "loss": 0.3745,
      "step": 22874
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5022863149642944,
      "learning_rate": 3.85651941572721e-06,
      "loss": 0.4313,
      "step": 22875
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5862061977386475,
      "learning_rate": 3.855715885923478e-06,
      "loss": 0.4047,
      "step": 22876
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7220793962478638,
      "learning_rate": 3.854912419845794e-06,
      "loss": 0.4874,
      "step": 22877
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5397670269012451,
      "learning_rate": 3.854109017502494e-06,
      "loss": 0.4237,
      "step": 22878
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5579979419708252,
      "learning_rate": 3.853305678901916e-06,
      "loss": 0.41,
      "step": 22879
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5852277278900146,
      "learning_rate": 3.852502404052387e-06,
      "loss": 0.413,
      "step": 22880
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.728944182395935,
      "learning_rate": 3.851699192962235e-06,
      "loss": 0.9298,
      "step": 22881
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4394862651824951,
      "learning_rate": 3.8508960456397926e-06,
      "loss": 0.9912,
      "step": 22882
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6596794128417969,
      "learning_rate": 3.8500929620933925e-06,
      "loss": 0.4226,
      "step": 22883
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4777337312698364,
      "learning_rate": 3.849289942331366e-06,
      "loss": 0.3807,
      "step": 22884
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5081639289855957,
      "learning_rate": 3.848486986362038e-06,
      "loss": 0.4558,
      "step": 22885
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6307415962219238,
      "learning_rate": 3.847684094193733e-06,
      "loss": 0.4259,
      "step": 22886
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6153783798217773,
      "learning_rate": 3.846881265834782e-06,
      "loss": 0.45,
      "step": 22887
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5753376483917236,
      "learning_rate": 3.846078501293512e-06,
      "loss": 0.4036,
      "step": 22888
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8073716163635254,
      "learning_rate": 3.845275800578252e-06,
      "loss": 0.4688,
      "step": 22889
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5716019868850708,
      "learning_rate": 3.844473163697323e-06,
      "loss": 0.4063,
      "step": 22890
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6217223405838013,
      "learning_rate": 3.843670590659048e-06,
      "loss": 0.4293,
      "step": 22891
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.496789574623108,
      "learning_rate": 3.842868081471753e-06,
      "loss": 0.3695,
      "step": 22892
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5602948665618896,
      "learning_rate": 3.8420656361437655e-06,
      "loss": 0.4208,
      "step": 22893
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4833447933197021,
      "learning_rate": 3.8412632546834004e-06,
      "loss": 0.3772,
      "step": 22894
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5711557865142822,
      "learning_rate": 3.840460937098988e-06,
      "loss": 0.4232,
      "step": 22895
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4867668151855469,
      "learning_rate": 3.8396586833988416e-06,
      "loss": 0.3661,
      "step": 22896
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5589812994003296,
      "learning_rate": 3.8388564935912885e-06,
      "loss": 0.4502,
      "step": 22897
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5311003923416138,
      "learning_rate": 3.838054367684643e-06,
      "loss": 0.3817,
      "step": 22898
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5094945430755615,
      "learning_rate": 3.837252305687229e-06,
      "loss": 0.4463,
      "step": 22899
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.506819725036621,
      "learning_rate": 3.8364503076073615e-06,
      "loss": 0.3541,
      "step": 22900
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5684924125671387,
      "learning_rate": 3.835648373453364e-06,
      "loss": 0.4425,
      "step": 22901
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4936189651489258,
      "learning_rate": 3.834846503233545e-06,
      "loss": 0.3876,
      "step": 22902
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7426538467407227,
      "learning_rate": 3.834044696956227e-06,
      "loss": 1.0436,
      "step": 22903
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4326471090316772,
      "learning_rate": 3.833242954629729e-06,
      "loss": 0.9971,
      "step": 22904
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4964535236358643,
      "learning_rate": 3.832441276262363e-06,
      "loss": 0.4198,
      "step": 22905
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4885915517807007,
      "learning_rate": 3.831639661862439e-06,
      "loss": 0.3657,
      "step": 22906
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6055834293365479,
      "learning_rate": 3.830838111438277e-06,
      "loss": 0.4759,
      "step": 22907
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4738743305206299,
      "learning_rate": 3.830036624998188e-06,
      "loss": 0.368,
      "step": 22908
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5487055778503418,
      "learning_rate": 3.8292352025504885e-06,
      "loss": 0.4208,
      "step": 22909
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5596928596496582,
      "learning_rate": 3.828433844103489e-06,
      "loss": 0.4127,
      "step": 22910
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5811916589736938,
      "learning_rate": 3.827632549665494e-06,
      "loss": 0.3934,
      "step": 22911
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.449999213218689,
      "learning_rate": 3.826831319244821e-06,
      "loss": 0.3584,
      "step": 22912
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.575026035308838,
      "learning_rate": 3.826030152849781e-06,
      "loss": 0.4544,
      "step": 22913
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5464893579483032,
      "learning_rate": 3.825229050488678e-06,
      "loss": 0.3768,
      "step": 22914
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6777093410491943,
      "learning_rate": 3.824428012169827e-06,
      "loss": 0.4918,
      "step": 22915
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8862360715866089,
      "learning_rate": 3.823627037901529e-06,
      "loss": 0.3776,
      "step": 22916
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.540065050125122,
      "learning_rate": 3.8228261276921e-06,
      "loss": 0.4386,
      "step": 22917
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4988484382629395,
      "learning_rate": 3.822025281549837e-06,
      "loss": 0.3812,
      "step": 22918
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5188162326812744,
      "learning_rate": 3.821224499483052e-06,
      "loss": 0.4508,
      "step": 22919
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4745503664016724,
      "learning_rate": 3.8204237815000515e-06,
      "loss": 0.369,
      "step": 22920
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5496522188186646,
      "learning_rate": 3.819623127609139e-06,
      "loss": 0.4065,
      "step": 22921
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5695240497589111,
      "learning_rate": 3.818822537818613e-06,
      "loss": 0.3824,
      "step": 22922
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5948911905288696,
      "learning_rate": 3.818022012136783e-06,
      "loss": 0.4556,
      "step": 22923
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.528038740158081,
      "learning_rate": 3.817221550571949e-06,
      "loss": 0.3894,
      "step": 22924
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4548969268798828,
      "learning_rate": 3.816421153132418e-06,
      "loss": 0.3896,
      "step": 22925
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.452207088470459,
      "learning_rate": 3.815620819826488e-06,
      "loss": 0.3764,
      "step": 22926
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5692819356918335,
      "learning_rate": 3.814820550662456e-06,
      "loss": 0.4666,
      "step": 22927
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.554188847541809,
      "learning_rate": 3.8140203456486246e-06,
      "loss": 0.3866,
      "step": 22928
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4948402643203735,
      "learning_rate": 3.8132202047932987e-06,
      "loss": 0.4169,
      "step": 22929
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5493892431259155,
      "learning_rate": 3.812420128104769e-06,
      "loss": 0.3852,
      "step": 22930
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6480633020401,
      "learning_rate": 3.8116201155913403e-06,
      "loss": 0.4218,
      "step": 22931
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.508271336555481,
      "learning_rate": 3.810820167261302e-06,
      "loss": 0.3575,
      "step": 22932
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5080010890960693,
      "learning_rate": 3.8100202831229603e-06,
      "loss": 0.4588,
      "step": 22933
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4560425281524658,
      "learning_rate": 3.809220463184603e-06,
      "loss": 0.3921,
      "step": 22934
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5126986503601074,
      "learning_rate": 3.8084207074545285e-06,
      "loss": 0.462,
      "step": 22935
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4905474185943604,
      "learning_rate": 3.8076210159410365e-06,
      "loss": 0.3904,
      "step": 22936
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7497954368591309,
      "learning_rate": 3.8068213886524163e-06,
      "loss": 0.9976,
      "step": 22937
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4449985027313232,
      "learning_rate": 3.8060218255969574e-06,
      "loss": 1.0307,
      "step": 22938
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5145374536514282,
      "learning_rate": 3.805222326782958e-06,
      "loss": 0.4305,
      "step": 22939
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4967398643493652,
      "learning_rate": 3.804422892218712e-06,
      "loss": 0.378,
      "step": 22940
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6475603580474854,
      "learning_rate": 3.8036235219125073e-06,
      "loss": 0.4292,
      "step": 22941
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5934758186340332,
      "learning_rate": 3.8028242158726325e-06,
      "loss": 0.3958,
      "step": 22942
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5319461822509766,
      "learning_rate": 3.80202497410738e-06,
      "loss": 0.4146,
      "step": 22943
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5424787998199463,
      "learning_rate": 3.8012257966250397e-06,
      "loss": 0.3781,
      "step": 22944
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.592678189277649,
      "learning_rate": 3.8004266834339045e-06,
      "loss": 0.4681,
      "step": 22945
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5392136573791504,
      "learning_rate": 3.7996276345422566e-06,
      "loss": 0.3946,
      "step": 22946
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4267181158065796,
      "learning_rate": 3.7988286499583826e-06,
      "loss": 0.3931,
      "step": 22947
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5064830780029297,
      "learning_rate": 3.798029729690571e-06,
      "loss": 0.3775,
      "step": 22948
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5696734189987183,
      "learning_rate": 3.797230873747113e-06,
      "loss": 0.4233,
      "step": 22949
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4809141159057617,
      "learning_rate": 3.7964320821362854e-06,
      "loss": 0.394,
      "step": 22950
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6126948595046997,
      "learning_rate": 3.7956333548663802e-06,
      "loss": 0.4179,
      "step": 22951
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.630550742149353,
      "learning_rate": 3.7948346919456757e-06,
      "loss": 0.4004,
      "step": 22952
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5546889305114746,
      "learning_rate": 3.7940360933824616e-06,
      "loss": 0.4023,
      "step": 22953
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4593067169189453,
      "learning_rate": 3.7932375591850135e-06,
      "loss": 0.3834,
      "step": 22954
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5354750156402588,
      "learning_rate": 3.792439089361617e-06,
      "loss": 0.459,
      "step": 22955
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4912011623382568,
      "learning_rate": 3.791640683920557e-06,
      "loss": 0.3978,
      "step": 22956
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5850155353546143,
      "learning_rate": 3.79084234287011e-06,
      "loss": 0.4584,
      "step": 22957
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.46687650680542,
      "learning_rate": 3.7900440662185532e-06,
      "loss": 0.395,
      "step": 22958
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5902351140975952,
      "learning_rate": 3.7892458539741708e-06,
      "loss": 0.4652,
      "step": 22959
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4972622394561768,
      "learning_rate": 3.7884477061452395e-06,
      "loss": 0.3936,
      "step": 22960
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6991971731185913,
      "learning_rate": 3.7876496227400427e-06,
      "loss": 0.4795,
      "step": 22961
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4809232950210571,
      "learning_rate": 3.7868516037668537e-06,
      "loss": 0.395,
      "step": 22962
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5209660530090332,
      "learning_rate": 3.786053649233944e-06,
      "loss": 0.3886,
      "step": 22963
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5718142986297607,
      "learning_rate": 3.7852557591495955e-06,
      "loss": 0.3842,
      "step": 22964
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5972462892532349,
      "learning_rate": 3.7844579335220854e-06,
      "loss": 0.4531,
      "step": 22965
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4632614850997925,
      "learning_rate": 3.783660172359682e-06,
      "loss": 0.3598,
      "step": 22966
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.562394380569458,
      "learning_rate": 3.7828624756706677e-06,
      "loss": 0.4421,
      "step": 22967
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.546828031539917,
      "learning_rate": 3.782064843463307e-06,
      "loss": 0.3721,
      "step": 22968
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5241514444351196,
      "learning_rate": 3.78126727574588e-06,
      "loss": 0.4196,
      "step": 22969
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5863336324691772,
      "learning_rate": 3.780469772526654e-06,
      "loss": 0.3722,
      "step": 22970
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5713127851486206,
      "learning_rate": 3.7796723338138995e-06,
      "loss": 0.4345,
      "step": 22971
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5154392719268799,
      "learning_rate": 3.778874959615895e-06,
      "loss": 0.3918,
      "step": 22972
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4635305404663086,
      "learning_rate": 3.7780776499409045e-06,
      "loss": 0.4012,
      "step": 22973
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4448997974395752,
      "learning_rate": 3.7772804047971947e-06,
      "loss": 0.3674,
      "step": 22974
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6003751754760742,
      "learning_rate": 3.776483224193037e-06,
      "loss": 0.4621,
      "step": 22975
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4670660495758057,
      "learning_rate": 3.7756861081367045e-06,
      "loss": 0.374,
      "step": 22976
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.575125813484192,
      "learning_rate": 3.7748890566364603e-06,
      "loss": 0.4714,
      "step": 22977
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6399458646774292,
      "learning_rate": 3.7740920697005677e-06,
      "loss": 0.4132,
      "step": 22978
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6163609027862549,
      "learning_rate": 3.773295147337296e-06,
      "loss": 0.4567,
      "step": 22979
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.512125849723816,
      "learning_rate": 3.77249828955491e-06,
      "loss": 0.3733,
      "step": 22980
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5647647380828857,
      "learning_rate": 3.7717014963616794e-06,
      "loss": 0.444,
      "step": 22981
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4897434711456299,
      "learning_rate": 3.770904767765863e-06,
      "loss": 0.3627,
      "step": 22982
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5512856245040894,
      "learning_rate": 3.7701081037757216e-06,
      "loss": 0.4687,
      "step": 22983
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5418286323547363,
      "learning_rate": 3.769311504399522e-06,
      "loss": 0.399,
      "step": 22984
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6777957677841187,
      "learning_rate": 3.768514969645528e-06,
      "loss": 0.4422,
      "step": 22985
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4795266389846802,
      "learning_rate": 3.767718499521995e-06,
      "loss": 0.366,
      "step": 22986
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.622115969657898,
      "learning_rate": 3.766922094037192e-06,
      "loss": 0.4805,
      "step": 22987
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4869298934936523,
      "learning_rate": 3.766125753199369e-06,
      "loss": 0.3722,
      "step": 22988
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5664182901382446,
      "learning_rate": 3.765329477016795e-06,
      "loss": 0.4464,
      "step": 22989
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5554163455963135,
      "learning_rate": 3.7645332654977195e-06,
      "loss": 0.4426,
      "step": 22990
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5813112258911133,
      "learning_rate": 3.7637371186504055e-06,
      "loss": 0.4309,
      "step": 22991
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5939388275146484,
      "learning_rate": 3.762941036483113e-06,
      "loss": 0.3905,
      "step": 22992
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.492038607597351,
      "learning_rate": 3.7621450190040966e-06,
      "loss": 0.4606,
      "step": 22993
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5392217636108398,
      "learning_rate": 3.761349066221607e-06,
      "loss": 0.4262,
      "step": 22994
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9958752393722534,
      "learning_rate": 3.7605531781439044e-06,
      "loss": 1.039,
      "step": 22995
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.433510184288025,
      "learning_rate": 3.759757354779241e-06,
      "loss": 0.9405,
      "step": 22996
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7226792573928833,
      "learning_rate": 3.758961596135877e-06,
      "loss": 0.9822,
      "step": 22997
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.427190899848938,
      "learning_rate": 3.7581659022220616e-06,
      "loss": 1.0163,
      "step": 22998
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5875033140182495,
      "learning_rate": 3.7573702730460425e-06,
      "loss": 0.4136,
      "step": 22999
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6034345626831055,
      "learning_rate": 3.7565747086160775e-06,
      "loss": 0.3739,
      "step": 23000
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5633049011230469,
      "learning_rate": 3.755779208940419e-06,
      "loss": 0.4223,
      "step": 23001
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5211355686187744,
      "learning_rate": 3.754983774027312e-06,
      "loss": 0.3714,
      "step": 23002
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.57647705078125,
      "learning_rate": 3.754188403885013e-06,
      "loss": 0.4577,
      "step": 23003
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.469298243522644,
      "learning_rate": 3.7533930985217648e-06,
      "loss": 0.3635,
      "step": 23004
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6130025386810303,
      "learning_rate": 3.752597857945821e-06,
      "loss": 0.4501,
      "step": 23005
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5195674896240234,
      "learning_rate": 3.751802682165425e-06,
      "loss": 0.3686,
      "step": 23006
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.696319341659546,
      "learning_rate": 3.751007571188827e-06,
      "loss": 0.4566,
      "step": 23007
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5949207544326782,
      "learning_rate": 3.750212525024276e-06,
      "loss": 0.4199,
      "step": 23008
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5956647396087646,
      "learning_rate": 3.7494175436800153e-06,
      "loss": 0.4237,
      "step": 23009
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5195927619934082,
      "learning_rate": 3.7486226271642867e-06,
      "loss": 0.3895,
      "step": 23010
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5745078325271606,
      "learning_rate": 3.7478277754853363e-06,
      "loss": 0.4916,
      "step": 23011
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5254015922546387,
      "learning_rate": 3.7470329886514112e-06,
      "loss": 0.4057,
      "step": 23012
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5751368999481201,
      "learning_rate": 3.74623826667076e-06,
      "loss": 0.398,
      "step": 23013
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6103911399841309,
      "learning_rate": 3.745443609551611e-06,
      "loss": 0.4262,
      "step": 23014
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5312433242797852,
      "learning_rate": 3.7446490173022132e-06,
      "loss": 0.4553,
      "step": 23015
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5480071306228638,
      "learning_rate": 3.7438544899308073e-06,
      "loss": 0.3786,
      "step": 23016
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6812994480133057,
      "learning_rate": 3.7430600274456386e-06,
      "loss": 0.4593,
      "step": 23017
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5794780254364014,
      "learning_rate": 3.742265629854942e-06,
      "loss": 0.3586,
      "step": 23018
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.584275484085083,
      "learning_rate": 3.7414712971669553e-06,
      "loss": 0.4493,
      "step": 23019
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5097596645355225,
      "learning_rate": 3.7406770293899185e-06,
      "loss": 0.3745,
      "step": 23020
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4951517581939697,
      "learning_rate": 3.739882826532073e-06,
      "loss": 0.4395,
      "step": 23021
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4712990522384644,
      "learning_rate": 3.7390886886016496e-06,
      "loss": 0.3712,
      "step": 23022
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6209588050842285,
      "learning_rate": 3.7382946156068922e-06,
      "loss": 0.4069,
      "step": 23023
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5629527568817139,
      "learning_rate": 3.7375006075560293e-06,
      "loss": 0.3787,
      "step": 23024
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.548353672027588,
      "learning_rate": 3.7367066644573023e-06,
      "loss": 0.4195,
      "step": 23025
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5198078155517578,
      "learning_rate": 3.73591278631894e-06,
      "loss": 0.3668,
      "step": 23026
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4787088632583618,
      "learning_rate": 3.7351189731491787e-06,
      "loss": 0.4377,
      "step": 23027
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.480920433998108,
      "learning_rate": 3.7343252249562545e-06,
      "loss": 0.3725,
      "step": 23028
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5316466093063354,
      "learning_rate": 3.7335315417483975e-06,
      "loss": 0.4075,
      "step": 23029
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5514352321624756,
      "learning_rate": 3.7327379235338356e-06,
      "loss": 0.3975,
      "step": 23030
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4789834022521973,
      "learning_rate": 3.7319443703208023e-06,
      "loss": 0.431,
      "step": 23031
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5004124641418457,
      "learning_rate": 3.7311508821175303e-06,
      "loss": 0.384,
      "step": 23032
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.541260838508606,
      "learning_rate": 3.730357458932251e-06,
      "loss": 0.4184,
      "step": 23033
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.516457438468933,
      "learning_rate": 3.7295641007731908e-06,
      "loss": 0.3794,
      "step": 23034
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.527508020401001,
      "learning_rate": 3.728770807648574e-06,
      "loss": 0.4298,
      "step": 23035
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5092557668685913,
      "learning_rate": 3.727977579566632e-06,
      "loss": 0.376,
      "step": 23036
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8530956506729126,
      "learning_rate": 3.7271844165355953e-06,
      "loss": 0.4166,
      "step": 23037
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5096492767333984,
      "learning_rate": 3.726391318563685e-06,
      "loss": 0.3638,
      "step": 23038
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5125010013580322,
      "learning_rate": 3.72559828565913e-06,
      "loss": 0.4774,
      "step": 23039
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.484834909439087,
      "learning_rate": 3.7248053178301524e-06,
      "loss": 0.3718,
      "step": 23040
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5597771406173706,
      "learning_rate": 3.72401241508498e-06,
      "loss": 0.4244,
      "step": 23041
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.530521273612976,
      "learning_rate": 3.7232195774318326e-06,
      "loss": 0.3836,
      "step": 23042
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5458474159240723,
      "learning_rate": 3.7224268048789347e-06,
      "loss": 0.4036,
      "step": 23043
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5087448358535767,
      "learning_rate": 3.7216340974345122e-06,
      "loss": 0.3701,
      "step": 23044
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4779795408248901,
      "learning_rate": 3.720841455106785e-06,
      "loss": 0.4155,
      "step": 23045
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6176345348358154,
      "learning_rate": 3.7200488779039677e-06,
      "loss": 0.4209,
      "step": 23046
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6032836437225342,
      "learning_rate": 3.7192563658342873e-06,
      "loss": 0.4701,
      "step": 23047
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5505024194717407,
      "learning_rate": 3.7184639189059613e-06,
      "loss": 0.3733,
      "step": 23048
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5719988346099854,
      "learning_rate": 3.7176715371272154e-06,
      "loss": 0.5001,
      "step": 23049
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6524540185928345,
      "learning_rate": 3.7168792205062554e-06,
      "loss": 0.3766,
      "step": 23050
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5113383531570435,
      "learning_rate": 3.7160869690513047e-06,
      "loss": 0.4417,
      "step": 23051
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.51359224319458,
      "learning_rate": 3.715294782770581e-06,
      "loss": 0.3885,
      "step": 23052
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7154481410980225,
      "learning_rate": 3.7145026616723023e-06,
      "loss": 0.4536,
      "step": 23053
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6024433374404907,
      "learning_rate": 3.713710605764683e-06,
      "loss": 0.3473,
      "step": 23054
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.527764081954956,
      "learning_rate": 3.7129186150559338e-06,
      "loss": 0.4439,
      "step": 23055
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5691848993301392,
      "learning_rate": 3.712126689554272e-06,
      "loss": 0.4353,
      "step": 23056
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5223137140274048,
      "learning_rate": 3.7113348292679152e-06,
      "loss": 0.4063,
      "step": 23057
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6976690292358398,
      "learning_rate": 3.7105430342050687e-06,
      "loss": 0.4,
      "step": 23058
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.517066240310669,
      "learning_rate": 3.7097513043739506e-06,
      "loss": 0.4119,
      "step": 23059
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5224510431289673,
      "learning_rate": 3.7089596397827675e-06,
      "loss": 0.3987,
      "step": 23060
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5556318759918213,
      "learning_rate": 3.708168040439736e-06,
      "loss": 0.478,
      "step": 23061
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5708645582199097,
      "learning_rate": 3.70737650635306e-06,
      "loss": 0.3933,
      "step": 23062
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6840134859085083,
      "learning_rate": 3.706585037530952e-06,
      "loss": 0.4621,
      "step": 23063
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4976080656051636,
      "learning_rate": 3.7057936339816238e-06,
      "loss": 0.369,
      "step": 23064
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5927188396453857,
      "learning_rate": 3.70500229571328e-06,
      "loss": 0.4176,
      "step": 23065
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5445575714111328,
      "learning_rate": 3.7042110227341264e-06,
      "loss": 0.3979,
      "step": 23066
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.625777244567871,
      "learning_rate": 3.7034198150523716e-06,
      "loss": 0.4376,
      "step": 23067
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5238946676254272,
      "learning_rate": 3.7026286726762217e-06,
      "loss": 0.3656,
      "step": 23068
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5093320608139038,
      "learning_rate": 3.7018375956138863e-06,
      "loss": 0.4099,
      "step": 23069
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5041409730911255,
      "learning_rate": 3.7010465838735666e-06,
      "loss": 0.3919,
      "step": 23070
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5657495260238647,
      "learning_rate": 3.7002556374634624e-06,
      "loss": 0.4255,
      "step": 23071
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4983619451522827,
      "learning_rate": 3.699464756391782e-06,
      "loss": 0.3586,
      "step": 23072
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5679477453231812,
      "learning_rate": 3.6986739406667295e-06,
      "loss": 0.4163,
      "step": 23073
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5578081607818604,
      "learning_rate": 3.6978831902965005e-06,
      "loss": 0.4101,
      "step": 23074
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5124585628509521,
      "learning_rate": 3.6970925052893058e-06,
      "loss": 0.4117,
      "step": 23075
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4946000576019287,
      "learning_rate": 3.696301885653336e-06,
      "loss": 0.3749,
      "step": 23076
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4955434799194336,
      "learning_rate": 3.6955113313967995e-06,
      "loss": 0.4368,
      "step": 23077
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4397438764572144,
      "learning_rate": 3.694720842527888e-06,
      "loss": 0.3768,
      "step": 23078
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5055826902389526,
      "learning_rate": 3.693930419054804e-06,
      "loss": 0.4293,
      "step": 23079
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5280975103378296,
      "learning_rate": 3.6931400609857506e-06,
      "loss": 0.4131,
      "step": 23080
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.516788363456726,
      "learning_rate": 3.6923497683289176e-06,
      "loss": 0.4282,
      "step": 23081
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.583969235420227,
      "learning_rate": 3.691559541092502e-06,
      "loss": 0.398,
      "step": 23082
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5241044759750366,
      "learning_rate": 3.690769379284701e-06,
      "loss": 0.3882,
      "step": 23083
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4736757278442383,
      "learning_rate": 3.6899792829137104e-06,
      "loss": 0.3744,
      "step": 23084
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7052903175354004,
      "learning_rate": 3.689189251987729e-06,
      "loss": 0.4635,
      "step": 23085
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.442929744720459,
      "learning_rate": 3.6883992865149463e-06,
      "loss": 0.3626,
      "step": 23086
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5713138580322266,
      "learning_rate": 3.6876093865035522e-06,
      "loss": 0.4498,
      "step": 23087
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5060476064682007,
      "learning_rate": 3.6868195519617423e-06,
      "loss": 0.3758,
      "step": 23088
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6143525838851929,
      "learning_rate": 3.6860297828977132e-06,
      "loss": 0.5119,
      "step": 23089
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5016013383865356,
      "learning_rate": 3.6852400793196517e-06,
      "loss": 0.3677,
      "step": 23090
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5396761894226074,
      "learning_rate": 3.684450441235744e-06,
      "loss": 0.4229,
      "step": 23091
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5021971464157104,
      "learning_rate": 3.683660868654185e-06,
      "loss": 0.401,
      "step": 23092
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6098207235336304,
      "learning_rate": 3.6828713615831666e-06,
      "loss": 0.4834,
      "step": 23093
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6667633056640625,
      "learning_rate": 3.682081920030871e-06,
      "loss": 0.393,
      "step": 23094
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6552400588989258,
      "learning_rate": 3.6812925440054916e-06,
      "loss": 0.4662,
      "step": 23095
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.562299132347107,
      "learning_rate": 3.680503233515209e-06,
      "loss": 0.4446,
      "step": 23096
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6088719367980957,
      "learning_rate": 3.6797139885682164e-06,
      "loss": 0.4385,
      "step": 23097
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.3924899101257324,
      "learning_rate": 3.6789248091726937e-06,
      "loss": 0.4012,
      "step": 23098
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.561955213546753,
      "learning_rate": 3.6781356953368286e-06,
      "loss": 0.4675,
      "step": 23099
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5629597902297974,
      "learning_rate": 3.677346647068808e-06,
      "loss": 0.3843,
      "step": 23100
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5140068531036377,
      "learning_rate": 3.6765576643768143e-06,
      "loss": 0.429,
      "step": 23101
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4799857139587402,
      "learning_rate": 3.675768747269025e-06,
      "loss": 0.37,
      "step": 23102
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5802053213119507,
      "learning_rate": 3.6749798957536276e-06,
      "loss": 0.4677,
      "step": 23103
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5997586250305176,
      "learning_rate": 3.6741911098388016e-06,
      "loss": 0.4013,
      "step": 23104
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6086759567260742,
      "learning_rate": 3.673402389532734e-06,
      "loss": 0.4139,
      "step": 23105
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5165114402770996,
      "learning_rate": 3.6726137348435998e-06,
      "loss": 0.3925,
      "step": 23106
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5012019872665405,
      "learning_rate": 3.6718251457795763e-06,
      "loss": 0.4055,
      "step": 23107
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5300718545913696,
      "learning_rate": 3.6710366223488448e-06,
      "loss": 0.4263,
      "step": 23108
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6371042728424072,
      "learning_rate": 3.6702481645595866e-06,
      "loss": 0.4638,
      "step": 23109
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5167121887207031,
      "learning_rate": 3.669459772419974e-06,
      "loss": 0.3738,
      "step": 23110
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.636021375656128,
      "learning_rate": 3.6686714459381913e-06,
      "loss": 0.4456,
      "step": 23111
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4598073959350586,
      "learning_rate": 3.667883185122405e-06,
      "loss": 0.3633,
      "step": 23112
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.548616886138916,
      "learning_rate": 3.6670949899807993e-06,
      "loss": 0.4796,
      "step": 23113
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5037318468093872,
      "learning_rate": 3.6663068605215426e-06,
      "loss": 0.3752,
      "step": 23114
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.533596396446228,
      "learning_rate": 3.6655187967528114e-06,
      "loss": 0.4154,
      "step": 23115
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4696462154388428,
      "learning_rate": 3.664730798682783e-06,
      "loss": 0.3707,
      "step": 23116
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5359050035476685,
      "learning_rate": 3.6639428663196262e-06,
      "loss": 0.455,
      "step": 23117
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6096775531768799,
      "learning_rate": 3.6631549996715123e-06,
      "loss": 0.3827,
      "step": 23118
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5771739482879639,
      "learning_rate": 3.6623671987466126e-06,
      "loss": 0.4328,
      "step": 23119
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5459682941436768,
      "learning_rate": 3.6615794635530998e-06,
      "loss": 0.4129,
      "step": 23120
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5087027549743652,
      "learning_rate": 3.6607917940991467e-06,
      "loss": 0.4527,
      "step": 23121
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5170209407806396,
      "learning_rate": 3.6600041903929194e-06,
      "loss": 0.4207,
      "step": 23122
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.618019461631775,
      "learning_rate": 3.6592166524425843e-06,
      "loss": 0.4684,
      "step": 23123
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4907293319702148,
      "learning_rate": 3.658429180256312e-06,
      "loss": 0.3825,
      "step": 23124
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5663987398147583,
      "learning_rate": 3.657641773842272e-06,
      "loss": 0.4645,
      "step": 23125
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6135660409927368,
      "learning_rate": 3.6568544332086296e-06,
      "loss": 0.4282,
      "step": 23126
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6030322313308716,
      "learning_rate": 3.6560671583635467e-06,
      "loss": 0.3889,
      "step": 23127
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.497290015220642,
      "learning_rate": 3.6552799493151915e-06,
      "loss": 0.4037,
      "step": 23128
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5982791185379028,
      "learning_rate": 3.6544928060717287e-06,
      "loss": 0.448,
      "step": 23129
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4877674579620361,
      "learning_rate": 3.653705728641327e-06,
      "loss": 0.3915,
      "step": 23130
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5235756635665894,
      "learning_rate": 3.6529187170321447e-06,
      "loss": 0.4575,
      "step": 23131
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.487090826034546,
      "learning_rate": 3.6521317712523407e-06,
      "loss": 0.3916,
      "step": 23132
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6034648418426514,
      "learning_rate": 3.6513448913100814e-06,
      "loss": 0.4947,
      "step": 23133
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.517472505569458,
      "learning_rate": 3.6505580772135308e-06,
      "loss": 0.3818,
      "step": 23134
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5472776889801025,
      "learning_rate": 3.6497713289708424e-06,
      "loss": 0.4167,
      "step": 23135
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.519740343093872,
      "learning_rate": 3.6489846465901835e-06,
      "loss": 0.4071,
      "step": 23136
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.528827428817749,
      "learning_rate": 3.648198030079707e-06,
      "loss": 0.415,
      "step": 23137
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5389745235443115,
      "learning_rate": 3.6474114794475767e-06,
      "loss": 0.411,
      "step": 23138
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4990177154541016,
      "learning_rate": 3.646624994701944e-06,
      "loss": 0.4331,
      "step": 23139
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5088433027267456,
      "learning_rate": 3.6458385758509705e-06,
      "loss": 0.3751,
      "step": 23140
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6198334693908691,
      "learning_rate": 3.6450522229028153e-06,
      "loss": 0.5394,
      "step": 23141
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.465235710144043,
      "learning_rate": 3.6442659358656297e-06,
      "loss": 0.3746,
      "step": 23142
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6273049116134644,
      "learning_rate": 3.6434797147475665e-06,
      "loss": 0.4726,
      "step": 23143
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4803837537765503,
      "learning_rate": 3.6426935595567835e-06,
      "loss": 0.396,
      "step": 23144
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8079992532730103,
      "learning_rate": 3.6419074703014336e-06,
      "loss": 0.4284,
      "step": 23145
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4878170490264893,
      "learning_rate": 3.641121446989675e-06,
      "loss": 0.3895,
      "step": 23146
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6676939725875854,
      "learning_rate": 3.640335489629654e-06,
      "loss": 0.4247,
      "step": 23147
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5341426134109497,
      "learning_rate": 3.6395495982295203e-06,
      "loss": 0.4024,
      "step": 23148
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6010369062423706,
      "learning_rate": 3.638763772797429e-06,
      "loss": 0.4269,
      "step": 23149
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5026029348373413,
      "learning_rate": 3.637978013341532e-06,
      "loss": 0.3614,
      "step": 23150
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5363550186157227,
      "learning_rate": 3.6371923198699743e-06,
      "loss": 0.4822,
      "step": 23151
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5463682413101196,
      "learning_rate": 3.6364066923909093e-06,
      "loss": 0.3808,
      "step": 23152
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5966752767562866,
      "learning_rate": 3.635621130912479e-06,
      "loss": 0.426,
      "step": 23153
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.69987154006958,
      "learning_rate": 3.6348356354428395e-06,
      "loss": 0.4152,
      "step": 23154
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5848376750946045,
      "learning_rate": 3.6340502059901283e-06,
      "loss": 0.4844,
      "step": 23155
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.566531777381897,
      "learning_rate": 3.633264842562496e-06,
      "loss": 0.3899,
      "step": 23156
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5104697942733765,
      "learning_rate": 3.6324795451680926e-06,
      "loss": 0.4562,
      "step": 23157
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5160928964614868,
      "learning_rate": 3.6316943138150586e-06,
      "loss": 0.3799,
      "step": 23158
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5332106351852417,
      "learning_rate": 3.6309091485115333e-06,
      "loss": 0.4082,
      "step": 23159
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.476586937904358,
      "learning_rate": 3.630124049265665e-06,
      "loss": 0.3546,
      "step": 23160
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.593555212020874,
      "learning_rate": 3.6293390160855967e-06,
      "loss": 0.4315,
      "step": 23161
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4598705768585205,
      "learning_rate": 3.6285540489794746e-06,
      "loss": 0.3786,
      "step": 23162
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4935462474822998,
      "learning_rate": 3.6277691479554335e-06,
      "loss": 0.4285,
      "step": 23163
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5189889669418335,
      "learning_rate": 3.626984313021614e-06,
      "loss": 0.4027,
      "step": 23164
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5355883836746216,
      "learning_rate": 3.6261995441861577e-06,
      "loss": 0.4507,
      "step": 23165
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5171109437942505,
      "learning_rate": 3.6254148414572076e-06,
      "loss": 0.3721,
      "step": 23166
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5286344289779663,
      "learning_rate": 3.6246302048428993e-06,
      "loss": 0.4057,
      "step": 23167
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.515598177909851,
      "learning_rate": 3.623845634351367e-06,
      "loss": 0.3851,
      "step": 23168
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6312824487686157,
      "learning_rate": 3.6230611299907516e-06,
      "loss": 0.4367,
      "step": 23169
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5402073860168457,
      "learning_rate": 3.622276691769192e-06,
      "loss": 0.3883,
      "step": 23170
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7449586391448975,
      "learning_rate": 3.621492319694819e-06,
      "loss": 0.4489,
      "step": 23171
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4548792839050293,
      "learning_rate": 3.6207080137757743e-06,
      "loss": 0.3695,
      "step": 23172
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8838109970092773,
      "learning_rate": 3.6199237740201843e-06,
      "loss": 0.8696,
      "step": 23173
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.519826054573059,
      "learning_rate": 3.6191396004361913e-06,
      "loss": 1.0376,
      "step": 23174
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5763287544250488,
      "learning_rate": 3.6183554930319198e-06,
      "loss": 0.4054,
      "step": 23175
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6011639833450317,
      "learning_rate": 3.617571451815507e-06,
      "loss": 0.4441,
      "step": 23176
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5669156312942505,
      "learning_rate": 3.616787476795087e-06,
      "loss": 0.4775,
      "step": 23177
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5464327335357666,
      "learning_rate": 3.6160035679787886e-06,
      "loss": 0.3839,
      "step": 23178
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5901912450790405,
      "learning_rate": 3.615219725374739e-06,
      "loss": 0.4372,
      "step": 23179
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5944371223449707,
      "learning_rate": 3.6144359489910697e-06,
      "loss": 0.3948,
      "step": 23180
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6155860424041748,
      "learning_rate": 3.613652238835912e-06,
      "loss": 0.4357,
      "step": 23181
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4771431684494019,
      "learning_rate": 3.6128685949173947e-06,
      "loss": 0.3906,
      "step": 23182
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5923702716827393,
      "learning_rate": 3.612085017243644e-06,
      "loss": 0.4293,
      "step": 23183
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5370417833328247,
      "learning_rate": 3.611301505822784e-06,
      "loss": 0.3891,
      "step": 23184
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5926132202148438,
      "learning_rate": 3.610518060662942e-06,
      "loss": 0.4107,
      "step": 23185
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5355005264282227,
      "learning_rate": 3.609734681772249e-06,
      "loss": 0.378,
      "step": 23186
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5789202451705933,
      "learning_rate": 3.608951369158823e-06,
      "loss": 0.4581,
      "step": 23187
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.588855266571045,
      "learning_rate": 3.6081681228307943e-06,
      "loss": 0.3661,
      "step": 23188
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6296706199645996,
      "learning_rate": 3.6073849427962793e-06,
      "loss": 0.5027,
      "step": 23189
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.557562232017517,
      "learning_rate": 3.6066018290634085e-06,
      "loss": 0.3794,
      "step": 23190
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.887730598449707,
      "learning_rate": 3.605818781640297e-06,
      "loss": 0.4475,
      "step": 23191
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4592506885528564,
      "learning_rate": 3.6050358005350695e-06,
      "loss": 0.3592,
      "step": 23192
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5948182344436646,
      "learning_rate": 3.60425288575585e-06,
      "loss": 0.4206,
      "step": 23193
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5357669591903687,
      "learning_rate": 3.6034700373107556e-06,
      "loss": 0.4264,
      "step": 23194
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5758439302444458,
      "learning_rate": 3.602687255207903e-06,
      "loss": 0.4455,
      "step": 23195
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.588101863861084,
      "learning_rate": 3.601904539455412e-06,
      "loss": 0.3847,
      "step": 23196
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5573598146438599,
      "learning_rate": 3.601121890061402e-06,
      "loss": 0.4209,
      "step": 23197
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5352346897125244,
      "learning_rate": 3.6003393070339944e-06,
      "loss": 0.3507,
      "step": 23198
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6033655405044556,
      "learning_rate": 3.599556790381301e-06,
      "loss": 0.4721,
      "step": 23199
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5770989656448364,
      "learning_rate": 3.598774340111436e-06,
      "loss": 0.4168,
      "step": 23200
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7211660146713257,
      "learning_rate": 3.597991956232517e-06,
      "loss": 0.4455,
      "step": 23201
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5039621591567993,
      "learning_rate": 3.597209638752662e-06,
      "loss": 0.3784,
      "step": 23202
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5299782752990723,
      "learning_rate": 3.596427387679978e-06,
      "loss": 0.4596,
      "step": 23203
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4680567979812622,
      "learning_rate": 3.5956452030225854e-06,
      "loss": 0.4012,
      "step": 23204
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.558809518814087,
      "learning_rate": 3.5948630847885902e-06,
      "loss": 0.4544,
      "step": 23205
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.503421664237976,
      "learning_rate": 3.5940810329861097e-06,
      "loss": 0.4044,
      "step": 23206
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5823458433151245,
      "learning_rate": 3.59329904762325e-06,
      "loss": 0.4796,
      "step": 23207
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.535032868385315,
      "learning_rate": 3.592517128708126e-06,
      "loss": 0.3911,
      "step": 23208
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5070401430130005,
      "learning_rate": 3.5917352762488434e-06,
      "loss": 0.4469,
      "step": 23209
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4135609865188599,
      "learning_rate": 3.5909534902535158e-06,
      "loss": 0.3586,
      "step": 23210
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6991685628890991,
      "learning_rate": 3.5901717707302464e-06,
      "loss": 0.4557,
      "step": 23211
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4743462800979614,
      "learning_rate": 3.589390117687145e-06,
      "loss": 0.3655,
      "step": 23212
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8257243633270264,
      "learning_rate": 3.588608531132324e-06,
      "loss": 0.4694,
      "step": 23213
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.440381646156311,
      "learning_rate": 3.5878270110738833e-06,
      "loss": 0.3658,
      "step": 23214
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5195871591567993,
      "learning_rate": 3.587045557519927e-06,
      "loss": 0.4177,
      "step": 23215
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5357612371444702,
      "learning_rate": 3.5862641704785627e-06,
      "loss": 0.3826,
      "step": 23216
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.57107412815094,
      "learning_rate": 3.5854828499578953e-06,
      "loss": 0.4562,
      "step": 23217
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5650434494018555,
      "learning_rate": 3.5847015959660323e-06,
      "loss": 0.386,
      "step": 23218
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5469263792037964,
      "learning_rate": 3.5839204085110713e-06,
      "loss": 0.461,
      "step": 23219
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1281964778900146,
      "learning_rate": 3.583139287601113e-06,
      "loss": 0.4154,
      "step": 23220
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.594445824623108,
      "learning_rate": 3.582358233244261e-06,
      "loss": 0.409,
      "step": 23221
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5292068719863892,
      "learning_rate": 3.5815772454486197e-06,
      "loss": 0.3912,
      "step": 23222
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5596731901168823,
      "learning_rate": 3.580796324222282e-06,
      "loss": 0.4402,
      "step": 23223
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5624897480010986,
      "learning_rate": 3.580015469573356e-06,
      "loss": 0.3712,
      "step": 23224
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5097107887268066,
      "learning_rate": 3.579234681509931e-06,
      "loss": 0.4218,
      "step": 23225
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4887654781341553,
      "learning_rate": 3.5784539600401138e-06,
      "loss": 0.3805,
      "step": 23226
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5141382217407227,
      "learning_rate": 3.5776733051719935e-06,
      "loss": 0.4256,
      "step": 23227
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.544225811958313,
      "learning_rate": 3.5768927169136723e-06,
      "loss": 0.4017,
      "step": 23228
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5378189086914062,
      "learning_rate": 3.576112195273247e-06,
      "loss": 0.4795,
      "step": 23229
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5501450300216675,
      "learning_rate": 3.5753317402588107e-06,
      "loss": 0.3967,
      "step": 23230
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.579634666442871,
      "learning_rate": 3.5745513518784548e-06,
      "loss": 0.4621,
      "step": 23231
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5083961486816406,
      "learning_rate": 3.573771030140275e-06,
      "loss": 0.3625,
      "step": 23232
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.542651653289795,
      "learning_rate": 3.572990775052366e-06,
      "loss": 0.4512,
      "step": 23233
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4421000480651855,
      "learning_rate": 3.572210586622824e-06,
      "loss": 0.3666,
      "step": 23234
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.448540449142456,
      "learning_rate": 3.5714304648597374e-06,
      "loss": 0.3724,
      "step": 23235
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5995949506759644,
      "learning_rate": 3.5706504097711925e-06,
      "loss": 0.3658,
      "step": 23236
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.627896785736084,
      "learning_rate": 3.5698704213652835e-06,
      "loss": 0.4158,
      "step": 23237
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4859611988067627,
      "learning_rate": 3.569090499650104e-06,
      "loss": 0.3582,
      "step": 23238
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5570809841156006,
      "learning_rate": 3.5683106446337368e-06,
      "loss": 0.4167,
      "step": 23239
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.489271879196167,
      "learning_rate": 3.5675308563242762e-06,
      "loss": 0.3707,
      "step": 23240
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5892101526260376,
      "learning_rate": 3.5667511347298035e-06,
      "loss": 0.4025,
      "step": 23241
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5792882442474365,
      "learning_rate": 3.5659714798584122e-06,
      "loss": 0.3997,
      "step": 23242
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5438663959503174,
      "learning_rate": 3.5651918917181803e-06,
      "loss": 0.4167,
      "step": 23243
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4954307079315186,
      "learning_rate": 3.564412370317204e-06,
      "loss": 0.3864,
      "step": 23244
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.853541374206543,
      "learning_rate": 3.5636329156635573e-06,
      "loss": 0.4522,
      "step": 23245
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.478514313697815,
      "learning_rate": 3.562853527765333e-06,
      "loss": 0.3739,
      "step": 23246
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5930966138839722,
      "learning_rate": 3.5620742066306092e-06,
      "loss": 0.4438,
      "step": 23247
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4938398599624634,
      "learning_rate": 3.561294952267469e-06,
      "loss": 0.3762,
      "step": 23248
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.558924913406372,
      "learning_rate": 3.5605157646840003e-06,
      "loss": 0.441,
      "step": 23249
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5855529308319092,
      "learning_rate": 3.559736643888281e-06,
      "loss": 0.3755,
      "step": 23250
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6040176153182983,
      "learning_rate": 3.5589575898883867e-06,
      "loss": 0.4735,
      "step": 23251
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.493051528930664,
      "learning_rate": 3.5581786026924026e-06,
      "loss": 0.3763,
      "step": 23252
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5450888872146606,
      "learning_rate": 3.5573996823084067e-06,
      "loss": 0.4181,
      "step": 23253
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.507217526435852,
      "learning_rate": 3.5566208287444824e-06,
      "loss": 0.4028,
      "step": 23254
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5790505409240723,
      "learning_rate": 3.555842042008704e-06,
      "loss": 0.4619,
      "step": 23255
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5152640342712402,
      "learning_rate": 3.5550633221091434e-06,
      "loss": 0.4109,
      "step": 23256
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5887689590454102,
      "learning_rate": 3.5542846690538833e-06,
      "loss": 0.4277,
      "step": 23257
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6097180843353271,
      "learning_rate": 3.5535060828510015e-06,
      "loss": 0.4028,
      "step": 23258
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.528329610824585,
      "learning_rate": 3.5527275635085667e-06,
      "loss": 0.4105,
      "step": 23259
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.492235541343689,
      "learning_rate": 3.5519491110346604e-06,
      "loss": 0.3739,
      "step": 23260
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5567620992660522,
      "learning_rate": 3.5511707254373494e-06,
      "loss": 0.4102,
      "step": 23261
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4779210090637207,
      "learning_rate": 3.550392406724713e-06,
      "loss": 0.371,
      "step": 23262
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5209167003631592,
      "learning_rate": 3.549614154904818e-06,
      "loss": 0.4699,
      "step": 23263
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5063806772232056,
      "learning_rate": 3.548835969985739e-06,
      "loss": 0.381,
      "step": 23264
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.696740984916687,
      "learning_rate": 3.5480578519755495e-06,
      "loss": 0.4527,
      "step": 23265
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5174221992492676,
      "learning_rate": 3.5472798008823185e-06,
      "loss": 0.3814,
      "step": 23266
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5486302375793457,
      "learning_rate": 3.54650181671411e-06,
      "loss": 0.4073,
      "step": 23267
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5335403680801392,
      "learning_rate": 3.5457238994789977e-06,
      "loss": 0.3553,
      "step": 23268
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.509244680404663,
      "learning_rate": 3.5449460491850494e-06,
      "loss": 0.4035,
      "step": 23269
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6052212715148926,
      "learning_rate": 3.5441682658403353e-06,
      "loss": 0.3786,
      "step": 23270
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6166527271270752,
      "learning_rate": 3.543390549452921e-06,
      "loss": 0.4101,
      "step": 23271
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5969876050949097,
      "learning_rate": 3.542612900030867e-06,
      "loss": 0.4176,
      "step": 23272
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.640666127204895,
      "learning_rate": 3.541835317582244e-06,
      "loss": 0.4219,
      "step": 23273
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6465566158294678,
      "learning_rate": 3.541057802115119e-06,
      "loss": 0.4085,
      "step": 23274
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5568783283233643,
      "learning_rate": 3.5402803536375486e-06,
      "loss": 0.4113,
      "step": 23275
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5056736469268799,
      "learning_rate": 3.539502972157605e-06,
      "loss": 0.3571,
      "step": 23276
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5986897945404053,
      "learning_rate": 3.538725657683343e-06,
      "loss": 0.5029,
      "step": 23277
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.617448091506958,
      "learning_rate": 3.537948410222831e-06,
      "loss": 0.4205,
      "step": 23278
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.599400281906128,
      "learning_rate": 3.537171229784124e-06,
      "loss": 0.4238,
      "step": 23279
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6150814294815063,
      "learning_rate": 3.5363941163752855e-06,
      "loss": 0.4365,
      "step": 23280
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6139142513275146,
      "learning_rate": 3.53561707000438e-06,
      "loss": 0.413,
      "step": 23281
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5489593744277954,
      "learning_rate": 3.5348400906794623e-06,
      "loss": 0.3748,
      "step": 23282
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5268478393554688,
      "learning_rate": 3.5340631784085877e-06,
      "loss": 0.4261,
      "step": 23283
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5314140319824219,
      "learning_rate": 3.533286333199818e-06,
      "loss": 0.3672,
      "step": 23284
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5378283262252808,
      "learning_rate": 3.532509555061213e-06,
      "loss": 0.4534,
      "step": 23285
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.500139594078064,
      "learning_rate": 3.531732844000826e-06,
      "loss": 0.3836,
      "step": 23286
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.547672152519226,
      "learning_rate": 3.5309562000267083e-06,
      "loss": 0.4164,
      "step": 23287
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6720507144927979,
      "learning_rate": 3.530179623146921e-06,
      "loss": 0.3911,
      "step": 23288
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5690559148788452,
      "learning_rate": 3.529403113369515e-06,
      "loss": 0.4183,
      "step": 23289
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5580852031707764,
      "learning_rate": 3.5286266707025507e-06,
      "loss": 0.426,
      "step": 23290
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5630031824111938,
      "learning_rate": 3.527850295154075e-06,
      "loss": 0.4413,
      "step": 23291
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4836959838867188,
      "learning_rate": 3.527073986732138e-06,
      "loss": 0.3566,
      "step": 23292
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5558897256851196,
      "learning_rate": 3.5262977454447954e-06,
      "loss": 0.4101,
      "step": 23293
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.660053014755249,
      "learning_rate": 3.5255215713001e-06,
      "loss": 0.4107,
      "step": 23294
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5246925354003906,
      "learning_rate": 3.5247454643060953e-06,
      "loss": 0.393,
      "step": 23295
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.100635051727295,
      "learning_rate": 3.5239694244708378e-06,
      "loss": 0.4212,
      "step": 23296
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5410722494125366,
      "learning_rate": 3.5231934518023702e-06,
      "loss": 0.4397,
      "step": 23297
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5079128742218018,
      "learning_rate": 3.522417546308747e-06,
      "loss": 0.3994,
      "step": 23298
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.565385341644287,
      "learning_rate": 3.521641707998008e-06,
      "loss": 0.4093,
      "step": 23299
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.512395977973938,
      "learning_rate": 3.520865936878205e-06,
      "loss": 0.3782,
      "step": 23300
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5970864295959473,
      "learning_rate": 3.520090232957385e-06,
      "loss": 0.4425,
      "step": 23301
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5297598838806152,
      "learning_rate": 3.5193145962435926e-06,
      "loss": 0.3943,
      "step": 23302
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8100417852401733,
      "learning_rate": 3.518539026744866e-06,
      "loss": 0.9754,
      "step": 23303
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4621955156326294,
      "learning_rate": 3.5177635244692552e-06,
      "loss": 0.936,
      "step": 23304
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6747440099716187,
      "learning_rate": 3.516988089424802e-06,
      "loss": 0.4399,
      "step": 23305
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5655947923660278,
      "learning_rate": 3.516212721619553e-06,
      "loss": 0.4035,
      "step": 23306
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5120964050292969,
      "learning_rate": 3.515437421061545e-06,
      "loss": 0.4078,
      "step": 23307
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.493309497833252,
      "learning_rate": 3.514662187758817e-06,
      "loss": 0.3899,
      "step": 23308
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7311711311340332,
      "learning_rate": 3.5138870217194145e-06,
      "loss": 0.5202,
      "step": 23309
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5523048639297485,
      "learning_rate": 3.5131119229513765e-06,
      "loss": 0.3606,
      "step": 23310
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5306472778320312,
      "learning_rate": 3.5123368914627387e-06,
      "loss": 0.4316,
      "step": 23311
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4729927778244019,
      "learning_rate": 3.511561927261545e-06,
      "loss": 0.3685,
      "step": 23312
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4983807802200317,
      "learning_rate": 3.5107870303558268e-06,
      "loss": 0.4152,
      "step": 23313
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4739519357681274,
      "learning_rate": 3.5100122007536265e-06,
      "loss": 0.3985,
      "step": 23314
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5892847776412964,
      "learning_rate": 3.5092374384629745e-06,
      "loss": 0.4164,
      "step": 23315
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5530890226364136,
      "learning_rate": 3.5084627434919093e-06,
      "loss": 0.3822,
      "step": 23316
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5487250089645386,
      "learning_rate": 3.5076881158484688e-06,
      "loss": 0.4929,
      "step": 23317
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4367567300796509,
      "learning_rate": 3.5069135555406852e-06,
      "loss": 0.3684,
      "step": 23318
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4836031198501587,
      "learning_rate": 3.506139062576588e-06,
      "loss": 0.4388,
      "step": 23319
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.635071873664856,
      "learning_rate": 3.5053646369642115e-06,
      "loss": 0.371,
      "step": 23320
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5063225030899048,
      "learning_rate": 3.5045902787115902e-06,
      "loss": 0.4132,
      "step": 23321
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0254476070404053,
      "learning_rate": 3.503815987826761e-06,
      "loss": 0.431,
      "step": 23322
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5399658679962158,
      "learning_rate": 3.5030417643177416e-06,
      "loss": 0.472,
      "step": 23323
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5228396654129028,
      "learning_rate": 3.5022676081925676e-06,
      "loss": 0.3812,
      "step": 23324
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.649735450744629,
      "learning_rate": 3.501493519459269e-06,
      "loss": 0.4389,
      "step": 23325
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4998133182525635,
      "learning_rate": 3.5007194981258776e-06,
      "loss": 0.3718,
      "step": 23326
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5966371297836304,
      "learning_rate": 3.4999455442004184e-06,
      "loss": 0.4543,
      "step": 23327
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.501811146736145,
      "learning_rate": 3.499171657690914e-06,
      "loss": 0.3769,
      "step": 23328
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6200430393218994,
      "learning_rate": 3.498397838605395e-06,
      "loss": 0.4267,
      "step": 23329
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5971883535385132,
      "learning_rate": 3.4976240869518907e-06,
      "loss": 0.3876,
      "step": 23330
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6501047611236572,
      "learning_rate": 3.4968504027384186e-06,
      "loss": 0.4409,
      "step": 23331
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5347378253936768,
      "learning_rate": 3.4960767859730103e-06,
      "loss": 0.3788,
      "step": 23332
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5856022834777832,
      "learning_rate": 3.495303236663683e-06,
      "loss": 0.454,
      "step": 23333
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5210283994674683,
      "learning_rate": 3.4945297548184654e-06,
      "loss": 0.4021,
      "step": 23334
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.674411654472351,
      "learning_rate": 3.4937563404453745e-06,
      "loss": 0.4673,
      "step": 23335
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5105125904083252,
      "learning_rate": 3.492982993552434e-06,
      "loss": 0.3745,
      "step": 23336
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.511360764503479,
      "learning_rate": 3.4922097141476684e-06,
      "loss": 0.4142,
      "step": 23337
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.469154953956604,
      "learning_rate": 3.491436502239094e-06,
      "loss": 0.3733,
      "step": 23338
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.661271095275879,
      "learning_rate": 3.4906633578347284e-06,
      "loss": 1.0169,
      "step": 23339
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.443302869796753,
      "learning_rate": 3.489890280942593e-06,
      "loss": 0.9346,
      "step": 23340
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5486236810684204,
      "learning_rate": 3.4891172715707044e-06,
      "loss": 0.4209,
      "step": 23341
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5347062349319458,
      "learning_rate": 3.4883443297270845e-06,
      "loss": 0.3818,
      "step": 23342
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4265457391738892,
      "learning_rate": 3.487571455419747e-06,
      "loss": 0.376,
      "step": 23343
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5206326246261597,
      "learning_rate": 3.486798648656705e-06,
      "loss": 0.3827,
      "step": 23344
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.714683771133423,
      "learning_rate": 3.4860259094459747e-06,
      "loss": 0.47,
      "step": 23345
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5272163152694702,
      "learning_rate": 3.485253237795576e-06,
      "loss": 0.3901,
      "step": 23346
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5696980953216553,
      "learning_rate": 3.484480633713516e-06,
      "loss": 0.474,
      "step": 23347
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6983027458190918,
      "learning_rate": 3.483708097207813e-06,
      "loss": 0.414,
      "step": 23348
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.604512095451355,
      "learning_rate": 3.4829356282864744e-06,
      "loss": 0.4483,
      "step": 23349
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.535475254058838,
      "learning_rate": 3.482163226957518e-06,
      "loss": 0.3726,
      "step": 23350
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.527517318725586,
      "learning_rate": 3.4813908932289485e-06,
      "loss": 0.4242,
      "step": 23351
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5514931678771973,
      "learning_rate": 3.480618627108778e-06,
      "loss": 0.3784,
      "step": 23352
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6199414730072021,
      "learning_rate": 3.4798464286050216e-06,
      "loss": 0.4442,
      "step": 23353
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5316435098648071,
      "learning_rate": 3.479074297725684e-06,
      "loss": 0.4036,
      "step": 23354
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.551347017288208,
      "learning_rate": 3.4783022344787698e-06,
      "loss": 0.4308,
      "step": 23355
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5289294719696045,
      "learning_rate": 3.477530238872291e-06,
      "loss": 0.3974,
      "step": 23356
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5018019676208496,
      "learning_rate": 3.4767583109142523e-06,
      "loss": 0.3954,
      "step": 23357
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5496721267700195,
      "learning_rate": 3.4759864506126683e-06,
      "loss": 0.4105,
      "step": 23358
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5807431936264038,
      "learning_rate": 3.4752146579755307e-06,
      "loss": 0.4259,
      "step": 23359
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5254930257797241,
      "learning_rate": 3.474442933010851e-06,
      "loss": 0.386,
      "step": 23360
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5805790424346924,
      "learning_rate": 3.4736712757266323e-06,
      "loss": 0.4702,
      "step": 23361
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6396275758743286,
      "learning_rate": 3.472899686130883e-06,
      "loss": 0.3919,
      "step": 23362
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6484522819519043,
      "learning_rate": 3.4721281642315994e-06,
      "loss": 0.4659,
      "step": 23363
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5345890522003174,
      "learning_rate": 3.471356710036784e-06,
      "loss": 0.3977,
      "step": 23364
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5708202123641968,
      "learning_rate": 3.4705853235544386e-06,
      "loss": 0.4297,
      "step": 23365
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5212303400039673,
      "learning_rate": 3.4698140047925677e-06,
      "loss": 0.3877,
      "step": 23366
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6846026182174683,
      "learning_rate": 3.4690427537591644e-06,
      "loss": 1.047,
      "step": 23367
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4752082824707031,
      "learning_rate": 3.468271570462235e-06,
      "loss": 0.9989,
      "step": 23368
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5346705913543701,
      "learning_rate": 3.46750045490977e-06,
      "loss": 0.4209,
      "step": 23369
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5564720630645752,
      "learning_rate": 3.4667294071097723e-06,
      "loss": 0.3748,
      "step": 23370
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.594539999961853,
      "learning_rate": 3.4659584270702406e-06,
      "loss": 0.4001,
      "step": 23371
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.542918086051941,
      "learning_rate": 3.4651875147991655e-06,
      "loss": 0.388,
      "step": 23372
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5076544284820557,
      "learning_rate": 3.464416670304549e-06,
      "loss": 0.4323,
      "step": 23373
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5215020179748535,
      "learning_rate": 3.4636458935943784e-06,
      "loss": 0.3921,
      "step": 23374
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.813888669013977,
      "learning_rate": 3.462875184676656e-06,
      "loss": 0.4226,
      "step": 23375
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5409905910491943,
      "learning_rate": 3.462104543559368e-06,
      "loss": 0.3631,
      "step": 23376
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4716017246246338,
      "learning_rate": 3.4613339702505113e-06,
      "loss": 0.3953,
      "step": 23377
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6110869646072388,
      "learning_rate": 3.460563464758079e-06,
      "loss": 0.3801,
      "step": 23378
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5369384288787842,
      "learning_rate": 3.4597930270900613e-06,
      "loss": 0.4533,
      "step": 23379
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.534142255783081,
      "learning_rate": 3.459022657254445e-06,
      "loss": 0.3778,
      "step": 23380
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5087085962295532,
      "learning_rate": 3.458252355259223e-06,
      "loss": 0.4679,
      "step": 23381
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.476371169090271,
      "learning_rate": 3.4574821211123854e-06,
      "loss": 0.3555,
      "step": 23382
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5712133646011353,
      "learning_rate": 3.456711954821923e-06,
      "loss": 0.4499,
      "step": 23383
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.4905295372009277,
      "learning_rate": 3.4559418563958215e-06,
      "loss": 0.3898,
      "step": 23384
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5414435863494873,
      "learning_rate": 3.455171825842063e-06,
      "loss": 0.4024,
      "step": 23385
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5214663743972778,
      "learning_rate": 3.454401863168639e-06,
      "loss": 0.3859,
      "step": 23386
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9627984762191772,
      "learning_rate": 3.4536319683835384e-06,
      "loss": 0.4187,
      "step": 23387
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.468993663787842,
      "learning_rate": 3.4528621414947395e-06,
      "loss": 0.3674,
      "step": 23388
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.533207654953003,
      "learning_rate": 3.4520923825102315e-06,
      "loss": 0.4073,
      "step": 23389
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6013740301132202,
      "learning_rate": 3.4513226914379937e-06,
      "loss": 0.4124,
      "step": 23390
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6457138061523438,
      "learning_rate": 3.4505530682860143e-06,
      "loss": 0.4693,
      "step": 23391
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6533002853393555,
      "learning_rate": 3.4497835130622703e-06,
      "loss": 0.3829,
      "step": 23392
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5798499584197998,
      "learning_rate": 3.449014025774744e-06,
      "loss": 0.4402,
      "step": 23393
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5370917320251465,
      "learning_rate": 3.4482446064314224e-06,
      "loss": 0.3707,
      "step": 23394
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4819309711456299,
      "learning_rate": 3.4474752550402812e-06,
      "loss": 0.4085,
      "step": 23395
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5498872995376587,
      "learning_rate": 3.4467059716092955e-06,
      "loss": 0.4093,
      "step": 23396
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5959943532943726,
      "learning_rate": 3.4459367561464484e-06,
      "loss": 0.4699,
      "step": 23397
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6622987985610962,
      "learning_rate": 3.4451676086597174e-06,
      "loss": 0.4032,
      "step": 23398
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7530713081359863,
      "learning_rate": 3.444398529157087e-06,
      "loss": 0.9823,
      "step": 23399
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4619165658950806,
      "learning_rate": 3.4436295176465197e-06,
      "loss": 1.0289,
      "step": 23400
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5430572032928467,
      "learning_rate": 3.4428605741359988e-06,
      "loss": 0.4116,
      "step": 23401
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5596294403076172,
      "learning_rate": 3.4420916986334995e-06,
      "loss": 0.4095,
      "step": 23402
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.526301383972168,
      "learning_rate": 3.441322891146999e-06,
      "loss": 0.4322,
      "step": 23403
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4658966064453125,
      "learning_rate": 3.440554151684468e-06,
      "loss": 0.3749,
      "step": 23404
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.606319785118103,
      "learning_rate": 3.4397854802538756e-06,
      "loss": 0.4431,
      "step": 23405
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5577534437179565,
      "learning_rate": 3.439016876863198e-06,
      "loss": 0.3778,
      "step": 23406
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5651824474334717,
      "learning_rate": 3.4382483415204114e-06,
      "loss": 0.4527,
      "step": 23407
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7316341400146484,
      "learning_rate": 3.437479874233478e-06,
      "loss": 0.3837,
      "step": 23408
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6073697805404663,
      "learning_rate": 3.436711475010376e-06,
      "loss": 0.4051,
      "step": 23409
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5613653659820557,
      "learning_rate": 3.4359431438590683e-06,
      "loss": 0.4116,
      "step": 23410
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5103946924209595,
      "learning_rate": 3.4351748807875294e-06,
      "loss": 0.437,
      "step": 23411
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.547533631324768,
      "learning_rate": 3.4344066858037217e-06,
      "loss": 0.377,
      "step": 23412
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6041430234909058,
      "learning_rate": 3.433638558915615e-06,
      "loss": 0.4441,
      "step": 23413
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.574562668800354,
      "learning_rate": 3.4328705001311803e-06,
      "loss": 0.3751,
      "step": 23414
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6198467016220093,
      "learning_rate": 3.4321025094583806e-06,
      "loss": 0.4505,
      "step": 23415
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5377179384231567,
      "learning_rate": 3.4313345869051762e-06,
      "loss": 0.3746,
      "step": 23416
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.583892822265625,
      "learning_rate": 3.4305667324795357e-06,
      "loss": 0.4637,
      "step": 23417
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.495287537574768,
      "learning_rate": 3.429798946189423e-06,
      "loss": 0.3742,
      "step": 23418
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5736908912658691,
      "learning_rate": 3.4290312280428064e-06,
      "loss": 0.393,
      "step": 23419
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5728018283843994,
      "learning_rate": 3.4282635780476427e-06,
      "loss": 0.3746,
      "step": 23420
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6755032539367676,
      "learning_rate": 3.4274959962118913e-06,
      "loss": 0.4343,
      "step": 23421
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4671216011047363,
      "learning_rate": 3.4267284825435156e-06,
      "loss": 0.3656,
      "step": 23422
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.603089690208435,
      "learning_rate": 3.4259610370504805e-06,
      "loss": 0.4291,
      "step": 23423
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4948768615722656,
      "learning_rate": 3.4251936597407386e-06,
      "loss": 0.3795,
      "step": 23424
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5537090301513672,
      "learning_rate": 3.4244263506222565e-06,
      "loss": 0.4264,
      "step": 23425
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8572709560394287,
      "learning_rate": 3.4236591097029824e-06,
      "loss": 0.4078,
      "step": 23426
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5446640253067017,
      "learning_rate": 3.422891936990884e-06,
      "loss": 0.4031,
      "step": 23427
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.540150761604309,
      "learning_rate": 3.4221248324939105e-06,
      "loss": 0.4091,
      "step": 23428
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5811283588409424,
      "learning_rate": 3.421357796220021e-06,
      "loss": 0.3981,
      "step": 23429
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5227094888687134,
      "learning_rate": 3.420590828177174e-06,
      "loss": 0.371,
      "step": 23430
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5660287141799927,
      "learning_rate": 3.4198239283733216e-06,
      "loss": 0.462,
      "step": 23431
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6124399900436401,
      "learning_rate": 3.419057096816413e-06,
      "loss": 0.392,
      "step": 23432
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7085236310958862,
      "learning_rate": 3.418290333514407e-06,
      "loss": 0.4394,
      "step": 23433
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.625622272491455,
      "learning_rate": 3.417523638475254e-06,
      "loss": 0.4111,
      "step": 23434
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5800877809524536,
      "learning_rate": 3.416757011706914e-06,
      "loss": 0.4231,
      "step": 23435
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.545638084411621,
      "learning_rate": 3.4159904532173236e-06,
      "loss": 0.3866,
      "step": 23436
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7044808864593506,
      "learning_rate": 3.4152239630144404e-06,
      "loss": 0.9336,
      "step": 23437
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4496614933013916,
      "learning_rate": 3.414457541106215e-06,
      "loss": 0.9918,
      "step": 23438
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5830451250076294,
      "learning_rate": 3.4136911875005985e-06,
      "loss": 0.4043,
      "step": 23439
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.601832628250122,
      "learning_rate": 3.4129249022055367e-06,
      "loss": 0.4102,
      "step": 23440
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.610464334487915,
      "learning_rate": 3.4121586852289747e-06,
      "loss": 0.4277,
      "step": 23441
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5271899700164795,
      "learning_rate": 3.4113925365788603e-06,
      "loss": 0.3664,
      "step": 23442
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5982087850570679,
      "learning_rate": 3.410626456263144e-06,
      "loss": 0.4428,
      "step": 23443
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4814945459365845,
      "learning_rate": 3.4098604442897656e-06,
      "loss": 0.3778,
      "step": 23444
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5330820083618164,
      "learning_rate": 3.4090945006666764e-06,
      "loss": 0.4342,
      "step": 23445
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.44899320602417,
      "learning_rate": 3.408328625401812e-06,
      "loss": 0.3739,
      "step": 23446
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.596632480621338,
      "learning_rate": 3.407562818503125e-06,
      "loss": 0.4643,
      "step": 23447
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5443214178085327,
      "learning_rate": 3.40679707997855e-06,
      "loss": 0.3841,
      "step": 23448
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5627785921096802,
      "learning_rate": 3.4060314098360312e-06,
      "loss": 0.4339,
      "step": 23449
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.564805030822754,
      "learning_rate": 3.4052658080835156e-06,
      "loss": 0.3895,
      "step": 23450
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4840376377105713,
      "learning_rate": 3.404500274728938e-06,
      "loss": 0.4058,
      "step": 23451
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5227129459381104,
      "learning_rate": 3.403734809780237e-06,
      "loss": 0.3745,
      "step": 23452
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5870274305343628,
      "learning_rate": 3.4029694132453526e-06,
      "loss": 0.4474,
      "step": 23453
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5073596239089966,
      "learning_rate": 3.4022040851322258e-06,
      "loss": 0.3796,
      "step": 23454
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5080313682556152,
      "learning_rate": 3.401438825448796e-06,
      "loss": 0.4048,
      "step": 23455
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5446362495422363,
      "learning_rate": 3.400673634202998e-06,
      "loss": 0.3908,
      "step": 23456
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.592499017715454,
      "learning_rate": 3.399908511402762e-06,
      "loss": 0.4243,
      "step": 23457
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4992892742156982,
      "learning_rate": 3.39914345705603e-06,
      "loss": 0.3784,
      "step": 23458
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7055715322494507,
      "learning_rate": 3.398378471170739e-06,
      "loss": 0.4424,
      "step": 23459
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5955476760864258,
      "learning_rate": 3.397613553754817e-06,
      "loss": 0.3729,
      "step": 23460
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5040068626403809,
      "learning_rate": 3.3968487048162025e-06,
      "loss": 0.4338,
      "step": 23461
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.54047691822052,
      "learning_rate": 3.396083924362823e-06,
      "loss": 0.404,
      "step": 23462
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5701581239700317,
      "learning_rate": 3.395319212402617e-06,
      "loss": 0.4109,
      "step": 23463
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5389258861541748,
      "learning_rate": 3.3945545689435087e-06,
      "loss": 0.3754,
      "step": 23464
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.612839937210083,
      "learning_rate": 3.3937899939934316e-06,
      "loss": 0.3891,
      "step": 23465
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5878804922103882,
      "learning_rate": 3.3930254875603197e-06,
      "loss": 0.3906,
      "step": 23466
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6043062210083008,
      "learning_rate": 3.392261049652099e-06,
      "loss": 0.4065,
      "step": 23467
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5713279247283936,
      "learning_rate": 3.3914966802766946e-06,
      "loss": 0.3867,
      "step": 23468
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6228678226470947,
      "learning_rate": 3.3907323794420367e-06,
      "loss": 0.4593,
      "step": 23469
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6404169797897339,
      "learning_rate": 3.3899681471560518e-06,
      "loss": 0.4006,
      "step": 23470
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.859116792678833,
      "learning_rate": 3.389203983426672e-06,
      "loss": 0.4114,
      "step": 23471
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4897339344024658,
      "learning_rate": 3.3884398882618173e-06,
      "loss": 0.3907,
      "step": 23472
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7036025524139404,
      "learning_rate": 3.3876758616694095e-06,
      "loss": 0.4205,
      "step": 23473
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.544008731842041,
      "learning_rate": 3.3869119036573774e-06,
      "loss": 0.4106,
      "step": 23474
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.538125991821289,
      "learning_rate": 3.3861480142336477e-06,
      "loss": 0.4065,
      "step": 23475
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5572870969772339,
      "learning_rate": 3.3853841934061383e-06,
      "loss": 0.3907,
      "step": 23476
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.851698398590088,
      "learning_rate": 3.3846204411827686e-06,
      "loss": 0.9597,
      "step": 23477
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5978786945343018,
      "learning_rate": 3.3838567575714644e-06,
      "loss": 1.1222,
      "step": 23478
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5420646667480469,
      "learning_rate": 3.383093142580147e-06,
      "loss": 0.4784,
      "step": 23479
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5476468801498413,
      "learning_rate": 3.3823295962167337e-06,
      "loss": 0.3895,
      "step": 23480
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6118379831314087,
      "learning_rate": 3.381566118489146e-06,
      "loss": 0.4538,
      "step": 23481
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5972586870193481,
      "learning_rate": 3.3808027094052988e-06,
      "loss": 0.3714,
      "step": 23482
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6104133129119873,
      "learning_rate": 3.380039368973115e-06,
      "loss": 0.4699,
      "step": 23483
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5075724124908447,
      "learning_rate": 3.379276097200506e-06,
      "loss": 0.3532,
      "step": 23484
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.605259656906128,
      "learning_rate": 3.37851289409539e-06,
      "loss": 0.4142,
      "step": 23485
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5409412384033203,
      "learning_rate": 3.3777497596656883e-06,
      "loss": 0.3882,
      "step": 23486
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5249453783035278,
      "learning_rate": 3.3769866939193098e-06,
      "loss": 0.4029,
      "step": 23487
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5024170875549316,
      "learning_rate": 3.376223696864167e-06,
      "loss": 0.3728,
      "step": 23488
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7147974967956543,
      "learning_rate": 3.3754607685081763e-06,
      "loss": 1.0169,
      "step": 23489
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4476940631866455,
      "learning_rate": 3.3746979088592515e-06,
      "loss": 0.9867,
      "step": 23490
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5206371545791626,
      "learning_rate": 3.3739351179253066e-06,
      "loss": 0.4262,
      "step": 23491
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4991132020950317,
      "learning_rate": 3.3731723957142504e-06,
      "loss": 0.3887,
      "step": 23492
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.609180212020874,
      "learning_rate": 3.3724097422339895e-06,
      "loss": 0.4095,
      "step": 23493
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6216291189193726,
      "learning_rate": 3.371647157492437e-06,
      "loss": 0.4068,
      "step": 23494
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.639573097229004,
      "learning_rate": 3.370884641497507e-06,
      "loss": 0.4612,
      "step": 23495
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.557700514793396,
      "learning_rate": 3.3701221942571007e-06,
      "loss": 0.3636,
      "step": 23496
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.706133484840393,
      "learning_rate": 3.3693598157791317e-06,
      "loss": 0.4386,
      "step": 23497
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5236369371414185,
      "learning_rate": 3.3685975060715013e-06,
      "loss": 0.3837,
      "step": 23498
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7027530670166016,
      "learning_rate": 3.367835265142122e-06,
      "loss": 0.462,
      "step": 23499
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.569029688835144,
      "learning_rate": 3.3670730929988925e-06,
      "loss": 0.3793,
      "step": 23500
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5328408479690552,
      "learning_rate": 3.3663109896497228e-06,
      "loss": 0.4259,
      "step": 23501
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5995798110961914,
      "learning_rate": 3.365548955102519e-06,
      "loss": 0.3931,
      "step": 23502
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5622239112854004,
      "learning_rate": 3.364786989365181e-06,
      "loss": 0.4305,
      "step": 23503
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5240668058395386,
      "learning_rate": 3.36402509244561e-06,
      "loss": 0.4004,
      "step": 23504
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8493932485580444,
      "learning_rate": 3.363263264351708e-06,
      "loss": 1.0367,
      "step": 23505
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4849258661270142,
      "learning_rate": 3.362501505091381e-06,
      "loss": 1.0085,
      "step": 23506
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5935635566711426,
      "learning_rate": 3.361739814672529e-06,
      "loss": 0.408,
      "step": 23507
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.543614387512207,
      "learning_rate": 3.360978193103052e-06,
      "loss": 0.4155,
      "step": 23508
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5150082111358643,
      "learning_rate": 3.360216640390842e-06,
      "loss": 0.4394,
      "step": 23509
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.520180583000183,
      "learning_rate": 3.3594551565438038e-06,
      "loss": 0.3752,
      "step": 23510
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.755135178565979,
      "learning_rate": 3.3586937415698382e-06,
      "loss": 0.952,
      "step": 23511
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4607384204864502,
      "learning_rate": 3.3579323954768384e-06,
      "loss": 1.032,
      "step": 23512
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6379152536392212,
      "learning_rate": 3.3571711182726975e-06,
      "loss": 0.4827,
      "step": 23513
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4688655138015747,
      "learning_rate": 3.3564099099653148e-06,
      "loss": 0.3686,
      "step": 23514
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5319340229034424,
      "learning_rate": 3.355648770562587e-06,
      "loss": 0.3966,
      "step": 23515
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6099214553833008,
      "learning_rate": 3.3548877000724045e-06,
      "loss": 0.3786,
      "step": 23516
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.552687644958496,
      "learning_rate": 3.3541266985026644e-06,
      "loss": 0.431,
      "step": 23517
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6092263460159302,
      "learning_rate": 3.353365765861256e-06,
      "loss": 0.4082,
      "step": 23518
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.1381566524505615,
      "learning_rate": 3.3526049021560757e-06,
      "loss": 0.4293,
      "step": 23519
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6534779071807861,
      "learning_rate": 3.351844107395009e-06,
      "loss": 0.3645,
      "step": 23520
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.578096866607666,
      "learning_rate": 3.35108338158595e-06,
      "loss": 0.3936,
      "step": 23521
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5208590030670166,
      "learning_rate": 3.350322724736791e-06,
      "loss": 0.3655,
      "step": 23522
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5545626878738403,
      "learning_rate": 3.34956213685542e-06,
      "loss": 0.4215,
      "step": 23523
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.552829384803772,
      "learning_rate": 3.3488016179497195e-06,
      "loss": 0.3807,
      "step": 23524
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5315771102905273,
      "learning_rate": 3.348041168027584e-06,
      "loss": 0.4157,
      "step": 23525
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.540956735610962,
      "learning_rate": 3.347280787096897e-06,
      "loss": 0.4035,
      "step": 23526
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4957915544509888,
      "learning_rate": 3.34652047516555e-06,
      "loss": 0.4164,
      "step": 23527
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5351091623306274,
      "learning_rate": 3.345760232241425e-06,
      "loss": 0.3886,
      "step": 23528
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.751029372215271,
      "learning_rate": 3.345000058332405e-06,
      "loss": 0.9061,
      "step": 23529
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.484100580215454,
      "learning_rate": 3.344239953446374e-06,
      "loss": 0.9842,
      "step": 23530
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.55074942111969,
      "learning_rate": 3.3434799175912236e-06,
      "loss": 0.4323,
      "step": 23531
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5663893222808838,
      "learning_rate": 3.3427199507748263e-06,
      "loss": 0.4031,
      "step": 23532
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6091442108154297,
      "learning_rate": 3.3419600530050722e-06,
      "loss": 0.4546,
      "step": 23533
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5602940320968628,
      "learning_rate": 3.3412002242898355e-06,
      "loss": 0.3987,
      "step": 23534
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5349669456481934,
      "learning_rate": 3.3404404646370036e-06,
      "loss": 0.4045,
      "step": 23535
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4613021612167358,
      "learning_rate": 3.33968077405445e-06,
      "loss": 0.3452,
      "step": 23536
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5599963665008545,
      "learning_rate": 3.3389211525500577e-06,
      "loss": 0.4399,
      "step": 23537
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5812674760818481,
      "learning_rate": 3.3381616001317074e-06,
      "loss": 0.3878,
      "step": 23538
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5420255661010742,
      "learning_rate": 3.3374021168072733e-06,
      "loss": 0.4091,
      "step": 23539
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.541161060333252,
      "learning_rate": 3.3366427025846315e-06,
      "loss": 0.3791,
      "step": 23540
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.649768352508545,
      "learning_rate": 3.33588335747166e-06,
      "loss": 0.4397,
      "step": 23541
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5320053100585938,
      "learning_rate": 3.335124081476234e-06,
      "loss": 0.3863,
      "step": 23542
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5380953550338745,
      "learning_rate": 3.3343648746062317e-06,
      "loss": 0.4205,
      "step": 23543
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5621839761734009,
      "learning_rate": 3.333605736869526e-06,
      "loss": 0.3811,
      "step": 23544
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.609812617301941,
      "learning_rate": 3.3328466682739846e-06,
      "loss": 0.4212,
      "step": 23545
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.670516848564148,
      "learning_rate": 3.332087668827485e-06,
      "loss": 0.4306,
      "step": 23546
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.601758599281311,
      "learning_rate": 3.3313287385379024e-06,
      "loss": 0.4521,
      "step": 23547
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.471286416053772,
      "learning_rate": 3.3305698774131003e-06,
      "loss": 0.3709,
      "step": 23548
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7777167558670044,
      "learning_rate": 3.3298110854609568e-06,
      "loss": 0.4381,
      "step": 23549
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4795536994934082,
      "learning_rate": 3.3290523626893355e-06,
      "loss": 0.3743,
      "step": 23550
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5522801876068115,
      "learning_rate": 3.328293709106113e-06,
      "loss": 0.4231,
      "step": 23551
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.507136344909668,
      "learning_rate": 3.327535124719149e-06,
      "loss": 0.3839,
      "step": 23552
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4880825281143188,
      "learning_rate": 3.3267766095363186e-06,
      "loss": 0.4453,
      "step": 23553
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5202518701553345,
      "learning_rate": 3.326018163565483e-06,
      "loss": 0.3773,
      "step": 23554
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.680603265762329,
      "learning_rate": 3.3252597868145143e-06,
      "loss": 0.4287,
      "step": 23555
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6358484029769897,
      "learning_rate": 3.3245014792912723e-06,
      "loss": 0.4221,
      "step": 23556
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.517655611038208,
      "learning_rate": 3.323743241003624e-06,
      "loss": 0.4212,
      "step": 23557
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5879778861999512,
      "learning_rate": 3.322985071959438e-06,
      "loss": 0.3981,
      "step": 23558
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5205817222595215,
      "learning_rate": 3.3222269721665734e-06,
      "loss": 0.4188,
      "step": 23559
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5138816833496094,
      "learning_rate": 3.3214689416328905e-06,
      "loss": 0.389,
      "step": 23560
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.547099232673645,
      "learning_rate": 3.320710980366254e-06,
      "loss": 0.441,
      "step": 23561
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5073319673538208,
      "learning_rate": 3.319953088374526e-06,
      "loss": 0.3778,
      "step": 23562
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5460628271102905,
      "learning_rate": 3.319195265665568e-06,
      "loss": 0.3964,
      "step": 23563
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5036362409591675,
      "learning_rate": 3.318437512247239e-06,
      "loss": 0.3636,
      "step": 23564
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5291187763214111,
      "learning_rate": 3.3176798281273947e-06,
      "loss": 0.4278,
      "step": 23565
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5857934951782227,
      "learning_rate": 3.3169222133138955e-06,
      "loss": 0.4713,
      "step": 23566
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6060113906860352,
      "learning_rate": 3.3161646678146042e-06,
      "loss": 0.4401,
      "step": 23567
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.491531252861023,
      "learning_rate": 3.3154071916373686e-06,
      "loss": 0.3806,
      "step": 23568
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5195189714431763,
      "learning_rate": 3.3146497847900536e-06,
      "loss": 0.4118,
      "step": 23569
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5099430084228516,
      "learning_rate": 3.313892447280508e-06,
      "loss": 0.3888,
      "step": 23570
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.532699704170227,
      "learning_rate": 3.313135179116591e-06,
      "loss": 0.4694,
      "step": 23571
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.558250904083252,
      "learning_rate": 3.312377980306153e-06,
      "loss": 0.4163,
      "step": 23572
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7324402332305908,
      "learning_rate": 3.311620850857049e-06,
      "loss": 1.0052,
      "step": 23573
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4359012842178345,
      "learning_rate": 3.310863790777136e-06,
      "loss": 1.0402,
      "step": 23574
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5769877433776855,
      "learning_rate": 3.310106800074261e-06,
      "loss": 0.4815,
      "step": 23575
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4725677967071533,
      "learning_rate": 3.3093498787562726e-06,
      "loss": 0.3506,
      "step": 23576
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5804736614227295,
      "learning_rate": 3.3085930268310264e-06,
      "loss": 0.4366,
      "step": 23577
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6018550395965576,
      "learning_rate": 3.3078362443063694e-06,
      "loss": 0.3914,
      "step": 23578
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5556408166885376,
      "learning_rate": 3.3070795311901547e-06,
      "loss": 0.4264,
      "step": 23579
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5669972896575928,
      "learning_rate": 3.3063228874902285e-06,
      "loss": 0.3847,
      "step": 23580
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7787586450576782,
      "learning_rate": 3.3055663132144355e-06,
      "loss": 0.4159,
      "step": 23581
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5070443153381348,
      "learning_rate": 3.3048098083706236e-06,
      "loss": 0.3858,
      "step": 23582
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6192779541015625,
      "learning_rate": 3.3040533729666434e-06,
      "loss": 0.437,
      "step": 23583
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5589337348937988,
      "learning_rate": 3.3032970070103343e-06,
      "loss": 0.4003,
      "step": 23584
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5658198595046997,
      "learning_rate": 3.3025407105095464e-06,
      "loss": 0.4517,
      "step": 23585
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6067559719085693,
      "learning_rate": 3.301784483472118e-06,
      "loss": 0.4078,
      "step": 23586
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5006201267242432,
      "learning_rate": 3.3010283259058996e-06,
      "loss": 0.4521,
      "step": 23587
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.483638882637024,
      "learning_rate": 3.3002722378187246e-06,
      "loss": 0.404,
      "step": 23588
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.642838478088379,
      "learning_rate": 3.2995162192184417e-06,
      "loss": 0.4751,
      "step": 23589
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5207912921905518,
      "learning_rate": 3.2987602701128927e-06,
      "loss": 0.3822,
      "step": 23590
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0618255138397217,
      "learning_rate": 3.2980043905099156e-06,
      "loss": 0.4442,
      "step": 23591
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6249421834945679,
      "learning_rate": 3.297248580417347e-06,
      "loss": 0.3897,
      "step": 23592
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4641045331954956,
      "learning_rate": 3.2964928398430287e-06,
      "loss": 0.404,
      "step": 23593
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5364890098571777,
      "learning_rate": 3.2957371687948013e-06,
      "loss": 0.3941,
      "step": 23594
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5699396133422852,
      "learning_rate": 3.2949815672805017e-06,
      "loss": 0.4553,
      "step": 23595
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.483839988708496,
      "learning_rate": 3.2942260353079615e-06,
      "loss": 0.3803,
      "step": 23596
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6272095441818237,
      "learning_rate": 3.29347057288502e-06,
      "loss": 0.3885,
      "step": 23597
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5669939517974854,
      "learning_rate": 3.2927151800195134e-06,
      "loss": 0.3913,
      "step": 23598
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5100927352905273,
      "learning_rate": 3.291959856719279e-06,
      "loss": 0.4069,
      "step": 23599
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.58340322971344,
      "learning_rate": 3.2912046029921474e-06,
      "loss": 0.3982,
      "step": 23600
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8364357948303223,
      "learning_rate": 3.290449418845948e-06,
      "loss": 0.9648,
      "step": 23601
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4797134399414062,
      "learning_rate": 3.2896943042885186e-06,
      "loss": 0.9955,
      "step": 23602
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.552498698234558,
      "learning_rate": 3.2889392593276927e-06,
      "loss": 0.4514,
      "step": 23603
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4991320371627808,
      "learning_rate": 3.288184283971294e-06,
      "loss": 0.3992,
      "step": 23604
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.601272463798523,
      "learning_rate": 3.2874293782271604e-06,
      "loss": 0.4121,
      "step": 23605
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5193681716918945,
      "learning_rate": 3.2866745421031152e-06,
      "loss": 0.4021,
      "step": 23606
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5105245113372803,
      "learning_rate": 3.285919775606994e-06,
      "loss": 0.436,
      "step": 23607
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4578698873519897,
      "learning_rate": 3.2851650787466173e-06,
      "loss": 0.3739,
      "step": 23608
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5740766525268555,
      "learning_rate": 3.284410451529816e-06,
      "loss": 0.4626,
      "step": 23609
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5077542066574097,
      "learning_rate": 3.2836558939644203e-06,
      "loss": 0.3988,
      "step": 23610
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.621546983718872,
      "learning_rate": 3.2829014060582498e-06,
      "loss": 0.4498,
      "step": 23611
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4377801418304443,
      "learning_rate": 3.282146987819137e-06,
      "loss": 0.3632,
      "step": 23612
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5743216276168823,
      "learning_rate": 3.2813926392548978e-06,
      "loss": 0.4358,
      "step": 23613
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5198999643325806,
      "learning_rate": 3.2806383603733618e-06,
      "loss": 0.3862,
      "step": 23614
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5997549295425415,
      "learning_rate": 3.279884151182353e-06,
      "loss": 0.4516,
      "step": 23615
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5109357833862305,
      "learning_rate": 3.2791300116896917e-06,
      "loss": 0.3902,
      "step": 23616
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.501821517944336,
      "learning_rate": 3.278375941903197e-06,
      "loss": 0.4771,
      "step": 23617
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6865648031234741,
      "learning_rate": 3.2776219418306908e-06,
      "loss": 0.3637,
      "step": 23618
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5318262577056885,
      "learning_rate": 3.2768680114799954e-06,
      "loss": 0.3899,
      "step": 23619
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.578922986984253,
      "learning_rate": 3.2761141508589335e-06,
      "loss": 0.4154,
      "step": 23620
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6831786632537842,
      "learning_rate": 3.2753603599753194e-06,
      "loss": 0.4482,
      "step": 23621
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.52968168258667,
      "learning_rate": 3.2746066388369678e-06,
      "loss": 0.3669,
      "step": 23622
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7628791332244873,
      "learning_rate": 3.2738529874517e-06,
      "loss": 0.4631,
      "step": 23623
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5916036367416382,
      "learning_rate": 3.273099405827336e-06,
      "loss": 0.3659,
      "step": 23624
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5312936305999756,
      "learning_rate": 3.2723458939716844e-06,
      "loss": 0.4281,
      "step": 23625
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5082125663757324,
      "learning_rate": 3.2715924518925677e-06,
      "loss": 0.3886,
      "step": 23626
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5255353450775146,
      "learning_rate": 3.2708390795977927e-06,
      "loss": 0.466,
      "step": 23627
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7366124391555786,
      "learning_rate": 3.270085777095181e-06,
      "loss": 0.4185,
      "step": 23628
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.520816445350647,
      "learning_rate": 3.2693325443925384e-06,
      "loss": 0.4351,
      "step": 23629
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5923328399658203,
      "learning_rate": 3.268579381497682e-06,
      "loss": 0.3961,
      "step": 23630
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8445438146591187,
      "learning_rate": 3.26782628841842e-06,
      "loss": 0.4169,
      "step": 23631
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4927804470062256,
      "learning_rate": 3.267073265162566e-06,
      "loss": 0.3846,
      "step": 23632
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8514610528945923,
      "learning_rate": 3.266320311737927e-06,
      "loss": 0.9502,
      "step": 23633
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4382686614990234,
      "learning_rate": 3.265567428152313e-06,
      "loss": 0.9787,
      "step": 23634
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5324031114578247,
      "learning_rate": 3.2648146144135375e-06,
      "loss": 0.4341,
      "step": 23635
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5346161127090454,
      "learning_rate": 3.2640618705294047e-06,
      "loss": 0.3754,
      "step": 23636
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6666370630264282,
      "learning_rate": 3.263309196507718e-06,
      "loss": 0.4606,
      "step": 23637
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4808342456817627,
      "learning_rate": 3.2625565923562875e-06,
      "loss": 0.3618,
      "step": 23638
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5996729135513306,
      "learning_rate": 3.261804058082919e-06,
      "loss": 0.4324,
      "step": 23639
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5707969665527344,
      "learning_rate": 3.26105159369542e-06,
      "loss": 0.3811,
      "step": 23640
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5298160314559937,
      "learning_rate": 3.260299199201593e-06,
      "loss": 0.4323,
      "step": 23641
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5967297554016113,
      "learning_rate": 3.2595468746092372e-06,
      "loss": 0.393,
      "step": 23642
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8584519624710083,
      "learning_rate": 3.258794619926159e-06,
      "loss": 0.9462,
      "step": 23643
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.440670371055603,
      "learning_rate": 3.2580424351601624e-06,
      "loss": 0.9715,
      "step": 23644
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5764890909194946,
      "learning_rate": 3.2572903203190453e-06,
      "loss": 0.439,
      "step": 23645
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.533019781112671,
      "learning_rate": 3.256538275410611e-06,
      "loss": 0.3806,
      "step": 23646
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.580928921699524,
      "learning_rate": 3.255786300442656e-06,
      "loss": 0.4663,
      "step": 23647
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6288909912109375,
      "learning_rate": 3.2550343954229846e-06,
      "loss": 0.4337,
      "step": 23648
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5508716106414795,
      "learning_rate": 3.25428256035939e-06,
      "loss": 0.469,
      "step": 23649
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4869177341461182,
      "learning_rate": 3.2535307952596705e-06,
      "loss": 0.3907,
      "step": 23650
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5003381967544556,
      "learning_rate": 3.252779100131629e-06,
      "loss": 0.3754,
      "step": 23651
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4823882579803467,
      "learning_rate": 3.252027474983057e-06,
      "loss": 0.3705,
      "step": 23652
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5986253023147583,
      "learning_rate": 3.2512759198217483e-06,
      "loss": 0.5161,
      "step": 23653
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4961293935775757,
      "learning_rate": 3.2505244346554997e-06,
      "loss": 0.3832,
      "step": 23654
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.649343490600586,
      "learning_rate": 3.2497730194921052e-06,
      "loss": 0.3796,
      "step": 23655
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.94402015209198,
      "learning_rate": 3.2490216743393623e-06,
      "loss": 0.3886,
      "step": 23656
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.498320460319519,
      "learning_rate": 3.24827039920506e-06,
      "loss": 0.4111,
      "step": 23657
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5014783143997192,
      "learning_rate": 3.2475191940969876e-06,
      "loss": 0.3835,
      "step": 23658
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7142674922943115,
      "learning_rate": 3.246768059022938e-06,
      "loss": 0.4178,
      "step": 23659
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5729936361312866,
      "learning_rate": 3.246016993990706e-06,
      "loss": 0.3951,
      "step": 23660
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5197077989578247,
      "learning_rate": 3.245265999008075e-06,
      "loss": 0.4063,
      "step": 23661
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.510690689086914,
      "learning_rate": 3.2445150740828393e-06,
      "loss": 0.3921,
      "step": 23662
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.516234040260315,
      "learning_rate": 3.2437642192227825e-06,
      "loss": 0.4376,
      "step": 23663
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5323827266693115,
      "learning_rate": 3.243013434435698e-06,
      "loss": 0.3795,
      "step": 23664
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6773854494094849,
      "learning_rate": 3.2422627197293655e-06,
      "loss": 0.4697,
      "step": 23665
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5813992023468018,
      "learning_rate": 3.2415120751115747e-06,
      "loss": 0.3742,
      "step": 23666
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7312756776809692,
      "learning_rate": 3.240761500590114e-06,
      "loss": 0.47,
      "step": 23667
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0724964141845703,
      "learning_rate": 3.2400109961727655e-06,
      "loss": 0.3682,
      "step": 23668
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.467750072479248,
      "learning_rate": 3.239260561867309e-06,
      "loss": 0.3842,
      "step": 23669
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5415066480636597,
      "learning_rate": 3.2385101976815325e-06,
      "loss": 0.3766,
      "step": 23670
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5151418447494507,
      "learning_rate": 3.23775990362322e-06,
      "loss": 0.4013,
      "step": 23671
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5309582948684692,
      "learning_rate": 3.2370096797001504e-06,
      "loss": 0.386,
      "step": 23672
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5053668022155762,
      "learning_rate": 3.2362595259201025e-06,
      "loss": 0.4164,
      "step": 23673
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6012773513793945,
      "learning_rate": 3.2355094422908583e-06,
      "loss": 0.3823,
      "step": 23674
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6108207702636719,
      "learning_rate": 3.234759428820198e-06,
      "loss": 0.4531,
      "step": 23675
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6001646518707275,
      "learning_rate": 3.2340094855159043e-06,
      "loss": 0.412,
      "step": 23676
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5243808031082153,
      "learning_rate": 3.233259612385752e-06,
      "loss": 0.439,
      "step": 23677
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.468639612197876,
      "learning_rate": 3.232509809437514e-06,
      "loss": 0.3704,
      "step": 23678
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5339235067367554,
      "learning_rate": 3.2317600766789716e-06,
      "loss": 0.4268,
      "step": 23679
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.533494234085083,
      "learning_rate": 3.231010414117903e-06,
      "loss": 0.3922,
      "step": 23680
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5119483470916748,
      "learning_rate": 3.230260821762078e-06,
      "loss": 0.424,
      "step": 23681
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5368072986602783,
      "learning_rate": 3.229511299619277e-06,
      "loss": 0.3965,
      "step": 23682
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5828540325164795,
      "learning_rate": 3.2287618476972658e-06,
      "loss": 0.4544,
      "step": 23683
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5791600942611694,
      "learning_rate": 3.228012466003827e-06,
      "loss": 0.3911,
      "step": 23684
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5925174951553345,
      "learning_rate": 3.2272631545467237e-06,
      "loss": 0.4282,
      "step": 23685
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.559739351272583,
      "learning_rate": 3.226513913333732e-06,
      "loss": 0.3612,
      "step": 23686
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5299322605133057,
      "learning_rate": 3.2257647423726245e-06,
      "loss": 0.4011,
      "step": 23687
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4627323150634766,
      "learning_rate": 3.2250156416711698e-06,
      "loss": 0.3576,
      "step": 23688
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5007376670837402,
      "learning_rate": 3.2242666112371344e-06,
      "loss": 0.3888,
      "step": 23689
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8803074359893799,
      "learning_rate": 3.2235176510782886e-06,
      "loss": 0.3999,
      "step": 23690
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4869462251663208,
      "learning_rate": 3.2227687612024006e-06,
      "loss": 0.452,
      "step": 23691
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4936696290969849,
      "learning_rate": 3.2220199416172406e-06,
      "loss": 0.3893,
      "step": 23692
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5279256105422974,
      "learning_rate": 3.221271192330573e-06,
      "loss": 0.4344,
      "step": 23693
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.500049114227295,
      "learning_rate": 3.2205225133501593e-06,
      "loss": 0.3766,
      "step": 23694
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5868313312530518,
      "learning_rate": 3.2197739046837685e-06,
      "loss": 0.4375,
      "step": 23695
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5091571807861328,
      "learning_rate": 3.219025366339168e-06,
      "loss": 0.3762,
      "step": 23696
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5162198543548584,
      "learning_rate": 3.218276898324114e-06,
      "loss": 0.4173,
      "step": 23697
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5964206457138062,
      "learning_rate": 3.2175285006463765e-06,
      "loss": 0.374,
      "step": 23698
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5392951965332031,
      "learning_rate": 3.2167801733137103e-06,
      "loss": 0.431,
      "step": 23699
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.563281536102295,
      "learning_rate": 3.216031916333885e-06,
      "loss": 0.4192,
      "step": 23700
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4749656915664673,
      "learning_rate": 3.2152837297146524e-06,
      "loss": 0.4132,
      "step": 23701
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5800731182098389,
      "learning_rate": 3.2145356134637774e-06,
      "loss": 0.4232,
      "step": 23702
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5157444477081299,
      "learning_rate": 3.2137875675890216e-06,
      "loss": 0.419,
      "step": 23703
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4709733724594116,
      "learning_rate": 3.2130395920981415e-06,
      "loss": 0.393,
      "step": 23704
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6068867444992065,
      "learning_rate": 3.2122916869988897e-06,
      "loss": 0.4691,
      "step": 23705
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5014886856079102,
      "learning_rate": 3.211543852299027e-06,
      "loss": 0.3816,
      "step": 23706
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.561862587928772,
      "learning_rate": 3.2107960880063096e-06,
      "loss": 0.4064,
      "step": 23707
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5341726541519165,
      "learning_rate": 3.2100483941284997e-06,
      "loss": 0.3762,
      "step": 23708
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.487513542175293,
      "learning_rate": 3.2093007706733403e-06,
      "loss": 0.378,
      "step": 23709
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4598063230514526,
      "learning_rate": 3.20855321764859e-06,
      "loss": 0.374,
      "step": 23710
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5943644046783447,
      "learning_rate": 3.207805735062003e-06,
      "loss": 0.4266,
      "step": 23711
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5078527927398682,
      "learning_rate": 3.2070583229213347e-06,
      "loss": 0.3685,
      "step": 23712
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5919699668884277,
      "learning_rate": 3.206310981234335e-06,
      "loss": 0.4312,
      "step": 23713
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.504302978515625,
      "learning_rate": 3.20556371000875e-06,
      "loss": 0.3683,
      "step": 23714
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.52440345287323,
      "learning_rate": 3.2048165092523344e-06,
      "loss": 0.4137,
      "step": 23715
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6493265628814697,
      "learning_rate": 3.2040693789728407e-06,
      "loss": 0.3956,
      "step": 23716
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.517173171043396,
      "learning_rate": 3.2033223191780117e-06,
      "loss": 0.425,
      "step": 23717
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5129832029342651,
      "learning_rate": 3.2025753298756023e-06,
      "loss": 0.3821,
      "step": 23718
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.514397144317627,
      "learning_rate": 3.2018284110733533e-06,
      "loss": 0.4313,
      "step": 23719
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5051270723342896,
      "learning_rate": 3.201081562779017e-06,
      "loss": 0.36,
      "step": 23720
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6198753118515015,
      "learning_rate": 3.200334785000334e-06,
      "loss": 0.4237,
      "step": 23721
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5263288021087646,
      "learning_rate": 3.1995880777450515e-06,
      "loss": 0.3727,
      "step": 23722
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6311246156692505,
      "learning_rate": 3.19884144102092e-06,
      "loss": 0.4597,
      "step": 23723
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5561801195144653,
      "learning_rate": 3.198094874835678e-06,
      "loss": 0.3634,
      "step": 23724
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5176912546157837,
      "learning_rate": 3.197348379197065e-06,
      "loss": 0.4537,
      "step": 23725
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5725340843200684,
      "learning_rate": 3.1966019541128266e-06,
      "loss": 0.3628,
      "step": 23726
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5212815999984741,
      "learning_rate": 3.195855599590707e-06,
      "loss": 0.4543,
      "step": 23727
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6356160640716553,
      "learning_rate": 3.195109315638447e-06,
      "loss": 0.3866,
      "step": 23728
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5450140237808228,
      "learning_rate": 3.1943631022637856e-06,
      "loss": 0.413,
      "step": 23729
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5164060592651367,
      "learning_rate": 3.1936169594744583e-06,
      "loss": 0.3983,
      "step": 23730
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5683382749557495,
      "learning_rate": 3.1928708872782066e-06,
      "loss": 0.4549,
      "step": 23731
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5677579641342163,
      "learning_rate": 3.1921248856827736e-06,
      "loss": 0.3819,
      "step": 23732
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6230651140213013,
      "learning_rate": 3.1913789546958873e-06,
      "loss": 0.4274,
      "step": 23733
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5572171211242676,
      "learning_rate": 3.190633094325293e-06,
      "loss": 0.407,
      "step": 23734
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6184775829315186,
      "learning_rate": 3.189887304578718e-06,
      "loss": 0.44,
      "step": 23735
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5107362270355225,
      "learning_rate": 3.189141585463905e-06,
      "loss": 0.3572,
      "step": 23736
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.489324688911438,
      "learning_rate": 3.188395936988582e-06,
      "loss": 0.4122,
      "step": 23737
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5156863927841187,
      "learning_rate": 3.187650359160485e-06,
      "loss": 0.375,
      "step": 23738
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.574853777885437,
      "learning_rate": 3.1869048519873514e-06,
      "loss": 0.4273,
      "step": 23739
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.557252049446106,
      "learning_rate": 3.1861594154769083e-06,
      "loss": 0.39,
      "step": 23740
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5186917781829834,
      "learning_rate": 3.1854140496368846e-06,
      "loss": 0.4104,
      "step": 23741
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5431864261627197,
      "learning_rate": 3.1846687544750145e-06,
      "loss": 0.394,
      "step": 23742
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6670395135879517,
      "learning_rate": 3.1839235299990278e-06,
      "loss": 0.4302,
      "step": 23743
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5050426721572876,
      "learning_rate": 3.1831783762166602e-06,
      "loss": 0.3668,
      "step": 23744
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.531017541885376,
      "learning_rate": 3.182433293135627e-06,
      "loss": 0.3839,
      "step": 23745
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4853843450546265,
      "learning_rate": 3.181688280763662e-06,
      "loss": 0.397,
      "step": 23746
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6226329803466797,
      "learning_rate": 3.180943339108493e-06,
      "loss": 0.4458,
      "step": 23747
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5617605447769165,
      "learning_rate": 3.1801984681778486e-06,
      "loss": 0.3904,
      "step": 23748
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5381654500961304,
      "learning_rate": 3.1794536679794517e-06,
      "loss": 0.4227,
      "step": 23749
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.470583438873291,
      "learning_rate": 3.1787089385210245e-06,
      "loss": 0.3669,
      "step": 23750
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5196723937988281,
      "learning_rate": 3.177964279810293e-06,
      "loss": 0.4098,
      "step": 23751
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5259082317352295,
      "learning_rate": 3.1772196918549826e-06,
      "loss": 0.4043,
      "step": 23752
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5360326766967773,
      "learning_rate": 3.176475174662812e-06,
      "loss": 0.4526,
      "step": 23753
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5276572704315186,
      "learning_rate": 3.1757307282415094e-06,
      "loss": 0.3847,
      "step": 23754
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5017716884613037,
      "learning_rate": 3.1749863525987868e-06,
      "loss": 0.4037,
      "step": 23755
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6039961576461792,
      "learning_rate": 3.174242047742374e-06,
      "loss": 0.3808,
      "step": 23756
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.511942982673645,
      "learning_rate": 3.173497813679982e-06,
      "loss": 0.4119,
      "step": 23757
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.512451410293579,
      "learning_rate": 3.1727536504193344e-06,
      "loss": 0.365,
      "step": 23758
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0466084480285645,
      "learning_rate": 3.172009557968152e-06,
      "loss": 0.415,
      "step": 23759
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.434882164001465,
      "learning_rate": 3.171265536334148e-06,
      "loss": 0.3976,
      "step": 23760
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6740440130233765,
      "learning_rate": 3.1705215855250383e-06,
      "loss": 0.4787,
      "step": 23761
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.580608606338501,
      "learning_rate": 3.1697777055485403e-06,
      "loss": 0.3888,
      "step": 23762
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.739425539970398,
      "learning_rate": 3.169033896412369e-06,
      "loss": 0.4719,
      "step": 23763
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4980441331863403,
      "learning_rate": 3.168290158124243e-06,
      "loss": 0.369,
      "step": 23764
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.742055058479309,
      "learning_rate": 3.167546490691873e-06,
      "loss": 0.9538,
      "step": 23765
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4773489236831665,
      "learning_rate": 3.1668028941229668e-06,
      "loss": 1.0178,
      "step": 23766
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7025763988494873,
      "learning_rate": 3.1660593684252428e-06,
      "loss": 0.9831,
      "step": 23767
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4977613687515259,
      "learning_rate": 3.1653159136064138e-06,
      "loss": 1.039,
      "step": 23768
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5647740364074707,
      "learning_rate": 3.164572529674185e-06,
      "loss": 0.4215,
      "step": 23769
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5719681978225708,
      "learning_rate": 3.1638292166362727e-06,
      "loss": 0.3761,
      "step": 23770
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5648770332336426,
      "learning_rate": 3.1630859745003796e-06,
      "loss": 0.4298,
      "step": 23771
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5419628620147705,
      "learning_rate": 3.162342803274221e-06,
      "loss": 0.3923,
      "step": 23772
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.743048906326294,
      "learning_rate": 3.161599702965499e-06,
      "loss": 0.4542,
      "step": 23773
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4746230840682983,
      "learning_rate": 3.1608566735819226e-06,
      "loss": 0.3839,
      "step": 23774
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5185362100601196,
      "learning_rate": 3.1601137151312024e-06,
      "loss": 0.4444,
      "step": 23775
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.507079005241394,
      "learning_rate": 3.15937082762104e-06,
      "loss": 0.39,
      "step": 23776
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7215458154678345,
      "learning_rate": 3.158628011059137e-06,
      "loss": 0.4813,
      "step": 23777
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.566127061843872,
      "learning_rate": 3.157885265453202e-06,
      "loss": 0.3898,
      "step": 23778
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6344696283340454,
      "learning_rate": 3.1571425908109387e-06,
      "loss": 0.4257,
      "step": 23779
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.535262107849121,
      "learning_rate": 3.156399987140052e-06,
      "loss": 0.3861,
      "step": 23780
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5564677715301514,
      "learning_rate": 3.15565745444824e-06,
      "loss": 0.4169,
      "step": 23781
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4958268404006958,
      "learning_rate": 3.154914992743202e-06,
      "loss": 0.3803,
      "step": 23782
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6898677349090576,
      "learning_rate": 3.1541726020326423e-06,
      "loss": 0.4178,
      "step": 23783
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6570813655853271,
      "learning_rate": 3.1534302823242625e-06,
      "loss": 0.3814,
      "step": 23784
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6668802499771118,
      "learning_rate": 3.1526880336257583e-06,
      "loss": 0.4494,
      "step": 23785
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.582251787185669,
      "learning_rate": 3.1519458559448257e-06,
      "loss": 0.3895,
      "step": 23786
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6016335487365723,
      "learning_rate": 3.1512037492891647e-06,
      "loss": 0.4399,
      "step": 23787
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7126169204711914,
      "learning_rate": 3.150461713666476e-06,
      "loss": 0.3954,
      "step": 23788
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5843119621276855,
      "learning_rate": 3.14971974908445e-06,
      "loss": 0.4306,
      "step": 23789
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6055927276611328,
      "learning_rate": 3.1489778555507866e-06,
      "loss": 0.3882,
      "step": 23790
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.519063115119934,
      "learning_rate": 3.1482360330731755e-06,
      "loss": 0.476,
      "step": 23791
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5551533699035645,
      "learning_rate": 3.1474942816593167e-06,
      "loss": 0.3717,
      "step": 23792
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.814868450164795,
      "learning_rate": 3.1467526013168958e-06,
      "loss": 0.3807,
      "step": 23793
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4890366792678833,
      "learning_rate": 3.1460109920536097e-06,
      "loss": 0.392,
      "step": 23794
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5542879104614258,
      "learning_rate": 3.1452694538771533e-06,
      "loss": 0.4151,
      "step": 23795
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4982768297195435,
      "learning_rate": 3.144527986795214e-06,
      "loss": 0.3757,
      "step": 23796
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5781794786453247,
      "learning_rate": 3.1437865908154773e-06,
      "loss": 0.4328,
      "step": 23797
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4887418746948242,
      "learning_rate": 3.143045265945639e-06,
      "loss": 0.3773,
      "step": 23798
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.503604769706726,
      "learning_rate": 3.142304012193386e-06,
      "loss": 0.4394,
      "step": 23799
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.54268217086792,
      "learning_rate": 3.141562829566408e-06,
      "loss": 0.3993,
      "step": 23800
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5908797979354858,
      "learning_rate": 3.140821718072392e-06,
      "loss": 0.4374,
      "step": 23801
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4597772359848022,
      "learning_rate": 3.1400806777190195e-06,
      "loss": 0.3505,
      "step": 23802
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5623644590377808,
      "learning_rate": 3.139339708513981e-06,
      "loss": 0.4461,
      "step": 23803
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6125433444976807,
      "learning_rate": 3.1385988104649623e-06,
      "loss": 0.4082,
      "step": 23804
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4810631275177002,
      "learning_rate": 3.137857983579643e-06,
      "loss": 0.4287,
      "step": 23805
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5673414468765259,
      "learning_rate": 3.1371172278657138e-06,
      "loss": 0.4049,
      "step": 23806
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4396072626113892,
      "learning_rate": 3.1363765433308493e-06,
      "loss": 0.3912,
      "step": 23807
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4691359996795654,
      "learning_rate": 3.1356359299827377e-06,
      "loss": 0.3671,
      "step": 23808
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.536646842956543,
      "learning_rate": 3.1348953878290557e-06,
      "loss": 0.4014,
      "step": 23809
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.501993179321289,
      "learning_rate": 3.1341549168774867e-06,
      "loss": 0.388,
      "step": 23810
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6549746990203857,
      "learning_rate": 3.1334145171357133e-06,
      "loss": 0.4122,
      "step": 23811
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.482059359550476,
      "learning_rate": 3.132674188611412e-06,
      "loss": 0.3683,
      "step": 23812
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5973087549209595,
      "learning_rate": 3.131933931312258e-06,
      "loss": 0.4392,
      "step": 23813
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4726873636245728,
      "learning_rate": 3.1311937452459295e-06,
      "loss": 0.3695,
      "step": 23814
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.659861445426941,
      "learning_rate": 3.1304536304201072e-06,
      "loss": 0.4598,
      "step": 23815
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9548481702804565,
      "learning_rate": 3.129713586842469e-06,
      "loss": 0.3842,
      "step": 23816
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6393375396728516,
      "learning_rate": 3.128973614520687e-06,
      "loss": 0.4408,
      "step": 23817
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4410959482192993,
      "learning_rate": 3.128233713462433e-06,
      "loss": 0.3601,
      "step": 23818
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.472554326057434,
      "learning_rate": 3.127493883675383e-06,
      "loss": 0.3869,
      "step": 23819
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5843642950057983,
      "learning_rate": 3.126754125167215e-06,
      "loss": 0.3952,
      "step": 23820
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.562568187713623,
      "learning_rate": 3.126014437945597e-06,
      "loss": 0.4544,
      "step": 23821
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.56877601146698,
      "learning_rate": 3.1252748220181984e-06,
      "loss": 0.385,
      "step": 23822
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6048731803894043,
      "learning_rate": 3.124535277392694e-06,
      "loss": 0.4918,
      "step": 23823
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5866100788116455,
      "learning_rate": 3.1237958040767557e-06,
      "loss": 0.3992,
      "step": 23824
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6193888187408447,
      "learning_rate": 3.1230564020780476e-06,
      "loss": 0.4223,
      "step": 23825
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.418427586555481,
      "learning_rate": 3.122317071404244e-06,
      "loss": 0.365,
      "step": 23826
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.530943751335144,
      "learning_rate": 3.121577812063007e-06,
      "loss": 0.4535,
      "step": 23827
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5265412330627441,
      "learning_rate": 3.12083862406201e-06,
      "loss": 0.391,
      "step": 23828
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4398555755615234,
      "learning_rate": 3.1200995074089157e-06,
      "loss": 0.3915,
      "step": 23829
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.459348440170288,
      "learning_rate": 3.119360462111389e-06,
      "loss": 0.3716,
      "step": 23830
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5224655866622925,
      "learning_rate": 3.1186214881771016e-06,
      "loss": 0.3927,
      "step": 23831
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.494553565979004,
      "learning_rate": 3.117882585613713e-06,
      "loss": 0.3638,
      "step": 23832
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6585462093353271,
      "learning_rate": 3.117143754428883e-06,
      "loss": 0.4246,
      "step": 23833
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5443379878997803,
      "learning_rate": 3.116404994630279e-06,
      "loss": 0.3959,
      "step": 23834
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6801389455795288,
      "learning_rate": 3.115666306225562e-06,
      "loss": 0.4544,
      "step": 23835
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4891830682754517,
      "learning_rate": 3.114927689222399e-06,
      "loss": 0.3837,
      "step": 23836
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4975217580795288,
      "learning_rate": 3.1141891436284445e-06,
      "loss": 0.4048,
      "step": 23837
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4840993881225586,
      "learning_rate": 3.1134506694513555e-06,
      "loss": 0.3691,
      "step": 23838
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5944350957870483,
      "learning_rate": 3.1127122666987964e-06,
      "loss": 0.4404,
      "step": 23839
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5172171592712402,
      "learning_rate": 3.1119739353784272e-06,
      "loss": 0.3822,
      "step": 23840
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5772664546966553,
      "learning_rate": 3.1112356754979002e-06,
      "loss": 0.4273,
      "step": 23841
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5308152437210083,
      "learning_rate": 3.1104974870648772e-06,
      "loss": 0.4057,
      "step": 23842
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5766679048538208,
      "learning_rate": 3.10975937008701e-06,
      "loss": 0.4449,
      "step": 23843
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5754694938659668,
      "learning_rate": 3.1090213245719582e-06,
      "loss": 0.3929,
      "step": 23844
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5078126192092896,
      "learning_rate": 3.1082833505273724e-06,
      "loss": 0.4338,
      "step": 23845
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5241471529006958,
      "learning_rate": 3.107545447960908e-06,
      "loss": 0.3775,
      "step": 23846
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8284454345703125,
      "learning_rate": 3.106807616880222e-06,
      "loss": 0.4467,
      "step": 23847
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.472806453704834,
      "learning_rate": 3.1060698572929637e-06,
      "loss": 0.3874,
      "step": 23848
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6107254028320312,
      "learning_rate": 3.1053321692067818e-06,
      "loss": 0.42,
      "step": 23849
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5947104692459106,
      "learning_rate": 3.1045945526293307e-06,
      "loss": 0.3982,
      "step": 23850
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4974790811538696,
      "learning_rate": 3.1038570075682605e-06,
      "loss": 0.4173,
      "step": 23851
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5160750150680542,
      "learning_rate": 3.1031195340312237e-06,
      "loss": 0.38,
      "step": 23852
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6610568761825562,
      "learning_rate": 3.1023821320258664e-06,
      "loss": 0.4397,
      "step": 23853
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5078930854797363,
      "learning_rate": 3.1016448015598323e-06,
      "loss": 0.3673,
      "step": 23854
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.353395938873291,
      "learning_rate": 3.100907542640773e-06,
      "loss": 0.5053,
      "step": 23855
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.534163236618042,
      "learning_rate": 3.1001703552763353e-06,
      "loss": 0.3969,
      "step": 23856
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7132095098495483,
      "learning_rate": 3.099433239474167e-06,
      "loss": 0.4431,
      "step": 23857
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.651406168937683,
      "learning_rate": 3.0986961952419104e-06,
      "loss": 0.4119,
      "step": 23858
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5406936407089233,
      "learning_rate": 3.0979592225872068e-06,
      "loss": 0.3899,
      "step": 23859
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5672383308410645,
      "learning_rate": 3.0972223215177033e-06,
      "loss": 0.3735,
      "step": 23860
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5353013277053833,
      "learning_rate": 3.096485492041045e-06,
      "loss": 0.4588,
      "step": 23861
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4612315893173218,
      "learning_rate": 3.095748734164872e-06,
      "loss": 0.3596,
      "step": 23862
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5526998043060303,
      "learning_rate": 3.0950120478968213e-06,
      "loss": 0.428,
      "step": 23863
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4847962856292725,
      "learning_rate": 3.094275433244538e-06,
      "loss": 0.3782,
      "step": 23864
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4824973344802856,
      "learning_rate": 3.0935388902156647e-06,
      "loss": 0.4203,
      "step": 23865
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5406146049499512,
      "learning_rate": 3.092802418817833e-06,
      "loss": 0.4007,
      "step": 23866
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5553497076034546,
      "learning_rate": 3.0920660190586893e-06,
      "loss": 0.4335,
      "step": 23867
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.679490566253662,
      "learning_rate": 3.0913296909458633e-06,
      "loss": 0.4111,
      "step": 23868
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5391435623168945,
      "learning_rate": 3.0905934344870005e-06,
      "loss": 0.4213,
      "step": 23869
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4781904220581055,
      "learning_rate": 3.0898572496897284e-06,
      "loss": 0.3628,
      "step": 23870
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5223619937896729,
      "learning_rate": 3.089121136561686e-06,
      "loss": 0.4427,
      "step": 23871
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6252890825271606,
      "learning_rate": 3.088385095110512e-06,
      "loss": 0.3907,
      "step": 23872
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.556449294090271,
      "learning_rate": 3.087649125343838e-06,
      "loss": 0.4096,
      "step": 23873
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6087923049926758,
      "learning_rate": 3.0869132272692915e-06,
      "loss": 0.421,
      "step": 23874
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.627091407775879,
      "learning_rate": 3.08617740089451e-06,
      "loss": 0.4448,
      "step": 23875
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5979714393615723,
      "learning_rate": 3.085441646227124e-06,
      "loss": 0.4193,
      "step": 23876
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5076673030853271,
      "learning_rate": 3.084705963274769e-06,
      "loss": 0.4171,
      "step": 23877
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4662193059921265,
      "learning_rate": 3.0839703520450703e-06,
      "loss": 0.3639,
      "step": 23878
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.834332823753357,
      "learning_rate": 3.0832348125456557e-06,
      "loss": 0.4322,
      "step": 23879
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6153870820999146,
      "learning_rate": 3.0824993447841567e-06,
      "loss": 0.4315,
      "step": 23880
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5121461153030396,
      "learning_rate": 3.0817639487682038e-06,
      "loss": 0.4469,
      "step": 23881
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.500586986541748,
      "learning_rate": 3.081028624505419e-06,
      "loss": 0.3818,
      "step": 23882
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5297272205352783,
      "learning_rate": 3.0802933720034344e-06,
      "loss": 0.44,
      "step": 23883
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5744191408157349,
      "learning_rate": 3.0795581912698704e-06,
      "loss": 0.4117,
      "step": 23884
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6833500862121582,
      "learning_rate": 3.0788230823123554e-06,
      "loss": 0.4453,
      "step": 23885
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5857200622558594,
      "learning_rate": 3.078088045138511e-06,
      "loss": 0.3811,
      "step": 23886
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6121528148651123,
      "learning_rate": 3.0773530797559614e-06,
      "loss": 0.3935,
      "step": 23887
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4942727088928223,
      "learning_rate": 3.0766181861723344e-06,
      "loss": 0.4011,
      "step": 23888
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5912067890167236,
      "learning_rate": 3.075883364395247e-06,
      "loss": 0.4157,
      "step": 23889
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5075500011444092,
      "learning_rate": 3.0751486144323184e-06,
      "loss": 0.3902,
      "step": 23890
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5829302072525024,
      "learning_rate": 3.0744139362911717e-06,
      "loss": 0.44,
      "step": 23891
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4875110387802124,
      "learning_rate": 3.0736793299794278e-06,
      "loss": 0.363,
      "step": 23892
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5871785879135132,
      "learning_rate": 3.0729447955047078e-06,
      "loss": 0.4229,
      "step": 23893
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5760325193405151,
      "learning_rate": 3.072210332874628e-06,
      "loss": 0.3854,
      "step": 23894
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.532924771308899,
      "learning_rate": 3.071475942096801e-06,
      "loss": 0.4115,
      "step": 23895
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4911309480667114,
      "learning_rate": 3.070741623178848e-06,
      "loss": 0.4192,
      "step": 23896
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.514681339263916,
      "learning_rate": 3.0700073761283877e-06,
      "loss": 0.4202,
      "step": 23897
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4815220832824707,
      "learning_rate": 3.0692732009530325e-06,
      "loss": 0.3722,
      "step": 23898
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.528552770614624,
      "learning_rate": 3.0685390976603945e-06,
      "loss": 0.421,
      "step": 23899
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4679814577102661,
      "learning_rate": 3.0678050662580893e-06,
      "loss": 0.3752,
      "step": 23900
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5758631229400635,
      "learning_rate": 3.067071106753734e-06,
      "loss": 0.4337,
      "step": 23901
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4682338237762451,
      "learning_rate": 3.0663372191549344e-06,
      "loss": 0.3826,
      "step": 23902
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5334171056747437,
      "learning_rate": 3.0656034034693093e-06,
      "loss": 0.4542,
      "step": 23903
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4799672365188599,
      "learning_rate": 3.0648696597044614e-06,
      "loss": 0.3995,
      "step": 23904
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5645442008972168,
      "learning_rate": 3.0641359878680087e-06,
      "loss": 0.423,
      "step": 23905
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5143063068389893,
      "learning_rate": 3.0634023879675535e-06,
      "loss": 0.4134,
      "step": 23906
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5636329650878906,
      "learning_rate": 3.062668860010708e-06,
      "loss": 0.401,
      "step": 23907
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5103774070739746,
      "learning_rate": 3.061935404005083e-06,
      "loss": 0.3584,
      "step": 23908
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7718536853790283,
      "learning_rate": 3.0612020199582817e-06,
      "loss": 0.9324,
      "step": 23909
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4168859720230103,
      "learning_rate": 3.0604687078779084e-06,
      "loss": 1.0053,
      "step": 23910
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8747378587722778,
      "learning_rate": 3.0597354677715718e-06,
      "loss": 0.4724,
      "step": 23911
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6965131759643555,
      "learning_rate": 3.059002299646876e-06,
      "loss": 0.3986,
      "step": 23912
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.59589421749115,
      "learning_rate": 3.058269203511428e-06,
      "loss": 0.4064,
      "step": 23913
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5614395141601562,
      "learning_rate": 3.0575361793728296e-06,
      "loss": 0.3932,
      "step": 23914
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5305384397506714,
      "learning_rate": 3.0568032272386795e-06,
      "loss": 0.4186,
      "step": 23915
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4830855131149292,
      "learning_rate": 3.056070347116583e-06,
      "loss": 0.3718,
      "step": 23916
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.599585771560669,
      "learning_rate": 3.055337539014144e-06,
      "loss": 0.4362,
      "step": 23917
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5301496982574463,
      "learning_rate": 3.054604802938956e-06,
      "loss": 0.3863,
      "step": 23918
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5120412111282349,
      "learning_rate": 3.0538721388986258e-06,
      "loss": 0.4093,
      "step": 23919
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5014028549194336,
      "learning_rate": 3.053139546900745e-06,
      "loss": 0.3847,
      "step": 23920
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5773893594741821,
      "learning_rate": 3.0524070269529206e-06,
      "loss": 0.457,
      "step": 23921
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5836715698242188,
      "learning_rate": 3.051674579062741e-06,
      "loss": 0.374,
      "step": 23922
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5688248872756958,
      "learning_rate": 3.050942203237807e-06,
      "loss": 0.4697,
      "step": 23923
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5448288917541504,
      "learning_rate": 3.0502098994857175e-06,
      "loss": 0.385,
      "step": 23924
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5880862474441528,
      "learning_rate": 3.049477667814065e-06,
      "loss": 0.4487,
      "step": 23925
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5629504919052124,
      "learning_rate": 3.0487455082304396e-06,
      "loss": 0.3713,
      "step": 23926
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.510785698890686,
      "learning_rate": 3.0480134207424395e-06,
      "loss": 0.4196,
      "step": 23927
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5413378477096558,
      "learning_rate": 3.0472814053576574e-06,
      "loss": 0.3867,
      "step": 23928
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.572831630706787,
      "learning_rate": 3.0465494620836877e-06,
      "loss": 0.4562,
      "step": 23929
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6290853023529053,
      "learning_rate": 3.0458175909281183e-06,
      "loss": 0.4063,
      "step": 23930
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5421587228775024,
      "learning_rate": 3.045085791898539e-06,
      "loss": 0.4438,
      "step": 23931
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4892520904541016,
      "learning_rate": 3.0443540650025417e-06,
      "loss": 0.3866,
      "step": 23932
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5712666511535645,
      "learning_rate": 3.0436224102477175e-06,
      "loss": 0.4156,
      "step": 23933
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4665608406066895,
      "learning_rate": 3.0428908276416503e-06,
      "loss": 0.372,
      "step": 23934
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4764105081558228,
      "learning_rate": 3.042159317191933e-06,
      "loss": 0.4163,
      "step": 23935
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5984996557235718,
      "learning_rate": 3.041427878906147e-06,
      "loss": 0.4114,
      "step": 23936
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9538594484329224,
      "learning_rate": 3.040696512791883e-06,
      "loss": 0.9717,
      "step": 23937
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4714157581329346,
      "learning_rate": 3.039965218856723e-06,
      "loss": 0.9999,
      "step": 23938
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5776793956756592,
      "learning_rate": 3.0392339971082563e-06,
      "loss": 0.443,
      "step": 23939
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5808520317077637,
      "learning_rate": 3.03850284755406e-06,
      "loss": 0.4287,
      "step": 23940
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4811251163482666,
      "learning_rate": 3.037771770201725e-06,
      "loss": 0.3919,
      "step": 23941
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5522404909133911,
      "learning_rate": 3.0370407650588263e-06,
      "loss": 0.395,
      "step": 23942
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4860483407974243,
      "learning_rate": 3.03630983213295e-06,
      "loss": 0.4227,
      "step": 23943
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5271204710006714,
      "learning_rate": 3.0355789714316787e-06,
      "loss": 0.4054,
      "step": 23944
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6079946756362915,
      "learning_rate": 3.0348481829625896e-06,
      "loss": 0.4896,
      "step": 23945
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5075594186782837,
      "learning_rate": 3.0341174667332607e-06,
      "loss": 0.4009,
      "step": 23946
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5506579875946045,
      "learning_rate": 3.033386822751272e-06,
      "loss": 0.4085,
      "step": 23947
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5524356365203857,
      "learning_rate": 3.032656251024202e-06,
      "loss": 0.42,
      "step": 23948
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5398343801498413,
      "learning_rate": 3.0319257515596313e-06,
      "loss": 0.4417,
      "step": 23949
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6057695150375366,
      "learning_rate": 3.0311953243651337e-06,
      "loss": 0.3849,
      "step": 23950
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4968726634979248,
      "learning_rate": 3.0304649694482802e-06,
      "loss": 0.4333,
      "step": 23951
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5847288370132446,
      "learning_rate": 3.02973468681665e-06,
      "loss": 0.4121,
      "step": 23952
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5074559450149536,
      "learning_rate": 3.0290044764778213e-06,
      "loss": 0.4126,
      "step": 23953
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6654998064041138,
      "learning_rate": 3.0282743384393587e-06,
      "loss": 0.3904,
      "step": 23954
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5136561393737793,
      "learning_rate": 3.0275442727088445e-06,
      "loss": 0.4364,
      "step": 23955
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.539577841758728,
      "learning_rate": 3.0268142792938406e-06,
      "loss": 0.4039,
      "step": 23956
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5289676189422607,
      "learning_rate": 3.026084358201927e-06,
      "loss": 0.4208,
      "step": 23957
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6210542917251587,
      "learning_rate": 3.025354509440669e-06,
      "loss": 0.3923,
      "step": 23958
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5203253030776978,
      "learning_rate": 3.024624733017636e-06,
      "loss": 0.4262,
      "step": 23959
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.443031668663025,
      "learning_rate": 3.0238950289404024e-06,
      "loss": 0.367,
      "step": 23960
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5398422479629517,
      "learning_rate": 3.023165397216532e-06,
      "loss": 0.4198,
      "step": 23961
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5825639963150024,
      "learning_rate": 3.022435837853591e-06,
      "loss": 0.3883,
      "step": 23962
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.885208606719971,
      "learning_rate": 3.021706350859147e-06,
      "loss": 0.4754,
      "step": 23963
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.491637110710144,
      "learning_rate": 3.020976936240768e-06,
      "loss": 0.3528,
      "step": 23964
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5761456489562988,
      "learning_rate": 3.0202475940060207e-06,
      "loss": 0.4143,
      "step": 23965
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4477425813674927,
      "learning_rate": 3.019518324162467e-06,
      "loss": 0.3541,
      "step": 23966
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5410425662994385,
      "learning_rate": 3.0187891267176685e-06,
      "loss": 0.421,
      "step": 23967
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5035041570663452,
      "learning_rate": 3.0180600016791883e-06,
      "loss": 0.3843,
      "step": 23968
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5695964097976685,
      "learning_rate": 3.0173309490545953e-06,
      "loss": 0.4505,
      "step": 23969
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4835554361343384,
      "learning_rate": 3.0166019688514426e-06,
      "loss": 0.3584,
      "step": 23970
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6419869661331177,
      "learning_rate": 3.015873061077298e-06,
      "loss": 0.4255,
      "step": 23971
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.503233790397644,
      "learning_rate": 3.0151442257397147e-06,
      "loss": 0.3693,
      "step": 23972
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.453518033027649,
      "learning_rate": 3.0144154628462573e-06,
      "loss": 0.3959,
      "step": 23973
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5138345956802368,
      "learning_rate": 3.0136867724044793e-06,
      "loss": 0.3847,
      "step": 23974
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.555195689201355,
      "learning_rate": 3.012958154421942e-06,
      "loss": 0.4463,
      "step": 23975
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5066016912460327,
      "learning_rate": 3.012229608906204e-06,
      "loss": 0.382,
      "step": 23976
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.658007264137268,
      "learning_rate": 3.0115011358648192e-06,
      "loss": 0.4643,
      "step": 23977
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.472163200378418,
      "learning_rate": 3.0107727353053396e-06,
      "loss": 0.3939,
      "step": 23978
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5655595064163208,
      "learning_rate": 3.010044407235323e-06,
      "loss": 0.4711,
      "step": 23979
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5395097732543945,
      "learning_rate": 3.0093161516623272e-06,
      "loss": 0.3936,
      "step": 23980
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6119035482406616,
      "learning_rate": 3.0085879685939e-06,
      "loss": 0.4955,
      "step": 23981
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5343732833862305,
      "learning_rate": 3.0078598580375938e-06,
      "loss": 0.4041,
      "step": 23982
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4948017597198486,
      "learning_rate": 3.0071318200009616e-06,
      "loss": 0.3952,
      "step": 23983
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4720696210861206,
      "learning_rate": 3.0064038544915543e-06,
      "loss": 0.36,
      "step": 23984
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5767537355422974,
      "learning_rate": 3.005675961516925e-06,
      "loss": 0.441,
      "step": 23985
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6028186082839966,
      "learning_rate": 3.0049481410846206e-06,
      "loss": 0.3922,
      "step": 23986
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.557083010673523,
      "learning_rate": 3.0042203932021873e-06,
      "loss": 0.4431,
      "step": 23987
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5019927024841309,
      "learning_rate": 3.0034927178771745e-06,
      "loss": 0.4206,
      "step": 23988
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5056421756744385,
      "learning_rate": 3.0027651151171345e-06,
      "loss": 0.4225,
      "step": 23989
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6077642440795898,
      "learning_rate": 3.0020375849296044e-06,
      "loss": 0.3642,
      "step": 23990
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.450082540512085,
      "learning_rate": 3.001310127322139e-06,
      "loss": 0.3897,
      "step": 23991
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4678400754928589,
      "learning_rate": 3.000582742302276e-06,
      "loss": 0.3596,
      "step": 23992
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4808502197265625,
      "learning_rate": 2.9998554298775662e-06,
      "loss": 0.4071,
      "step": 23993
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5600847005844116,
      "learning_rate": 2.9991281900555457e-06,
      "loss": 0.3912,
      "step": 23994
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7141448259353638,
      "learning_rate": 2.998401022843761e-06,
      "loss": 0.4572,
      "step": 23995
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5478135347366333,
      "learning_rate": 2.997673928249757e-06,
      "loss": 0.3931,
      "step": 23996
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5577672719955444,
      "learning_rate": 2.996946906281072e-06,
      "loss": 0.4241,
      "step": 23997
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5053832530975342,
      "learning_rate": 2.996219956945242e-06,
      "loss": 0.3743,
      "step": 23998
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6530356407165527,
      "learning_rate": 2.9954930802498106e-06,
      "loss": 0.4575,
      "step": 23999
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.569066047668457,
      "learning_rate": 2.994766276202318e-06,
      "loss": 0.4108,
      "step": 24000
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5410066843032837,
      "learning_rate": 2.9940395448103034e-06,
      "loss": 0.4294,
      "step": 24001
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5381121635437012,
      "learning_rate": 2.993312886081301e-06,
      "loss": 0.4004,
      "step": 24002
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5615649223327637,
      "learning_rate": 2.992586300022846e-06,
      "loss": 0.4201,
      "step": 24003
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5359090566635132,
      "learning_rate": 2.991859786642476e-06,
      "loss": 0.3644,
      "step": 24004
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.558540940284729,
      "learning_rate": 2.991133345947731e-06,
      "loss": 0.4415,
      "step": 24005
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5008959770202637,
      "learning_rate": 2.9904069779461366e-06,
      "loss": 0.3869,
      "step": 24006
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5457727909088135,
      "learning_rate": 2.989680682645234e-06,
      "loss": 0.4044,
      "step": 24007
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4933040142059326,
      "learning_rate": 2.98895446005255e-06,
      "loss": 0.3979,
      "step": 24008
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6740893125534058,
      "learning_rate": 2.9882283101756226e-06,
      "loss": 0.448,
      "step": 24009
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.505971074104309,
      "learning_rate": 2.987502233021976e-06,
      "loss": 0.3712,
      "step": 24010
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6450600624084473,
      "learning_rate": 2.9867762285991463e-06,
      "loss": 0.4587,
      "step": 24011
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5101797580718994,
      "learning_rate": 2.9860502969146645e-06,
      "loss": 0.4066,
      "step": 24012
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.692504644393921,
      "learning_rate": 2.985324437976056e-06,
      "loss": 0.4136,
      "step": 24013
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5056055784225464,
      "learning_rate": 2.9845986517908476e-06,
      "loss": 0.3892,
      "step": 24014
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7138999700546265,
      "learning_rate": 2.98387293836657e-06,
      "loss": 0.4705,
      "step": 24015
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5524227619171143,
      "learning_rate": 2.9831472977107524e-06,
      "loss": 0.3575,
      "step": 24016
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.518298864364624,
      "learning_rate": 2.9824217298309164e-06,
      "loss": 0.4121,
      "step": 24017
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.512956976890564,
      "learning_rate": 2.9816962347345866e-06,
      "loss": 0.3839,
      "step": 24018
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5463327169418335,
      "learning_rate": 2.98097081242929e-06,
      "loss": 0.42,
      "step": 24019
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5666131973266602,
      "learning_rate": 2.9802454629225496e-06,
      "loss": 0.3824,
      "step": 24020
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.519905924797058,
      "learning_rate": 2.9795201862218927e-06,
      "loss": 0.4136,
      "step": 24021
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4839030504226685,
      "learning_rate": 2.978794982334837e-06,
      "loss": 0.3771,
      "step": 24022
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5751557350158691,
      "learning_rate": 2.978069851268902e-06,
      "loss": 0.4506,
      "step": 24023
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5258355140686035,
      "learning_rate": 2.9773447930316113e-06,
      "loss": 0.3895,
      "step": 24024
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5965325832366943,
      "learning_rate": 2.9766198076304877e-06,
      "loss": 0.4601,
      "step": 24025
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.516572117805481,
      "learning_rate": 2.9758948950730447e-06,
      "loss": 0.3785,
      "step": 24026
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5077424049377441,
      "learning_rate": 2.9751700553668074e-06,
      "loss": 0.419,
      "step": 24027
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5188932418823242,
      "learning_rate": 2.9744452885192865e-06,
      "loss": 0.3837,
      "step": 24028
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.846770167350769,
      "learning_rate": 2.9737205945380056e-06,
      "loss": 0.9908,
      "step": 24029
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4704134464263916,
      "learning_rate": 2.9729959734304734e-06,
      "loss": 0.9546,
      "step": 24030
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5051392316818237,
      "learning_rate": 2.9722714252042107e-06,
      "loss": 0.4107,
      "step": 24031
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5242033004760742,
      "learning_rate": 2.9715469498667347e-06,
      "loss": 0.3705,
      "step": 24032
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6269973516464233,
      "learning_rate": 2.970822547425556e-06,
      "loss": 0.4601,
      "step": 24033
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5008974075317383,
      "learning_rate": 2.970098217888183e-06,
      "loss": 0.3573,
      "step": 24034
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5389264822006226,
      "learning_rate": 2.969373961262134e-06,
      "loss": 0.4084,
      "step": 24035
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4509496688842773,
      "learning_rate": 2.9686497775549204e-06,
      "loss": 0.3948,
      "step": 24036
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6033774614334106,
      "learning_rate": 2.967925666774054e-06,
      "loss": 0.4364,
      "step": 24037
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5849987268447876,
      "learning_rate": 2.9672016289270442e-06,
      "loss": 0.3962,
      "step": 24038
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.551863670349121,
      "learning_rate": 2.9664776640213954e-06,
      "loss": 0.4176,
      "step": 24039
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5102763175964355,
      "learning_rate": 2.965753772064621e-06,
      "loss": 0.4242,
      "step": 24040
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5799380540847778,
      "learning_rate": 2.9650299530642323e-06,
      "loss": 0.4064,
      "step": 24041
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.482187032699585,
      "learning_rate": 2.9643062070277284e-06,
      "loss": 0.3642,
      "step": 24042
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5878921747207642,
      "learning_rate": 2.9635825339626232e-06,
      "loss": 0.4613,
      "step": 24043
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6151477098464966,
      "learning_rate": 2.9628589338764157e-06,
      "loss": 0.4031,
      "step": 24044
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5335086584091187,
      "learning_rate": 2.9621354067766173e-06,
      "loss": 0.4218,
      "step": 24045
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5249046087265015,
      "learning_rate": 2.961411952670725e-06,
      "loss": 0.3789,
      "step": 24046
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9100948572158813,
      "learning_rate": 2.960688571566246e-06,
      "loss": 0.4226,
      "step": 24047
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.494784951210022,
      "learning_rate": 2.9599652634706864e-06,
      "loss": 0.3815,
      "step": 24048
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5069994926452637,
      "learning_rate": 2.9592420283915456e-06,
      "loss": 0.4282,
      "step": 24049
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.627855896949768,
      "learning_rate": 2.9585188663363197e-06,
      "loss": 0.4307,
      "step": 24050
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6514216661453247,
      "learning_rate": 2.9577957773125133e-06,
      "loss": 0.4206,
      "step": 24051
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5093090534210205,
      "learning_rate": 2.957072761327625e-06,
      "loss": 0.3922,
      "step": 24052
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7159597873687744,
      "learning_rate": 2.9563498183891613e-06,
      "loss": 0.4314,
      "step": 24053
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4467780590057373,
      "learning_rate": 2.9556269485046072e-06,
      "loss": 0.3777,
      "step": 24054
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.576911449432373,
      "learning_rate": 2.9549041516814657e-06,
      "loss": 0.5041,
      "step": 24055
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5570175647735596,
      "learning_rate": 2.954181427927234e-06,
      "loss": 0.389,
      "step": 24056
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5331627130508423,
      "learning_rate": 2.953458777249412e-06,
      "loss": 0.4249,
      "step": 24057
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6874970197677612,
      "learning_rate": 2.952736199655488e-06,
      "loss": 0.3758,
      "step": 24058
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.549465537071228,
      "learning_rate": 2.9520136951529577e-06,
      "loss": 0.4237,
      "step": 24059
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4891023635864258,
      "learning_rate": 2.951291263749315e-06,
      "loss": 0.3761,
      "step": 24060
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5291169881820679,
      "learning_rate": 2.950568905452057e-06,
      "loss": 0.4121,
      "step": 24061
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4828219413757324,
      "learning_rate": 2.949846620268668e-06,
      "loss": 0.3648,
      "step": 24062
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5692554712295532,
      "learning_rate": 2.949124408206647e-06,
      "loss": 0.3981,
      "step": 24063
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.525033950805664,
      "learning_rate": 2.948402269273477e-06,
      "loss": 0.3359,
      "step": 24064
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6055792570114136,
      "learning_rate": 2.947680203476656e-06,
      "loss": 0.4229,
      "step": 24065
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5636268854141235,
      "learning_rate": 2.946958210823664e-06,
      "loss": 0.3916,
      "step": 24066
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6214542388916016,
      "learning_rate": 2.9462362913219944e-06,
      "loss": 0.3989,
      "step": 24067
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5675965547561646,
      "learning_rate": 2.9455144449791374e-06,
      "loss": 0.3937,
      "step": 24068
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8144961595535278,
      "learning_rate": 2.9447926718025755e-06,
      "loss": 0.4987,
      "step": 24069
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5050028562545776,
      "learning_rate": 2.944070971799793e-06,
      "loss": 0.3817,
      "step": 24070
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.52048921585083,
      "learning_rate": 2.943349344978278e-06,
      "loss": 0.4657,
      "step": 24071
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6385314464569092,
      "learning_rate": 2.9426277913455138e-06,
      "loss": 0.3747,
      "step": 24072
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5673528909683228,
      "learning_rate": 2.9419063109089874e-06,
      "loss": 0.384,
      "step": 24073
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4871925115585327,
      "learning_rate": 2.9411849036761795e-06,
      "loss": 0.362,
      "step": 24074
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.737152338027954,
      "learning_rate": 2.940463569654568e-06,
      "loss": 0.9937,
      "step": 24075
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.447945475578308,
      "learning_rate": 2.9397423088516384e-06,
      "loss": 0.985,
      "step": 24076
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9064204692840576,
      "learning_rate": 2.9390211212748753e-06,
      "loss": 0.4192,
      "step": 24077
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.572020173072815,
      "learning_rate": 2.9383000069317503e-06,
      "loss": 0.378,
      "step": 24078
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5607309341430664,
      "learning_rate": 2.9375789658297495e-06,
      "loss": 0.4214,
      "step": 24079
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.505996584892273,
      "learning_rate": 2.9368579979763446e-06,
      "loss": 0.3737,
      "step": 24080
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5637083053588867,
      "learning_rate": 2.93613710337902e-06,
      "loss": 0.4378,
      "step": 24081
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5191144943237305,
      "learning_rate": 2.935416282045247e-06,
      "loss": 0.3727,
      "step": 24082
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5902948379516602,
      "learning_rate": 2.9346955339825034e-06,
      "loss": 0.4354,
      "step": 24083
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5462759733200073,
      "learning_rate": 2.9339748591982686e-06,
      "loss": 0.3658,
      "step": 24084
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5334019660949707,
      "learning_rate": 2.9332542577000134e-06,
      "loss": 0.4459,
      "step": 24085
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5552091598510742,
      "learning_rate": 2.9325337294952082e-06,
      "loss": 0.3895,
      "step": 24086
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.573826789855957,
      "learning_rate": 2.9318132745913308e-06,
      "loss": 0.4213,
      "step": 24087
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5069187879562378,
      "learning_rate": 2.9310928929958514e-06,
      "loss": 0.4069,
      "step": 24088
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.502267837524414,
      "learning_rate": 2.930372584716249e-06,
      "loss": 0.4263,
      "step": 24089
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5266644954681396,
      "learning_rate": 2.9296523497599816e-06,
      "loss": 0.3874,
      "step": 24090
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4989674091339111,
      "learning_rate": 2.9289321881345257e-06,
      "loss": 0.4207,
      "step": 24091
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4938507080078125,
      "learning_rate": 2.9282120998473495e-06,
      "loss": 0.3888,
      "step": 24092
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.579947590827942,
      "learning_rate": 2.927492084905923e-06,
      "loss": 0.4473,
      "step": 24093
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6242311000823975,
      "learning_rate": 2.9267721433177186e-06,
      "loss": 0.413,
      "step": 24094
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.494030237197876,
      "learning_rate": 2.926052275090192e-06,
      "loss": 0.4045,
      "step": 24095
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5166245698928833,
      "learning_rate": 2.9253324802308146e-06,
      "loss": 0.3817,
      "step": 24096
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5902328491210938,
      "learning_rate": 2.924612758747053e-06,
      "loss": 0.4022,
      "step": 24097
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.618097186088562,
      "learning_rate": 2.923893110646374e-06,
      "loss": 0.4096,
      "step": 24098
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5283623933792114,
      "learning_rate": 2.923173535936239e-06,
      "loss": 0.44,
      "step": 24099
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5432095527648926,
      "learning_rate": 2.922454034624108e-06,
      "loss": 0.3631,
      "step": 24100
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5972381830215454,
      "learning_rate": 2.9217346067174457e-06,
      "loss": 0.4481,
      "step": 24101
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4470505714416504,
      "learning_rate": 2.9210152522237166e-06,
      "loss": 0.367,
      "step": 24102
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5484198331832886,
      "learning_rate": 2.9202959711503777e-06,
      "loss": 0.415,
      "step": 24103
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5402579307556152,
      "learning_rate": 2.919576763504893e-06,
      "loss": 0.3726,
      "step": 24104
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5680527687072754,
      "learning_rate": 2.918857629294716e-06,
      "loss": 0.4168,
      "step": 24105
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5492266416549683,
      "learning_rate": 2.918138568527312e-06,
      "loss": 0.3737,
      "step": 24106
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5880132913589478,
      "learning_rate": 2.9174195812101324e-06,
      "loss": 0.4309,
      "step": 24107
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5218772888183594,
      "learning_rate": 2.916700667350637e-06,
      "loss": 0.3723,
      "step": 24108
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.57374906539917,
      "learning_rate": 2.9159818269562867e-06,
      "loss": 0.4335,
      "step": 24109
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4721084833145142,
      "learning_rate": 2.9152630600345324e-06,
      "loss": 0.3686,
      "step": 24110
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.527402400970459,
      "learning_rate": 2.9145443665928252e-06,
      "loss": 0.4255,
      "step": 24111
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.511216163635254,
      "learning_rate": 2.9138257466386234e-06,
      "loss": 0.3783,
      "step": 24112
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.58755624294281,
      "learning_rate": 2.9131072001793804e-06,
      "loss": 0.4579,
      "step": 24113
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5857465267181396,
      "learning_rate": 2.9123887272225517e-06,
      "loss": 0.4033,
      "step": 24114
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5088615417480469,
      "learning_rate": 2.9116703277755853e-06,
      "loss": 0.4014,
      "step": 24115
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5763517618179321,
      "learning_rate": 2.910952001845928e-06,
      "loss": 0.3776,
      "step": 24116
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1158511638641357,
      "learning_rate": 2.910233749441036e-06,
      "loss": 0.5093,
      "step": 24117
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5584731101989746,
      "learning_rate": 2.9095155705683597e-06,
      "loss": 0.4019,
      "step": 24118
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5136805772781372,
      "learning_rate": 2.908797465235342e-06,
      "loss": 0.4427,
      "step": 24119
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.530860424041748,
      "learning_rate": 2.9080794334494378e-06,
      "loss": 0.3618,
      "step": 24120
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.794772744178772,
      "learning_rate": 2.907361475218087e-06,
      "loss": 0.4277,
      "step": 24121
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.586168646812439,
      "learning_rate": 2.906643590548742e-06,
      "loss": 0.3687,
      "step": 24122
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5273054838180542,
      "learning_rate": 2.9059257794488428e-06,
      "loss": 0.4474,
      "step": 24123
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4805468320846558,
      "learning_rate": 2.905208041925838e-06,
      "loss": 0.3705,
      "step": 24124
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9001832008361816,
      "learning_rate": 2.9044903779871725e-06,
      "loss": 1.024,
      "step": 24125
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4999126195907593,
      "learning_rate": 2.90377278764029e-06,
      "loss": 1.0409,
      "step": 24126
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5927419662475586,
      "learning_rate": 2.903055270892626e-06,
      "loss": 0.3982,
      "step": 24127
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5439871549606323,
      "learning_rate": 2.902337827751629e-06,
      "loss": 0.4027,
      "step": 24128
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5182952880859375,
      "learning_rate": 2.9016204582247388e-06,
      "loss": 0.4182,
      "step": 24129
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5662527084350586,
      "learning_rate": 2.9009031623194007e-06,
      "loss": 0.4198,
      "step": 24130
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.454064965248108,
      "learning_rate": 2.900185940043042e-06,
      "loss": 0.4217,
      "step": 24131
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.626707911491394,
      "learning_rate": 2.8994687914031092e-06,
      "loss": 0.3899,
      "step": 24132
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5092294216156006,
      "learning_rate": 2.898751716407039e-06,
      "loss": 0.421,
      "step": 24133
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5367807149887085,
      "learning_rate": 2.8980347150622725e-06,
      "loss": 0.3756,
      "step": 24134
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6146998405456543,
      "learning_rate": 2.8973177873762415e-06,
      "loss": 0.4108,
      "step": 24135
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5001111030578613,
      "learning_rate": 2.8966009333563805e-06,
      "loss": 0.3689,
      "step": 24136
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.469610333442688,
      "learning_rate": 2.895884153010127e-06,
      "loss": 0.4057,
      "step": 24137
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5148828029632568,
      "learning_rate": 2.8951674463449174e-06,
      "loss": 0.4037,
      "step": 24138
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5478285551071167,
      "learning_rate": 2.8944508133681804e-06,
      "loss": 0.5194,
      "step": 24139
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6483482122421265,
      "learning_rate": 2.8937342540873535e-06,
      "loss": 0.3883,
      "step": 24140
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6439019441604614,
      "learning_rate": 2.8930177685098628e-06,
      "loss": 0.4515,
      "step": 24141
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5143119096755981,
      "learning_rate": 2.8923013566431446e-06,
      "loss": 0.392,
      "step": 24142
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6573950052261353,
      "learning_rate": 2.891585018494625e-06,
      "loss": 0.4478,
      "step": 24143
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.458841323852539,
      "learning_rate": 2.890868754071736e-06,
      "loss": 0.3654,
      "step": 24144
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6420841217041016,
      "learning_rate": 2.8901525633819084e-06,
      "loss": 0.4178,
      "step": 24145
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.483956217765808,
      "learning_rate": 2.8894364464325684e-06,
      "loss": 0.3721,
      "step": 24146
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5176384449005127,
      "learning_rate": 2.88872040323114e-06,
      "loss": 0.384,
      "step": 24147
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.551600456237793,
      "learning_rate": 2.888004433785052e-06,
      "loss": 0.4077,
      "step": 24148
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6022549867630005,
      "learning_rate": 2.8872885381017304e-06,
      "loss": 0.4443,
      "step": 24149
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.516736388206482,
      "learning_rate": 2.8865727161886035e-06,
      "loss": 0.3679,
      "step": 24150
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9218653440475464,
      "learning_rate": 2.8858569680530937e-06,
      "loss": 0.5045,
      "step": 24151
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5037537813186646,
      "learning_rate": 2.8851412937026192e-06,
      "loss": 0.3911,
      "step": 24152
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5900099277496338,
      "learning_rate": 2.884425693144607e-06,
      "loss": 0.4247,
      "step": 24153
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4958277940750122,
      "learning_rate": 2.883710166386482e-06,
      "loss": 0.376,
      "step": 24154
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5367579460144043,
      "learning_rate": 2.8829947134356582e-06,
      "loss": 0.4532,
      "step": 24155
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.582188606262207,
      "learning_rate": 2.8822793342995635e-06,
      "loss": 0.3961,
      "step": 24156
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4854862689971924,
      "learning_rate": 2.8815640289856105e-06,
      "loss": 0.409,
      "step": 24157
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5660959482192993,
      "learning_rate": 2.8808487975012257e-06,
      "loss": 0.3942,
      "step": 24158
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5867236852645874,
      "learning_rate": 2.880133639853818e-06,
      "loss": 0.4579,
      "step": 24159
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5249691009521484,
      "learning_rate": 2.8794185560508113e-06,
      "loss": 0.3817,
      "step": 24160
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.692476511001587,
      "learning_rate": 2.8787035460996225e-06,
      "loss": 0.4394,
      "step": 24161
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.521958351135254,
      "learning_rate": 2.877988610007667e-06,
      "loss": 0.3917,
      "step": 24162
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5979183912277222,
      "learning_rate": 2.877273747782354e-06,
      "loss": 0.4318,
      "step": 24163
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6369034051895142,
      "learning_rate": 2.876558959431103e-06,
      "loss": 0.3945,
      "step": 24164
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5044156312942505,
      "learning_rate": 2.8758442449613257e-06,
      "loss": 0.436,
      "step": 24165
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.444059133529663,
      "learning_rate": 2.875129604380439e-06,
      "loss": 0.3844,
      "step": 24166
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6116276979446411,
      "learning_rate": 2.8744150376958513e-06,
      "loss": 0.4402,
      "step": 24167
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.788199543952942,
      "learning_rate": 2.87370054491497e-06,
      "loss": 0.4002,
      "step": 24168
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5753060579299927,
      "learning_rate": 2.872986126045212e-06,
      "loss": 0.4077,
      "step": 24169
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.51509690284729,
      "learning_rate": 2.872271781093986e-06,
      "loss": 0.3774,
      "step": 24170
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.539990782737732,
      "learning_rate": 2.8715575100687e-06,
      "loss": 0.4016,
      "step": 24171
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4987998008728027,
      "learning_rate": 2.870843312976758e-06,
      "loss": 0.3727,
      "step": 24172
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5970927476882935,
      "learning_rate": 2.870129189825571e-06,
      "loss": 0.4554,
      "step": 24173
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5029685497283936,
      "learning_rate": 2.869415140622549e-06,
      "loss": 0.3868,
      "step": 24174
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5396445989608765,
      "learning_rate": 2.868701165375091e-06,
      "loss": 0.4331,
      "step": 24175
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6713348627090454,
      "learning_rate": 2.867987264090609e-06,
      "loss": 0.4254,
      "step": 24176
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.667370319366455,
      "learning_rate": 2.8672734367765e-06,
      "loss": 0.4648,
      "step": 24177
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6725554466247559,
      "learning_rate": 2.8665596834401744e-06,
      "loss": 0.4083,
      "step": 24178
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5522685050964355,
      "learning_rate": 2.865846004089029e-06,
      "loss": 0.4557,
      "step": 24179
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4780280590057373,
      "learning_rate": 2.8651323987304692e-06,
      "loss": 0.3724,
      "step": 24180
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5699518918991089,
      "learning_rate": 2.8644188673718987e-06,
      "loss": 0.4674,
      "step": 24181
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4774378538131714,
      "learning_rate": 2.863705410020714e-06,
      "loss": 0.3869,
      "step": 24182
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.588860034942627,
      "learning_rate": 2.8629920266843124e-06,
      "loss": 0.4372,
      "step": 24183
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5177148580551147,
      "learning_rate": 2.8622787173700974e-06,
      "loss": 0.3965,
      "step": 24184
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5592739582061768,
      "learning_rate": 2.8615654820854654e-06,
      "loss": 0.4375,
      "step": 24185
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.0024359226226807,
      "learning_rate": 2.860852320837817e-06,
      "loss": 0.4099,
      "step": 24186
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5306587219238281,
      "learning_rate": 2.860139233634547e-06,
      "loss": 0.423,
      "step": 24187
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5499075651168823,
      "learning_rate": 2.859426220483046e-06,
      "loss": 0.3873,
      "step": 24188
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5748159885406494,
      "learning_rate": 2.858713281390714e-06,
      "loss": 0.4627,
      "step": 24189
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5754731893539429,
      "learning_rate": 2.8580004163649478e-06,
      "loss": 0.3837,
      "step": 24190
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6334928274154663,
      "learning_rate": 2.8572876254131353e-06,
      "loss": 1.0036,
      "step": 24191
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4455000162124634,
      "learning_rate": 2.856574908542674e-06,
      "loss": 1.0173,
      "step": 24192
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.588759422302246,
      "learning_rate": 2.8558622657609527e-06,
      "loss": 0.465,
      "step": 24193
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5181070566177368,
      "learning_rate": 2.8551496970753645e-06,
      "loss": 0.3911,
      "step": 24194
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5256415605545044,
      "learning_rate": 2.854437202493298e-06,
      "loss": 0.3978,
      "step": 24195
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5049744844436646,
      "learning_rate": 2.8537247820221425e-06,
      "loss": 0.3809,
      "step": 24196
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5102756023406982,
      "learning_rate": 2.8530124356692923e-06,
      "loss": 0.4022,
      "step": 24197
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5224441289901733,
      "learning_rate": 2.852300163442132e-06,
      "loss": 0.3598,
      "step": 24198
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.539817452430725,
      "learning_rate": 2.851587965348045e-06,
      "loss": 0.3913,
      "step": 24199
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5855515003204346,
      "learning_rate": 2.8508758413944213e-06,
      "loss": 0.3878,
      "step": 24200
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.646470069885254,
      "learning_rate": 2.8501637915886484e-06,
      "loss": 0.4873,
      "step": 24201
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5133081674575806,
      "learning_rate": 2.8494518159381125e-06,
      "loss": 0.3847,
      "step": 24202
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4451615810394287,
      "learning_rate": 2.848739914450196e-06,
      "loss": 0.4164,
      "step": 24203
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5378937721252441,
      "learning_rate": 2.8480280871322786e-06,
      "loss": 0.3895,
      "step": 24204
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4935065507888794,
      "learning_rate": 2.8473163339917465e-06,
      "loss": 0.4157,
      "step": 24205
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5336904525756836,
      "learning_rate": 2.846604655035985e-06,
      "loss": 0.3795,
      "step": 24206
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.49406099319458,
      "learning_rate": 2.8458930502723725e-06,
      "loss": 0.4484,
      "step": 24207
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5094565153121948,
      "learning_rate": 2.8451815197082854e-06,
      "loss": 0.3618,
      "step": 24208
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5971614122390747,
      "learning_rate": 2.844470063351107e-06,
      "loss": 0.4883,
      "step": 24209
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.548580288887024,
      "learning_rate": 2.8437586812082195e-06,
      "loss": 0.3944,
      "step": 24210
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.552141547203064,
      "learning_rate": 2.8430473732869954e-06,
      "loss": 0.419,
      "step": 24211
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4665651321411133,
      "learning_rate": 2.842336139594817e-06,
      "loss": 0.3374,
      "step": 24212
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5918829441070557,
      "learning_rate": 2.841624980139056e-06,
      "loss": 0.4319,
      "step": 24213
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5556516647338867,
      "learning_rate": 2.8409138949270932e-06,
      "loss": 0.3978,
      "step": 24214
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4362181425094604,
      "learning_rate": 2.840202883966299e-06,
      "loss": 0.4014,
      "step": 24215
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5195497274398804,
      "learning_rate": 2.839491947264049e-06,
      "loss": 0.3917,
      "step": 24216
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5094714164733887,
      "learning_rate": 2.838781084827721e-06,
      "loss": 0.4427,
      "step": 24217
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4991375207901,
      "learning_rate": 2.8380702966646843e-06,
      "loss": 0.3876,
      "step": 24218
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6745809316635132,
      "learning_rate": 2.8373595827823085e-06,
      "loss": 0.4599,
      "step": 24219
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5082327127456665,
      "learning_rate": 2.8366489431879683e-06,
      "loss": 0.3653,
      "step": 24220
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6955698728561401,
      "learning_rate": 2.8359383778890327e-06,
      "loss": 0.4399,
      "step": 24221
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6393901109695435,
      "learning_rate": 2.835227886892875e-06,
      "loss": 0.4287,
      "step": 24222
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6133458614349365,
      "learning_rate": 2.8345174702068613e-06,
      "loss": 0.438,
      "step": 24223
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5874507427215576,
      "learning_rate": 2.833807127838356e-06,
      "loss": 0.3904,
      "step": 24224
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6014600992202759,
      "learning_rate": 2.8330968597947305e-06,
      "loss": 0.9164,
      "step": 24225
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4126551151275635,
      "learning_rate": 2.8323866660833543e-06,
      "loss": 1.0056,
      "step": 24226
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5636484622955322,
      "learning_rate": 2.8316765467115858e-06,
      "loss": 0.4132,
      "step": 24227
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5480036735534668,
      "learning_rate": 2.830966501686798e-06,
      "loss": 0.3699,
      "step": 24228
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6831341981887817,
      "learning_rate": 2.830256531016349e-06,
      "loss": 0.461,
      "step": 24229
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5132139921188354,
      "learning_rate": 2.8295466347076072e-06,
      "loss": 0.3604,
      "step": 24230
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5972791910171509,
      "learning_rate": 2.82883681276793e-06,
      "loss": 0.4383,
      "step": 24231
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.54099440574646,
      "learning_rate": 2.828127065204682e-06,
      "loss": 0.4168,
      "step": 24232
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.534145712852478,
      "learning_rate": 2.827417392025229e-06,
      "loss": 0.3976,
      "step": 24233
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5001174211502075,
      "learning_rate": 2.826707793236926e-06,
      "loss": 0.349,
      "step": 24234
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5812652111053467,
      "learning_rate": 2.825998268847131e-06,
      "loss": 0.4352,
      "step": 24235
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5937042236328125,
      "learning_rate": 2.825288818863207e-06,
      "loss": 0.4204,
      "step": 24236
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.514260172843933,
      "learning_rate": 2.824579443292509e-06,
      "loss": 0.4104,
      "step": 24237
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.493004322052002,
      "learning_rate": 2.8238701421424007e-06,
      "loss": 0.3785,
      "step": 24238
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.72919225692749,
      "learning_rate": 2.8231609154202335e-06,
      "loss": 0.4195,
      "step": 24239
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6314032077789307,
      "learning_rate": 2.8224517631333604e-06,
      "loss": 0.3755,
      "step": 24240
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6127516031265259,
      "learning_rate": 2.8217426852891416e-06,
      "loss": 0.464,
      "step": 24241
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.6398167610168457,
      "learning_rate": 2.8210336818949312e-06,
      "loss": 0.3791,
      "step": 24242
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5533230304718018,
      "learning_rate": 2.820324752958079e-06,
      "loss": 0.4073,
      "step": 24243
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.576886773109436,
      "learning_rate": 2.8196158984859436e-06,
      "loss": 0.37,
      "step": 24244
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6382360458374023,
      "learning_rate": 2.8189071184858695e-06,
      "loss": 0.3848,
      "step": 24245
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5409002304077148,
      "learning_rate": 2.818198412965215e-06,
      "loss": 0.3914,
      "step": 24246
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5424175262451172,
      "learning_rate": 2.817489781931324e-06,
      "loss": 0.4196,
      "step": 24247
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.519120693206787,
      "learning_rate": 2.8167812253915527e-06,
      "loss": 0.3861,
      "step": 24248
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5905097723007202,
      "learning_rate": 2.816072743353243e-06,
      "loss": 0.4586,
      "step": 24249
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7392518520355225,
      "learning_rate": 2.8153643358237502e-06,
      "loss": 0.3854,
      "step": 24250
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6256698369979858,
      "learning_rate": 2.8146560028104155e-06,
      "loss": 0.412,
      "step": 24251
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5681718587875366,
      "learning_rate": 2.8139477443205866e-06,
      "loss": 0.3748,
      "step": 24252
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6329584121704102,
      "learning_rate": 2.813239560361615e-06,
      "loss": 0.4134,
      "step": 24253
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6048823595046997,
      "learning_rate": 2.8125314509408407e-06,
      "loss": 0.3959,
      "step": 24254
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5823229551315308,
      "learning_rate": 2.8118234160656054e-06,
      "loss": 0.4332,
      "step": 24255
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.426942229270935,
      "learning_rate": 2.811115455743256e-06,
      "loss": 0.3418,
      "step": 24256
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5010919570922852,
      "learning_rate": 2.8104075699811353e-06,
      "loss": 0.4001,
      "step": 24257
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7013885974884033,
      "learning_rate": 2.809699758786588e-06,
      "loss": 0.3663,
      "step": 24258
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.564104676246643,
      "learning_rate": 2.808992022166952e-06,
      "loss": 0.4545,
      "step": 24259
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4947861433029175,
      "learning_rate": 2.808284360129564e-06,
      "loss": 0.3601,
      "step": 24260
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.529821515083313,
      "learning_rate": 2.8075767726817684e-06,
      "loss": 0.4331,
      "step": 24261
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5464057922363281,
      "learning_rate": 2.8068692598309043e-06,
      "loss": 0.3821,
      "step": 24262
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5271111726760864,
      "learning_rate": 2.806161821584307e-06,
      "loss": 0.4172,
      "step": 24263
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4247198104858398,
      "learning_rate": 2.805454457949317e-06,
      "loss": 0.3706,
      "step": 24264
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5609776973724365,
      "learning_rate": 2.804747168933266e-06,
      "loss": 0.4176,
      "step": 24265
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.510847568511963,
      "learning_rate": 2.804039954543497e-06,
      "loss": 0.3477,
      "step": 24266
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5855779647827148,
      "learning_rate": 2.8033328147873363e-06,
      "loss": 0.4379,
      "step": 24267
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4772433042526245,
      "learning_rate": 2.8026257496721234e-06,
      "loss": 0.371,
      "step": 24268
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.8675713539123535,
      "learning_rate": 2.8019187592051922e-06,
      "loss": 0.418,
      "step": 24269
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5818811655044556,
      "learning_rate": 2.8012118433938752e-06,
      "loss": 0.403,
      "step": 24270
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5164573192596436,
      "learning_rate": 2.8005050022454984e-06,
      "loss": 0.4232,
      "step": 24271
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4729869365692139,
      "learning_rate": 2.7997982357673982e-06,
      "loss": 0.3768,
      "step": 24272
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5480767488479614,
      "learning_rate": 2.799091543966904e-06,
      "loss": 0.4235,
      "step": 24273
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4795913696289062,
      "learning_rate": 2.7983849268513485e-06,
      "loss": 0.3752,
      "step": 24274
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5055707693099976,
      "learning_rate": 2.7976783844280564e-06,
      "loss": 0.4227,
      "step": 24275
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5108779668807983,
      "learning_rate": 2.796971916704355e-06,
      "loss": 0.3658,
      "step": 24276
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5318270921707153,
      "learning_rate": 2.796265523687571e-06,
      "loss": 0.4423,
      "step": 24277
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5976879596710205,
      "learning_rate": 2.7955592053850376e-06,
      "loss": 0.3865,
      "step": 24278
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.583425521850586,
      "learning_rate": 2.794852961804072e-06,
      "loss": 0.41,
      "step": 24279
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7058371305465698,
      "learning_rate": 2.794146792952006e-06,
      "loss": 0.3883,
      "step": 24280
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6372830867767334,
      "learning_rate": 2.7934406988361586e-06,
      "loss": 0.4084,
      "step": 24281
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5173077583312988,
      "learning_rate": 2.7927346794638565e-06,
      "loss": 0.3815,
      "step": 24282
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5259592533111572,
      "learning_rate": 2.792028734842418e-06,
      "loss": 0.4322,
      "step": 24283
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5250385999679565,
      "learning_rate": 2.7913228649791678e-06,
      "loss": 0.3795,
      "step": 24284
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5881931781768799,
      "learning_rate": 2.7906170698814293e-06,
      "loss": 0.4112,
      "step": 24285
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5668485164642334,
      "learning_rate": 2.789911349556521e-06,
      "loss": 0.3933,
      "step": 24286
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5620710849761963,
      "learning_rate": 2.7892057040117584e-06,
      "loss": 0.4153,
      "step": 24287
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4984819889068604,
      "learning_rate": 2.7885001332544635e-06,
      "loss": 0.398,
      "step": 24288
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6664057970046997,
      "learning_rate": 2.7877946372919563e-06,
      "loss": 0.4351,
      "step": 24289
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5449880361557007,
      "learning_rate": 2.787089216131552e-06,
      "loss": 0.388,
      "step": 24290
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4694774150848389,
      "learning_rate": 2.7863838697805633e-06,
      "loss": 0.3941,
      "step": 24291
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5397473573684692,
      "learning_rate": 2.7856785982463095e-06,
      "loss": 0.3867,
      "step": 24292
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5701537132263184,
      "learning_rate": 2.784973401536104e-06,
      "loss": 0.4786,
      "step": 24293
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.561484932899475,
      "learning_rate": 2.7842682796572662e-06,
      "loss": 0.3926,
      "step": 24294
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8243770599365234,
      "learning_rate": 2.7835632326171037e-06,
      "loss": 0.4628,
      "step": 24295
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5082017183303833,
      "learning_rate": 2.7828582604229272e-06,
      "loss": 0.3627,
      "step": 24296
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4661638736724854,
      "learning_rate": 2.782153363082052e-06,
      "loss": 0.4146,
      "step": 24297
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8183094263076782,
      "learning_rate": 2.781448540601791e-06,
      "loss": 0.3974,
      "step": 24298
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.543304204940796,
      "learning_rate": 2.7807437929894488e-06,
      "loss": 0.4785,
      "step": 24299
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4845550060272217,
      "learning_rate": 2.78003912025234e-06,
      "loss": 0.3783,
      "step": 24300
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5780102014541626,
      "learning_rate": 2.7793345223977687e-06,
      "loss": 0.4159,
      "step": 24301
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5866156816482544,
      "learning_rate": 2.778629999433047e-06,
      "loss": 0.392,
      "step": 24302
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.624335765838623,
      "learning_rate": 2.7779255513654767e-06,
      "loss": 0.4195,
      "step": 24303
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5087165832519531,
      "learning_rate": 2.7772211782023683e-06,
      "loss": 0.3722,
      "step": 24304
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6343902349472046,
      "learning_rate": 2.7765168799510278e-06,
      "loss": 0.4541,
      "step": 24305
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5087929964065552,
      "learning_rate": 2.775812656618759e-06,
      "loss": 0.3619,
      "step": 24306
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6260019540786743,
      "learning_rate": 2.7751085082128614e-06,
      "loss": 0.9602,
      "step": 24307
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.397059440612793,
      "learning_rate": 2.774404434740643e-06,
      "loss": 0.9985,
      "step": 24308
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6339415311813354,
      "learning_rate": 2.773700436209403e-06,
      "loss": 0.4194,
      "step": 24309
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.577695608139038,
      "learning_rate": 2.7729965126264503e-06,
      "loss": 0.3899,
      "step": 24310
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5848733186721802,
      "learning_rate": 2.7722926639990786e-06,
      "loss": 0.4323,
      "step": 24311
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.494256615638733,
      "learning_rate": 2.7715888903345865e-06,
      "loss": 0.3517,
      "step": 24312
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5623502731323242,
      "learning_rate": 2.770885191640278e-06,
      "loss": 0.4556,
      "step": 24313
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5281561613082886,
      "learning_rate": 2.7701815679234513e-06,
      "loss": 0.3856,
      "step": 24314
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5569851398468018,
      "learning_rate": 2.7694780191914005e-06,
      "loss": 0.4126,
      "step": 24315
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5295411348342896,
      "learning_rate": 2.7687745454514283e-06,
      "loss": 0.3847,
      "step": 24316
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6112964153289795,
      "learning_rate": 2.7680711467108232e-06,
      "loss": 0.4161,
      "step": 24317
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4723068475723267,
      "learning_rate": 2.76736782297689e-06,
      "loss": 0.3639,
      "step": 24318
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5294287204742432,
      "learning_rate": 2.7666645742569133e-06,
      "loss": 0.4045,
      "step": 24319
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.522756814956665,
      "learning_rate": 2.7659614005581913e-06,
      "loss": 0.3791,
      "step": 24320
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5285894870758057,
      "learning_rate": 2.7652583018880207e-06,
      "loss": 0.4263,
      "step": 24321
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4850609302520752,
      "learning_rate": 2.764555278253691e-06,
      "loss": 0.3496,
      "step": 24322
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4635850191116333,
      "learning_rate": 2.7638523296624896e-06,
      "loss": 0.4155,
      "step": 24323
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.605065107345581,
      "learning_rate": 2.763149456121711e-06,
      "loss": 0.3911,
      "step": 24324
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5173711776733398,
      "learning_rate": 2.7624466576386478e-06,
      "loss": 0.4128,
      "step": 24325
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.501794695854187,
      "learning_rate": 2.7617439342205865e-06,
      "loss": 0.3746,
      "step": 24326
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5555858612060547,
      "learning_rate": 2.7610412858748115e-06,
      "loss": 0.491,
      "step": 24327
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.443286418914795,
      "learning_rate": 2.7603387126086147e-06,
      "loss": 0.3663,
      "step": 24328
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5145686864852905,
      "learning_rate": 2.759636214429282e-06,
      "loss": 0.405,
      "step": 24329
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4652493000030518,
      "learning_rate": 2.7589337913441027e-06,
      "loss": 0.3838,
      "step": 24330
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6845223903656006,
      "learning_rate": 2.7582314433603587e-06,
      "loss": 0.4504,
      "step": 24331
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6127585172653198,
      "learning_rate": 2.7575291704853325e-06,
      "loss": 0.4318,
      "step": 24332
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.528860330581665,
      "learning_rate": 2.7568269727263096e-06,
      "loss": 0.3951,
      "step": 24333
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5344575643539429,
      "learning_rate": 2.7561248500905736e-06,
      "loss": 0.3761,
      "step": 24334
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7396494150161743,
      "learning_rate": 2.755422802585409e-06,
      "loss": 0.9722,
      "step": 24335
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4138365983963013,
      "learning_rate": 2.754720830218095e-06,
      "loss": 0.9837,
      "step": 24336
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6161853075027466,
      "learning_rate": 2.754018932995909e-06,
      "loss": 0.4284,
      "step": 24337
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6114351749420166,
      "learning_rate": 2.7533171109261335e-06,
      "loss": 0.3803,
      "step": 24338
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5203359127044678,
      "learning_rate": 2.7526153640160504e-06,
      "loss": 0.4065,
      "step": 24339
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.489663004875183,
      "learning_rate": 2.7519136922729318e-06,
      "loss": 0.3564,
      "step": 24340
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6599358320236206,
      "learning_rate": 2.751212095704062e-06,
      "loss": 0.4412,
      "step": 24341
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5705233812332153,
      "learning_rate": 2.7505105743167105e-06,
      "loss": 0.3804,
      "step": 24342
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6440482139587402,
      "learning_rate": 2.7498091281181607e-06,
      "loss": 0.4231,
      "step": 24343
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5147761106491089,
      "learning_rate": 2.749107757115681e-06,
      "loss": 0.3464,
      "step": 24344
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5630712509155273,
      "learning_rate": 2.7484064613165475e-06,
      "loss": 0.4503,
      "step": 24345
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5569766759872437,
      "learning_rate": 2.747705240728039e-06,
      "loss": 0.3786,
      "step": 24346
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.500908613204956,
      "learning_rate": 2.747004095357424e-06,
      "loss": 0.434,
      "step": 24347
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5632978677749634,
      "learning_rate": 2.7463030252119706e-06,
      "loss": 0.3921,
      "step": 24348
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5329324007034302,
      "learning_rate": 2.745602030298955e-06,
      "loss": 0.4712,
      "step": 24349
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7954058647155762,
      "learning_rate": 2.7449011106256462e-06,
      "loss": 0.4459,
      "step": 24350
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5781983137130737,
      "learning_rate": 2.7442002661993185e-06,
      "loss": 0.4336,
      "step": 24351
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5106576681137085,
      "learning_rate": 2.7434994970272353e-06,
      "loss": 0.3587,
      "step": 24352
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5328419208526611,
      "learning_rate": 2.742798803116664e-06,
      "loss": 0.402,
      "step": 24353
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5035338401794434,
      "learning_rate": 2.742098184474874e-06,
      "loss": 0.4161,
      "step": 24354
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5817346572875977,
      "learning_rate": 2.7413976411091338e-06,
      "loss": 0.4058,
      "step": 24355
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.621580958366394,
      "learning_rate": 2.740697173026705e-06,
      "loss": 0.3972,
      "step": 24356
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.498713731765747,
      "learning_rate": 2.739996780234857e-06,
      "loss": 0.4227,
      "step": 24357
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4840757846832275,
      "learning_rate": 2.739296462740849e-06,
      "loss": 0.3454,
      "step": 24358
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.532350778579712,
      "learning_rate": 2.738596220551951e-06,
      "loss": 0.3783,
      "step": 24359
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5068120956420898,
      "learning_rate": 2.7378960536754186e-06,
      "loss": 0.3954,
      "step": 24360
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4878329038619995,
      "learning_rate": 2.737195962118516e-06,
      "loss": 0.3947,
      "step": 24361
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5636390447616577,
      "learning_rate": 2.7364959458885087e-06,
      "loss": 0.4041,
      "step": 24362
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.5161030292510986,
      "learning_rate": 2.7357960049926523e-06,
      "loss": 0.4673,
      "step": 24363
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5363987684249878,
      "learning_rate": 2.7350961394382057e-06,
      "loss": 0.3953,
      "step": 24364
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5634528398513794,
      "learning_rate": 2.7343963492324286e-06,
      "loss": 0.4041,
      "step": 24365
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5100810527801514,
      "learning_rate": 2.7336966343825832e-06,
      "loss": 0.3941,
      "step": 24366
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7132179737091064,
      "learning_rate": 2.7329969948959222e-06,
      "loss": 0.987,
      "step": 24367
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.464392066001892,
      "learning_rate": 2.7322974307797e-06,
      "loss": 0.9831,
      "step": 24368
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4517635107040405,
      "learning_rate": 2.7315979420411766e-06,
      "loss": 0.4418,
      "step": 24369
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4317649602890015,
      "learning_rate": 2.7308985286876033e-06,
      "loss": 0.3445,
      "step": 24370
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5773520469665527,
      "learning_rate": 2.7301991907262405e-06,
      "loss": 0.4261,
      "step": 24371
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5411828756332397,
      "learning_rate": 2.7294999281643363e-06,
      "loss": 0.3754,
      "step": 24372
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5747556686401367,
      "learning_rate": 2.7288007410091413e-06,
      "loss": 0.434,
      "step": 24373
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5217692852020264,
      "learning_rate": 2.72810162926791e-06,
      "loss": 0.3979,
      "step": 24374
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.885215401649475,
      "learning_rate": 2.7274025929478964e-06,
      "loss": 0.4131,
      "step": 24375
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5066304206848145,
      "learning_rate": 2.7267036320563435e-06,
      "loss": 0.3565,
      "step": 24376
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5351500511169434,
      "learning_rate": 2.726004746600508e-06,
      "loss": 0.3871,
      "step": 24377
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5365833044052124,
      "learning_rate": 2.7253059365876324e-06,
      "loss": 0.3513,
      "step": 24378
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5389407873153687,
      "learning_rate": 2.724607202024969e-06,
      "loss": 0.4108,
      "step": 24379
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4557080268859863,
      "learning_rate": 2.723908542919761e-06,
      "loss": 0.375,
      "step": 24380
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5494623184204102,
      "learning_rate": 2.723209959279256e-06,
      "loss": 0.4126,
      "step": 24381
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5278699398040771,
      "learning_rate": 2.7225114511107033e-06,
      "loss": 0.3742,
      "step": 24382
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4915543794631958,
      "learning_rate": 2.7218130184213433e-06,
      "loss": 0.3975,
      "step": 24383
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4872246980667114,
      "learning_rate": 2.7211146612184183e-06,
      "loss": 0.3707,
      "step": 24384
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.565613865852356,
      "learning_rate": 2.720416379509173e-06,
      "loss": 0.4775,
      "step": 24385
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5049993991851807,
      "learning_rate": 2.7197181733008514e-06,
      "loss": 0.3678,
      "step": 24386
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6232441663742065,
      "learning_rate": 2.7190200426006973e-06,
      "loss": 0.9361,
      "step": 24387
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.437182903289795,
      "learning_rate": 2.718321987415947e-06,
      "loss": 0.9573,
      "step": 24388
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5471047163009644,
      "learning_rate": 2.7176240077538394e-06,
      "loss": 0.453,
      "step": 24389
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5560678243637085,
      "learning_rate": 2.7169261036216164e-06,
      "loss": 0.3834,
      "step": 24390
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5622612237930298,
      "learning_rate": 2.716228275026519e-06,
      "loss": 0.4288,
      "step": 24391
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4985668659210205,
      "learning_rate": 2.715530521975778e-06,
      "loss": 0.3767,
      "step": 24392
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5754175186157227,
      "learning_rate": 2.714832844476637e-06,
      "loss": 0.4391,
      "step": 24393
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6417443752288818,
      "learning_rate": 2.7141352425363266e-06,
      "loss": 0.3763,
      "step": 24394
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5735760927200317,
      "learning_rate": 2.713437716162087e-06,
      "loss": 0.4082,
      "step": 24395
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5635534524917603,
      "learning_rate": 2.7127402653611466e-06,
      "loss": 0.4143,
      "step": 24396
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5880932807922363,
      "learning_rate": 2.712042890140744e-06,
      "loss": 0.4224,
      "step": 24397
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5421878099441528,
      "learning_rate": 2.711345590508112e-06,
      "loss": 0.3803,
      "step": 24398
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5897754430770874,
      "learning_rate": 2.7106483664704817e-06,
      "loss": 0.45,
      "step": 24399
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4574922323226929,
      "learning_rate": 2.7099512180350817e-06,
      "loss": 0.3617,
      "step": 24400
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5338943004608154,
      "learning_rate": 2.7092541452091448e-06,
      "loss": 0.4158,
      "step": 24401
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.491983413696289,
      "learning_rate": 2.708557147999904e-06,
      "loss": 0.3669,
      "step": 24402
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4954006671905518,
      "learning_rate": 2.707860226414586e-06,
      "loss": 0.3895,
      "step": 24403
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6582156419754028,
      "learning_rate": 2.7071633804604135e-06,
      "loss": 0.3941,
      "step": 24404
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4296817779541016,
      "learning_rate": 2.706466610144619e-06,
      "loss": 0.4064,
      "step": 24405
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5222642421722412,
      "learning_rate": 2.7057699154744297e-06,
      "loss": 0.4038,
      "step": 24406
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.623846411705017,
      "learning_rate": 2.7050732964570724e-06,
      "loss": 0.4488,
      "step": 24407
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.546434760093689,
      "learning_rate": 2.704376753099771e-06,
      "loss": 0.3998,
      "step": 24408
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.631381869316101,
      "learning_rate": 2.7036802854097454e-06,
      "loss": 0.4731,
      "step": 24409
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.497686505317688,
      "learning_rate": 2.702983893394222e-06,
      "loss": 0.3637,
      "step": 24410
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.786963701248169,
      "learning_rate": 2.7022875770604283e-06,
      "loss": 0.4541,
      "step": 24411
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4661943912506104,
      "learning_rate": 2.701591336415579e-06,
      "loss": 0.3593,
      "step": 24412
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6226364374160767,
      "learning_rate": 2.700895171466901e-06,
      "loss": 0.4068,
      "step": 24413
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5047177076339722,
      "learning_rate": 2.7001990822216082e-06,
      "loss": 0.3678,
      "step": 24414
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5759450197219849,
      "learning_rate": 2.6995030686869283e-06,
      "loss": 0.4671,
      "step": 24415
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5803810358047485,
      "learning_rate": 2.6988071308700715e-06,
      "loss": 0.4096,
      "step": 24416
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.515576720237732,
      "learning_rate": 2.6981112687782605e-06,
      "loss": 0.4197,
      "step": 24417
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.52411949634552,
      "learning_rate": 2.697415482418715e-06,
      "loss": 0.3802,
      "step": 24418
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5982264280319214,
      "learning_rate": 2.696719771798648e-06,
      "loss": 0.4458,
      "step": 24419
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5752700567245483,
      "learning_rate": 2.6960241369252725e-06,
      "loss": 0.3831,
      "step": 24420
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6374115943908691,
      "learning_rate": 2.695328577805807e-06,
      "loss": 0.4882,
      "step": 24421
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6137844324111938,
      "learning_rate": 2.694633094447464e-06,
      "loss": 0.3912,
      "step": 24422
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4645695686340332,
      "learning_rate": 2.693937686857462e-06,
      "loss": 0.347,
      "step": 24423
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5591539144515991,
      "learning_rate": 2.6932423550430076e-06,
      "loss": 0.4017,
      "step": 24424
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5344606637954712,
      "learning_rate": 2.6925470990113125e-06,
      "loss": 0.4236,
      "step": 24425
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6602694988250732,
      "learning_rate": 2.6918519187695877e-06,
      "loss": 0.384,
      "step": 24426
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6097255945205688,
      "learning_rate": 2.691156814325049e-06,
      "loss": 0.4729,
      "step": 24427
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.2712156772613525,
      "learning_rate": 2.6904617856848982e-06,
      "loss": 0.4017,
      "step": 24428
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6052846908569336,
      "learning_rate": 2.68976683285635e-06,
      "loss": 0.4279,
      "step": 24429
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5603065490722656,
      "learning_rate": 2.6890719558466063e-06,
      "loss": 0.3826,
      "step": 24430
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7643567323684692,
      "learning_rate": 2.688377154662881e-06,
      "loss": 0.4601,
      "step": 24431
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6249628067016602,
      "learning_rate": 2.687682429312373e-06,
      "loss": 0.404,
      "step": 24432
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.580331802368164,
      "learning_rate": 2.6869877798022915e-06,
      "loss": 0.4617,
      "step": 24433
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4540784358978271,
      "learning_rate": 2.6862932061398427e-06,
      "loss": 0.356,
      "step": 24434
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5408668518066406,
      "learning_rate": 2.6855987083322298e-06,
      "loss": 0.419,
      "step": 24435
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5514235496520996,
      "learning_rate": 2.6849042863866515e-06,
      "loss": 0.3931,
      "step": 24436
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.518203616142273,
      "learning_rate": 2.6842099403103128e-06,
      "loss": 0.4034,
      "step": 24437
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5755904912948608,
      "learning_rate": 2.6835156701104147e-06,
      "loss": 0.4333,
      "step": 24438
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5218287706375122,
      "learning_rate": 2.6828214757941663e-06,
      "loss": 0.4638,
      "step": 24439
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5468393564224243,
      "learning_rate": 2.6821273573687534e-06,
      "loss": 0.3798,
      "step": 24440
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5826966762542725,
      "learning_rate": 2.6814333148413818e-06,
      "loss": 0.4674,
      "step": 24441
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4760454893112183,
      "learning_rate": 2.680739348219249e-06,
      "loss": 0.3668,
      "step": 24442
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5775582790374756,
      "learning_rate": 2.680045457509557e-06,
      "loss": 0.4638,
      "step": 24443
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4888720512390137,
      "learning_rate": 2.679351642719499e-06,
      "loss": 0.3696,
      "step": 24444
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5302846431732178,
      "learning_rate": 2.6786579038562666e-06,
      "loss": 0.4236,
      "step": 24445
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6522300243377686,
      "learning_rate": 2.67796424092706e-06,
      "loss": 0.3874,
      "step": 24446
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5794459581375122,
      "learning_rate": 2.6772706539390757e-06,
      "loss": 0.4377,
      "step": 24447
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5827819108963013,
      "learning_rate": 2.6765771428995014e-06,
      "loss": 0.3824,
      "step": 24448
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.191621780395508,
      "learning_rate": 2.675883707815535e-06,
      "loss": 0.4285,
      "step": 24449
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5976238250732422,
      "learning_rate": 2.6751903486943654e-06,
      "loss": 0.3918,
      "step": 24450
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5791916847229004,
      "learning_rate": 2.6744970655431855e-06,
      "loss": 0.4104,
      "step": 24451
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4861021041870117,
      "learning_rate": 2.673803858369184e-06,
      "loss": 0.352,
      "step": 24452
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5305052995681763,
      "learning_rate": 2.673110727179551e-06,
      "loss": 0.4243,
      "step": 24453
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5652247667312622,
      "learning_rate": 2.6724176719814785e-06,
      "loss": 0.4045,
      "step": 24454
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.565658450126648,
      "learning_rate": 2.6717246927821527e-06,
      "loss": 0.4201,
      "step": 24455
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5164105892181396,
      "learning_rate": 2.6710317895887573e-06,
      "loss": 0.3881,
      "step": 24456
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5498090982437134,
      "learning_rate": 2.670338962408482e-06,
      "loss": 0.4394,
      "step": 24457
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5109037160873413,
      "learning_rate": 2.669646211248512e-06,
      "loss": 0.3705,
      "step": 24458
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.602961778640747,
      "learning_rate": 2.6689535361160357e-06,
      "loss": 0.4248,
      "step": 24459
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4738116264343262,
      "learning_rate": 2.6682609370182345e-06,
      "loss": 0.3714,
      "step": 24460
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5876779556274414,
      "learning_rate": 2.667568413962288e-06,
      "loss": 0.9454,
      "step": 24461
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.499871850013733,
      "learning_rate": 2.666875966955382e-06,
      "loss": 1.0615,
      "step": 24462
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4770478010177612,
      "learning_rate": 2.666183596004702e-06,
      "loss": 0.4251,
      "step": 24463
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5014901161193848,
      "learning_rate": 2.6654913011174223e-06,
      "loss": 0.3596,
      "step": 24464
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5679179430007935,
      "learning_rate": 2.66479908230073e-06,
      "loss": 0.4411,
      "step": 24465
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4949718713760376,
      "learning_rate": 2.664106939561797e-06,
      "loss": 0.3895,
      "step": 24466
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9078973531723022,
      "learning_rate": 2.663414872907809e-06,
      "loss": 0.4276,
      "step": 24467
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5162932872772217,
      "learning_rate": 2.6627228823459373e-06,
      "loss": 0.4027,
      "step": 24468
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5283358097076416,
      "learning_rate": 2.662030967883362e-06,
      "loss": 0.414,
      "step": 24469
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4932910203933716,
      "learning_rate": 2.661339129527263e-06,
      "loss": 0.3889,
      "step": 24470
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5535340309143066,
      "learning_rate": 2.6606473672848123e-06,
      "loss": 0.4122,
      "step": 24471
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4785385131835938,
      "learning_rate": 2.659955681163181e-06,
      "loss": 0.374,
      "step": 24472
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5476371049880981,
      "learning_rate": 2.659264071169547e-06,
      "loss": 0.4271,
      "step": 24473
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5109429359436035,
      "learning_rate": 2.658572537311084e-06,
      "loss": 0.3804,
      "step": 24474
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5196046829223633,
      "learning_rate": 2.6578810795949683e-06,
      "loss": 0.417,
      "step": 24475
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4948296546936035,
      "learning_rate": 2.6571896980283606e-06,
      "loss": 0.3694,
      "step": 24476
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.52326238155365,
      "learning_rate": 2.6564983926184374e-06,
      "loss": 0.4757,
      "step": 24477
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5863924026489258,
      "learning_rate": 2.655807163372369e-06,
      "loss": 0.3831,
      "step": 24478
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.620821475982666,
      "learning_rate": 2.6551160102973273e-06,
      "loss": 0.4538,
      "step": 24479
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4691232442855835,
      "learning_rate": 2.654424933400478e-06,
      "loss": 0.3854,
      "step": 24480
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6063891649246216,
      "learning_rate": 2.653733932688984e-06,
      "loss": 0.3961,
      "step": 24481
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5357905626296997,
      "learning_rate": 2.6530430081700177e-06,
      "loss": 0.3521,
      "step": 24482
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5723716020584106,
      "learning_rate": 2.6523521598507463e-06,
      "loss": 0.4042,
      "step": 24483
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5233300924301147,
      "learning_rate": 2.651661387738329e-06,
      "loss": 0.3498,
      "step": 24484
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5812859535217285,
      "learning_rate": 2.650970691839937e-06,
      "loss": 0.4373,
      "step": 24485
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.435766577720642,
      "learning_rate": 2.6502800721627277e-06,
      "loss": 0.3435,
      "step": 24486
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7026362419128418,
      "learning_rate": 2.6495895287138706e-06,
      "loss": 0.9905,
      "step": 24487
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.442645788192749,
      "learning_rate": 2.6488990615005194e-06,
      "loss": 1.0462,
      "step": 24488
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4688608646392822,
      "learning_rate": 2.6482086705298414e-06,
      "loss": 0.4036,
      "step": 24489
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.494920015335083,
      "learning_rate": 2.647518355808998e-06,
      "loss": 0.3779,
      "step": 24490
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6005064249038696,
      "learning_rate": 2.646828117345146e-06,
      "loss": 0.3997,
      "step": 24491
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5094248056411743,
      "learning_rate": 2.6461379551454424e-06,
      "loss": 0.364,
      "step": 24492
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.654697299003601,
      "learning_rate": 2.6454478692170484e-06,
      "loss": 0.466,
      "step": 24493
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4713209867477417,
      "learning_rate": 2.6447578595671184e-06,
      "loss": 0.3564,
      "step": 24494
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5297236442565918,
      "learning_rate": 2.644067926202816e-06,
      "loss": 0.4479,
      "step": 24495
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4887717962265015,
      "learning_rate": 2.6433780691312905e-06,
      "loss": 0.3783,
      "step": 24496
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7035359144210815,
      "learning_rate": 2.642688288359696e-06,
      "loss": 0.4839,
      "step": 24497
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.559462070465088,
      "learning_rate": 2.641998583895189e-06,
      "loss": 0.3882,
      "step": 24498
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6504572629928589,
      "learning_rate": 2.6413089557449257e-06,
      "loss": 0.4501,
      "step": 24499
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4824844598770142,
      "learning_rate": 2.6406194039160525e-06,
      "loss": 0.3768,
      "step": 24500
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5092647075653076,
      "learning_rate": 2.6399299284157276e-06,
      "loss": 0.405,
      "step": 24501
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5275815725326538,
      "learning_rate": 2.639240529251095e-06,
      "loss": 0.3742,
      "step": 24502
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5472080707550049,
      "learning_rate": 2.6385512064293117e-06,
      "loss": 0.4506,
      "step": 24503
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5474226474761963,
      "learning_rate": 2.6378619599575216e-06,
      "loss": 0.3694,
      "step": 24504
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.614067792892456,
      "learning_rate": 2.6371727898428755e-06,
      "loss": 0.4449,
      "step": 24505
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.586604356765747,
      "learning_rate": 2.6364836960925233e-06,
      "loss": 0.4063,
      "step": 24506
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5873911380767822,
      "learning_rate": 2.635794678713611e-06,
      "loss": 0.4288,
      "step": 24507
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4801561832427979,
      "learning_rate": 2.6351057377132804e-06,
      "loss": 0.3842,
      "step": 24508
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5030198097229004,
      "learning_rate": 2.634416873098681e-06,
      "loss": 0.4351,
      "step": 24509
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.48586106300354,
      "learning_rate": 2.6337280848769566e-06,
      "loss": 0.3846,
      "step": 24510
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6259263753890991,
      "learning_rate": 2.633039373055254e-06,
      "loss": 0.4246,
      "step": 24511
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.438761591911316,
      "learning_rate": 2.632350737640713e-06,
      "loss": 0.3556,
      "step": 24512
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.558334231376648,
      "learning_rate": 2.6316621786404738e-06,
      "loss": 0.4392,
      "step": 24513
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5443240404129028,
      "learning_rate": 2.6309736960616807e-06,
      "loss": 0.3487,
      "step": 24514
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5968897342681885,
      "learning_rate": 2.6302852899114773e-06,
      "loss": 0.412,
      "step": 24515
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5014166831970215,
      "learning_rate": 2.6295969601969993e-06,
      "loss": 0.3898,
      "step": 24516
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6320345401763916,
      "learning_rate": 2.628908706925384e-06,
      "loss": 0.4263,
      "step": 24517
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5935691595077515,
      "learning_rate": 2.6282205301037724e-06,
      "loss": 0.4077,
      "step": 24518
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.547265887260437,
      "learning_rate": 2.6275324297393044e-06,
      "loss": 0.4273,
      "step": 24519
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4813148975372314,
      "learning_rate": 2.626844405839112e-06,
      "loss": 0.3703,
      "step": 24520
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6980485916137695,
      "learning_rate": 2.626156458410336e-06,
      "loss": 0.9357,
      "step": 24521
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4908442497253418,
      "learning_rate": 2.6254685874601072e-06,
      "loss": 1.0147,
      "step": 24522
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5009355545043945,
      "learning_rate": 2.6247807929955628e-06,
      "loss": 0.388,
      "step": 24523
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5415376424789429,
      "learning_rate": 2.6240930750238338e-06,
      "loss": 0.3627,
      "step": 24524
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5739980936050415,
      "learning_rate": 2.623405433552053e-06,
      "loss": 0.4174,
      "step": 24525
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6503764390945435,
      "learning_rate": 2.6227178685873567e-06,
      "loss": 0.3611,
      "step": 24526
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7750033140182495,
      "learning_rate": 2.622030380136873e-06,
      "loss": 0.434,
      "step": 24527
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5508763790130615,
      "learning_rate": 2.6213429682077295e-06,
      "loss": 0.3782,
      "step": 24528
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5079303979873657,
      "learning_rate": 2.6206556328070587e-06,
      "loss": 0.4197,
      "step": 24529
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4657469987869263,
      "learning_rate": 2.619968373941989e-06,
      "loss": 0.3663,
      "step": 24530
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.568536639213562,
      "learning_rate": 2.6192811916196513e-06,
      "loss": 0.4347,
      "step": 24531
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4274914264678955,
      "learning_rate": 2.6185940858471704e-06,
      "loss": 0.3516,
      "step": 24532
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6864737272262573,
      "learning_rate": 2.61790705663167e-06,
      "loss": 0.9736,
      "step": 24533
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4141477346420288,
      "learning_rate": 2.617220103980277e-06,
      "loss": 0.9301,
      "step": 24534
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5612174272537231,
      "learning_rate": 2.6165332279001197e-06,
      "loss": 0.4053,
      "step": 24535
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.520991563796997,
      "learning_rate": 2.6158464283983175e-06,
      "loss": 0.4132,
      "step": 24536
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.648258090019226,
      "learning_rate": 2.615159705481999e-06,
      "loss": 0.4346,
      "step": 24537
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4979976415634155,
      "learning_rate": 2.614473059158279e-06,
      "loss": 0.3695,
      "step": 24538
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.721742868423462,
      "learning_rate": 2.613786489434287e-06,
      "loss": 0.4787,
      "step": 24539
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5676085948944092,
      "learning_rate": 2.6130999963171376e-06,
      "loss": 0.3862,
      "step": 24540
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6601648330688477,
      "learning_rate": 2.612413579813953e-06,
      "loss": 0.4139,
      "step": 24541
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5969722270965576,
      "learning_rate": 2.6117272399318562e-06,
      "loss": 0.3846,
      "step": 24542
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5298088788986206,
      "learning_rate": 2.6110409766779623e-06,
      "loss": 0.4202,
      "step": 24543
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5136241912841797,
      "learning_rate": 2.610354790059386e-06,
      "loss": 0.3838,
      "step": 24544
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5161200761795044,
      "learning_rate": 2.6096686800832473e-06,
      "loss": 0.4148,
      "step": 24545
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5550687313079834,
      "learning_rate": 2.6089826467566624e-06,
      "loss": 0.3729,
      "step": 24546
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4716713428497314,
      "learning_rate": 2.608296690086749e-06,
      "loss": 0.412,
      "step": 24547
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4943572282791138,
      "learning_rate": 2.6076108100806188e-06,
      "loss": 0.3586,
      "step": 24548
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.544630765914917,
      "learning_rate": 2.606925006745382e-06,
      "loss": 0.411,
      "step": 24549
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4682927131652832,
      "learning_rate": 2.6062392800881565e-06,
      "loss": 0.3738,
      "step": 24550
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.575562596321106,
      "learning_rate": 2.6055536301160543e-06,
      "loss": 0.4757,
      "step": 24551
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4807791709899902,
      "learning_rate": 2.604868056836183e-06,
      "loss": 0.3717,
      "step": 24552
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4930437803268433,
      "learning_rate": 2.604182560255658e-06,
      "loss": 0.3925,
      "step": 24553
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5714964866638184,
      "learning_rate": 2.6034971403815845e-06,
      "loss": 0.3766,
      "step": 24554
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6980901956558228,
      "learning_rate": 2.6028117972210754e-06,
      "loss": 0.4769,
      "step": 24555
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.559199571609497,
      "learning_rate": 2.6021265307812337e-06,
      "loss": 0.3939,
      "step": 24556
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5324597358703613,
      "learning_rate": 2.601441341069172e-06,
      "loss": 0.4048,
      "step": 24557
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6760808229446411,
      "learning_rate": 2.600756228091993e-06,
      "loss": 0.4146,
      "step": 24558
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6215373277664185,
      "learning_rate": 2.6000711918568057e-06,
      "loss": 0.3931,
      "step": 24559
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6113340854644775,
      "learning_rate": 2.599386232370711e-06,
      "loss": 0.3641,
      "step": 24560
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6005144119262695,
      "learning_rate": 2.5987013496408142e-06,
      "loss": 0.4531,
      "step": 24561
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.578892707824707,
      "learning_rate": 2.5980165436742233e-06,
      "loss": 0.359,
      "step": 24562
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5280647277832031,
      "learning_rate": 2.5973318144780367e-06,
      "loss": 0.4134,
      "step": 24563
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4726759195327759,
      "learning_rate": 2.5966471620593546e-06,
      "loss": 0.367,
      "step": 24564
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.556368350982666,
      "learning_rate": 2.5959625864252802e-06,
      "loss": 0.4099,
      "step": 24565
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5266674757003784,
      "learning_rate": 2.595278087582912e-06,
      "loss": 0.3918,
      "step": 24566
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.625037670135498,
      "learning_rate": 2.5945936655393545e-06,
      "loss": 0.4245,
      "step": 24567
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5728203058242798,
      "learning_rate": 2.5939093203017027e-06,
      "loss": 0.4248,
      "step": 24568
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.517166018486023,
      "learning_rate": 2.5932250518770517e-06,
      "loss": 0.3823,
      "step": 24569
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6464790105819702,
      "learning_rate": 2.5925408602725e-06,
      "loss": 0.3974,
      "step": 24570
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6975088119506836,
      "learning_rate": 2.591856745495148e-06,
      "loss": 0.4266,
      "step": 24571
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6933479309082031,
      "learning_rate": 2.5911727075520855e-06,
      "loss": 0.422,
      "step": 24572
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5604238510131836,
      "learning_rate": 2.5904887464504115e-06,
      "loss": 0.4122,
      "step": 24573
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5616356134414673,
      "learning_rate": 2.5898048621972136e-06,
      "loss": 0.3933,
      "step": 24574
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5945847034454346,
      "learning_rate": 2.5891210547995905e-06,
      "loss": 0.4668,
      "step": 24575
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5006741285324097,
      "learning_rate": 2.5884373242646355e-06,
      "loss": 0.3725,
      "step": 24576
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5649127960205078,
      "learning_rate": 2.5877536705994332e-06,
      "loss": 0.4209,
      "step": 24577
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5284756422042847,
      "learning_rate": 2.5870700938110815e-06,
      "loss": 0.3908,
      "step": 24578
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5415802001953125,
      "learning_rate": 2.5863865939066644e-06,
      "loss": 0.4616,
      "step": 24579
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5130118131637573,
      "learning_rate": 2.585703170893276e-06,
      "loss": 0.3824,
      "step": 24580
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.695210576057434,
      "learning_rate": 2.585019824777999e-06,
      "loss": 0.4558,
      "step": 24581
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6032758951187134,
      "learning_rate": 2.584336555567922e-06,
      "loss": 0.3912,
      "step": 24582
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5727512836456299,
      "learning_rate": 2.5836533632701377e-06,
      "loss": 0.4196,
      "step": 24583
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7195791006088257,
      "learning_rate": 2.5829702478917273e-06,
      "loss": 0.3722,
      "step": 24584
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8375290632247925,
      "learning_rate": 2.5822872094397735e-06,
      "loss": 0.9298,
      "step": 24585
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4845614433288574,
      "learning_rate": 2.581604247921361e-06,
      "loss": 1.0462,
      "step": 24586
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.588607668876648,
      "learning_rate": 2.580921363343576e-06,
      "loss": 0.3962,
      "step": 24587
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5475351810455322,
      "learning_rate": 2.5802385557135034e-06,
      "loss": 0.3954,
      "step": 24588
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7034623622894287,
      "learning_rate": 2.5795558250382215e-06,
      "loss": 0.4988,
      "step": 24589
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5406213998794556,
      "learning_rate": 2.578873171324808e-06,
      "loss": 0.3867,
      "step": 24590
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.538696527481079,
      "learning_rate": 2.5781905945803477e-06,
      "loss": 0.439,
      "step": 24591
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5405536890029907,
      "learning_rate": 2.5775080948119224e-06,
      "loss": 0.3671,
      "step": 24592
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5227482318878174,
      "learning_rate": 2.576825672026606e-06,
      "loss": 0.4089,
      "step": 24593
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5678913593292236,
      "learning_rate": 2.576143326231475e-06,
      "loss": 0.3987,
      "step": 24594
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.610416054725647,
      "learning_rate": 2.5754610574336104e-06,
      "loss": 0.4102,
      "step": 24595
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5603960752487183,
      "learning_rate": 2.5747788656400886e-06,
      "loss": 0.3763,
      "step": 24596
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5754410028457642,
      "learning_rate": 2.5740967508579808e-06,
      "loss": 0.3887,
      "step": 24597
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.529342532157898,
      "learning_rate": 2.5734147130943676e-06,
      "loss": 0.354,
      "step": 24598
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5196454524993896,
      "learning_rate": 2.572732752356316e-06,
      "loss": 0.4266,
      "step": 24599
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5257998704910278,
      "learning_rate": 2.572050868650906e-06,
      "loss": 0.3978,
      "step": 24600
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.555025339126587,
      "learning_rate": 2.5713690619852026e-06,
      "loss": 0.4546,
      "step": 24601
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4745784997940063,
      "learning_rate": 2.5706873323662816e-06,
      "loss": 0.3603,
      "step": 24602
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5460577011108398,
      "learning_rate": 2.5700056798012164e-06,
      "loss": 0.4341,
      "step": 24603
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4812599420547485,
      "learning_rate": 2.569324104297073e-06,
      "loss": 0.3728,
      "step": 24604
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6006866693496704,
      "learning_rate": 2.568642605860918e-06,
      "loss": 0.4657,
      "step": 24605
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5046391487121582,
      "learning_rate": 2.567961184499823e-06,
      "loss": 0.3895,
      "step": 24606
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.555413842201233,
      "learning_rate": 2.567279840220854e-06,
      "loss": 0.4099,
      "step": 24607
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4842265844345093,
      "learning_rate": 2.5665985730310826e-06,
      "loss": 0.398,
      "step": 24608
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6683385372161865,
      "learning_rate": 2.5659173829375694e-06,
      "loss": 0.4011,
      "step": 24609
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6756471395492554,
      "learning_rate": 2.5652362699473786e-06,
      "loss": 0.3984,
      "step": 24610
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5598820447921753,
      "learning_rate": 2.5645552340675763e-06,
      "loss": 0.432,
      "step": 24611
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5083171129226685,
      "learning_rate": 2.5638742753052293e-06,
      "loss": 0.3723,
      "step": 24612
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5533548593521118,
      "learning_rate": 2.5631933936673937e-06,
      "loss": 0.4512,
      "step": 24613
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4655609130859375,
      "learning_rate": 2.562512589161138e-06,
      "loss": 0.3648,
      "step": 24614
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5528557300567627,
      "learning_rate": 2.5618318617935166e-06,
      "loss": 0.4252,
      "step": 24615
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.483083963394165,
      "learning_rate": 2.5611512115715955e-06,
      "loss": 0.386,
      "step": 24616
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6427572965621948,
      "learning_rate": 2.5604706385024293e-06,
      "loss": 0.4154,
      "step": 24617
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5014286041259766,
      "learning_rate": 2.5597901425930794e-06,
      "loss": 0.3821,
      "step": 24618
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4875731468200684,
      "learning_rate": 2.5591097238506045e-06,
      "loss": 0.423,
      "step": 24619
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4457483291625977,
      "learning_rate": 2.558429382282063e-06,
      "loss": 0.361,
      "step": 24620
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.773658275604248,
      "learning_rate": 2.5577491178945035e-06,
      "loss": 0.4688,
      "step": 24621
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5470305681228638,
      "learning_rate": 2.557068930694987e-06,
      "loss": 0.3726,
      "step": 24622
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9606220722198486,
      "learning_rate": 2.5563888206905672e-06,
      "loss": 0.4301,
      "step": 24623
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5989323854446411,
      "learning_rate": 2.5557087878883014e-06,
      "loss": 0.3985,
      "step": 24624
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.51996648311615,
      "learning_rate": 2.5550288322952398e-06,
      "loss": 0.423,
      "step": 24625
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4362977743148804,
      "learning_rate": 2.554348953918431e-06,
      "loss": 0.3535,
      "step": 24626
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6448756456375122,
      "learning_rate": 2.5536691527649306e-06,
      "loss": 0.3858,
      "step": 24627
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5258554220199585,
      "learning_rate": 2.55298942884179e-06,
      "loss": 0.4534,
      "step": 24628
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5715134143829346,
      "learning_rate": 2.552309782156055e-06,
      "loss": 0.4109,
      "step": 24629
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4453173875808716,
      "learning_rate": 2.5516302127147808e-06,
      "loss": 0.3712,
      "step": 24630
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5715899467468262,
      "learning_rate": 2.5509507205250073e-06,
      "loss": 0.4566,
      "step": 24631
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.539177417755127,
      "learning_rate": 2.55027130559379e-06,
      "loss": 0.39,
      "step": 24632
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4810744524002075,
      "learning_rate": 2.54959196792817e-06,
      "loss": 0.3868,
      "step": 24633
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5212236642837524,
      "learning_rate": 2.5489127075351972e-06,
      "loss": 0.3665,
      "step": 24634
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6698671579360962,
      "learning_rate": 2.5482335244219114e-06,
      "loss": 0.4998,
      "step": 24635
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5670111179351807,
      "learning_rate": 2.547554418595364e-06,
      "loss": 0.4117,
      "step": 24636
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5465644598007202,
      "learning_rate": 2.54687539006259e-06,
      "loss": 0.4289,
      "step": 24637
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5253229141235352,
      "learning_rate": 2.5461964388306372e-06,
      "loss": 0.3669,
      "step": 24638
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5970399379730225,
      "learning_rate": 2.545517564906549e-06,
      "loss": 0.4724,
      "step": 24639
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.648549199104309,
      "learning_rate": 2.5448387682973643e-06,
      "loss": 0.3979,
      "step": 24640
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5840286016464233,
      "learning_rate": 2.5441600490101203e-06,
      "loss": 0.4584,
      "step": 24641
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4651577472686768,
      "learning_rate": 2.5434814070518586e-06,
      "loss": 0.3734,
      "step": 24642
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5455169677734375,
      "learning_rate": 2.542802842429618e-06,
      "loss": 0.4605,
      "step": 24643
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.644917368888855,
      "learning_rate": 2.5421243551504404e-06,
      "loss": 0.3851,
      "step": 24644
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.49183189868927,
      "learning_rate": 2.541445945221358e-06,
      "loss": 0.3958,
      "step": 24645
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4785882234573364,
      "learning_rate": 2.5407676126494063e-06,
      "loss": 0.3684,
      "step": 24646
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5288422107696533,
      "learning_rate": 2.540089357441621e-06,
      "loss": 0.3944,
      "step": 24647
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6212579011917114,
      "learning_rate": 2.5394111796050426e-06,
      "loss": 0.3961,
      "step": 24648
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6389473676681519,
      "learning_rate": 2.5387330791466957e-06,
      "loss": 0.4279,
      "step": 24649
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4039138555526733,
      "learning_rate": 2.5380550560736215e-06,
      "loss": 0.3511,
      "step": 24650
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4927525520324707,
      "learning_rate": 2.5373771103928456e-06,
      "loss": 0.4289,
      "step": 24651
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5655834674835205,
      "learning_rate": 2.536699242111406e-06,
      "loss": 0.3967,
      "step": 24652
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7489972114562988,
      "learning_rate": 2.5360214512363268e-06,
      "loss": 0.9272,
      "step": 24653
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3620117902755737,
      "learning_rate": 2.5353437377746394e-06,
      "loss": 0.9981,
      "step": 24654
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5077917575836182,
      "learning_rate": 2.5346661017333783e-06,
      "loss": 0.3914,
      "step": 24655
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5160048007965088,
      "learning_rate": 2.5339885431195666e-06,
      "loss": 0.3929,
      "step": 24656
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5173841714859009,
      "learning_rate": 2.53331106194023e-06,
      "loss": 0.4534,
      "step": 24657
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5371631383895874,
      "learning_rate": 2.532633658202397e-06,
      "loss": 0.3982,
      "step": 24658
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5602494478225708,
      "learning_rate": 2.5319563319130945e-06,
      "loss": 0.4468,
      "step": 24659
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5014050006866455,
      "learning_rate": 2.53127908307935e-06,
      "loss": 0.3764,
      "step": 24660
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5375357866287231,
      "learning_rate": 2.5306019117081837e-06,
      "loss": 0.417,
      "step": 24661
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5230523347854614,
      "learning_rate": 2.529924817806617e-06,
      "loss": 0.3773,
      "step": 24662
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5916887521743774,
      "learning_rate": 2.5292478013816747e-06,
      "loss": 0.4467,
      "step": 24663
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5087085962295532,
      "learning_rate": 2.528570862440383e-06,
      "loss": 0.389,
      "step": 24664
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5604571104049683,
      "learning_rate": 2.5278940009897557e-06,
      "loss": 0.4337,
      "step": 24665
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6709409952163696,
      "learning_rate": 2.527217217036818e-06,
      "loss": 0.3768,
      "step": 24666
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5314432382583618,
      "learning_rate": 2.5265405105885855e-06,
      "loss": 0.4256,
      "step": 24667
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4636603593826294,
      "learning_rate": 2.525863881652082e-06,
      "loss": 0.3715,
      "step": 24668
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7624279260635376,
      "learning_rate": 2.5251873302343177e-06,
      "loss": 0.4372,
      "step": 24669
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.495527982711792,
      "learning_rate": 2.5245108563423137e-06,
      "loss": 0.3578,
      "step": 24670
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.566876769065857,
      "learning_rate": 2.5238344599830898e-06,
      "loss": 0.4105,
      "step": 24671
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5066678524017334,
      "learning_rate": 2.5231581411636564e-06,
      "loss": 0.4129,
      "step": 24672
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.611533761024475,
      "learning_rate": 2.522481899891027e-06,
      "loss": 0.4404,
      "step": 24673
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5629518032073975,
      "learning_rate": 2.521805736172218e-06,
      "loss": 0.368,
      "step": 24674
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.560023307800293,
      "learning_rate": 2.5211296500142434e-06,
      "loss": 0.4247,
      "step": 24675
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4803518056869507,
      "learning_rate": 2.520453641424114e-06,
      "loss": 0.3841,
      "step": 24676
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5489412546157837,
      "learning_rate": 2.5197777104088385e-06,
      "loss": 0.4146,
      "step": 24677
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5443415641784668,
      "learning_rate": 2.519101856975429e-06,
      "loss": 0.3906,
      "step": 24678
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5467735528945923,
      "learning_rate": 2.5184260811308958e-06,
      "loss": 0.4203,
      "step": 24679
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4903514385223389,
      "learning_rate": 2.5177503828822514e-06,
      "loss": 0.3741,
      "step": 24680
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6470175981521606,
      "learning_rate": 2.5170747622365e-06,
      "loss": 0.4034,
      "step": 24681
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.577120304107666,
      "learning_rate": 2.516399219200646e-06,
      "loss": 0.3664,
      "step": 24682
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5591968297958374,
      "learning_rate": 2.515723753781699e-06,
      "loss": 0.4351,
      "step": 24683
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5136098861694336,
      "learning_rate": 2.5150483659866677e-06,
      "loss": 0.3637,
      "step": 24684
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.7653632164001465,
      "learning_rate": 2.5143730558225497e-06,
      "loss": 0.4169,
      "step": 24685
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5840139389038086,
      "learning_rate": 2.513697823296358e-06,
      "loss": 0.3816,
      "step": 24686
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8168436288833618,
      "learning_rate": 2.513022668415087e-06,
      "loss": 0.9026,
      "step": 24687
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3937197923660278,
      "learning_rate": 2.5123475911857463e-06,
      "loss": 1.0088,
      "step": 24688
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5630531311035156,
      "learning_rate": 2.5116725916153316e-06,
      "loss": 0.4289,
      "step": 24689
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7060716152191162,
      "learning_rate": 2.5109976697108475e-06,
      "loss": 0.3789,
      "step": 24690
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.494454026222229,
      "learning_rate": 2.5103228254792945e-06,
      "loss": 0.3967,
      "step": 24691
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5912765264511108,
      "learning_rate": 2.509648058927672e-06,
      "loss": 0.3601,
      "step": 24692
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6054633855819702,
      "learning_rate": 2.5089733700629724e-06,
      "loss": 0.448,
      "step": 24693
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4162076711654663,
      "learning_rate": 2.508298758892198e-06,
      "loss": 0.3687,
      "step": 24694
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7178065776824951,
      "learning_rate": 2.507624225422346e-06,
      "loss": 0.4308,
      "step": 24695
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4881165027618408,
      "learning_rate": 2.506949769660414e-06,
      "loss": 0.3763,
      "step": 24696
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5103881359100342,
      "learning_rate": 2.5062753916133952e-06,
      "loss": 0.425,
      "step": 24697
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6689949035644531,
      "learning_rate": 2.505601091288281e-06,
      "loss": 0.3967,
      "step": 24698
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5492255687713623,
      "learning_rate": 2.504926868692067e-06,
      "loss": 0.404,
      "step": 24699
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5058547258377075,
      "learning_rate": 2.5042527238317504e-06,
      "loss": 0.3812,
      "step": 24700
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6246094703674316,
      "learning_rate": 2.503578656714315e-06,
      "loss": 0.4142,
      "step": 24701
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4928940534591675,
      "learning_rate": 2.5029046673467603e-06,
      "loss": 0.3526,
      "step": 24702
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5522693395614624,
      "learning_rate": 2.5022307557360693e-06,
      "loss": 0.4075,
      "step": 24703
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5413002967834473,
      "learning_rate": 2.5015569218892387e-06,
      "loss": 0.3718,
      "step": 24704
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.58201003074646,
      "learning_rate": 2.5008831658132493e-06,
      "loss": 0.4427,
      "step": 24705
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0412890911102295,
      "learning_rate": 2.500209487515094e-06,
      "loss": 0.4238,
      "step": 24706
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5951035022735596,
      "learning_rate": 2.4995358870017616e-06,
      "loss": 0.4097,
      "step": 24707
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.486159086227417,
      "learning_rate": 2.498862364280237e-06,
      "loss": 0.3857,
      "step": 24708
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5158082246780396,
      "learning_rate": 2.4981889193575e-06,
      "loss": 0.4403,
      "step": 24709
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5045645236968994,
      "learning_rate": 2.4975155522405403e-06,
      "loss": 0.3668,
      "step": 24710
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.486363172531128,
      "learning_rate": 2.4968422629363455e-06,
      "loss": 0.3922,
      "step": 24711
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5874937772750854,
      "learning_rate": 2.4961690514518933e-06,
      "loss": 0.3909,
      "step": 24712
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5517412424087524,
      "learning_rate": 2.495495917794165e-06,
      "loss": 0.444,
      "step": 24713
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5260536670684814,
      "learning_rate": 2.494822861970144e-06,
      "loss": 0.3788,
      "step": 24714
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5413450002670288,
      "learning_rate": 2.4941498839868118e-06,
      "loss": 0.4312,
      "step": 24715
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5759000778198242,
      "learning_rate": 2.4934769838511495e-06,
      "loss": 0.4005,
      "step": 24716
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.622962236404419,
      "learning_rate": 2.492804161570136e-06,
      "loss": 0.3957,
      "step": 24717
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4842077493667603,
      "learning_rate": 2.4921314171507437e-06,
      "loss": 0.3608,
      "step": 24718
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5850582122802734,
      "learning_rate": 2.4914587505999554e-06,
      "loss": 0.4426,
      "step": 24719
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7683510780334473,
      "learning_rate": 2.4907861619247487e-06,
      "loss": 0.3979,
      "step": 24720
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5814104080200195,
      "learning_rate": 2.4901136511320945e-06,
      "loss": 0.4663,
      "step": 24721
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6287086009979248,
      "learning_rate": 2.489441218228974e-06,
      "loss": 0.4204,
      "step": 24722
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.825132131576538,
      "learning_rate": 2.488768863222355e-06,
      "loss": 0.9985,
      "step": 24723
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3948023319244385,
      "learning_rate": 2.4880965861192174e-06,
      "loss": 0.9721,
      "step": 24724
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.563057780265808,
      "learning_rate": 2.487424386926527e-06,
      "loss": 0.416,
      "step": 24725
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5331158638000488,
      "learning_rate": 2.4867522656512587e-06,
      "loss": 0.3625,
      "step": 24726
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7337653636932373,
      "learning_rate": 2.4860802223003865e-06,
      "loss": 0.5179,
      "step": 24727
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5086729526519775,
      "learning_rate": 2.485408256880878e-06,
      "loss": 0.3707,
      "step": 24728
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5424364805221558,
      "learning_rate": 2.4847363693996994e-06,
      "loss": 0.4233,
      "step": 24729
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5476404428482056,
      "learning_rate": 2.4840645598638223e-06,
      "loss": 0.4066,
      "step": 24730
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4900314807891846,
      "learning_rate": 2.4833928282802134e-06,
      "loss": 0.4204,
      "step": 24731
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4999239444732666,
      "learning_rate": 2.4827211746558444e-06,
      "loss": 0.3531,
      "step": 24732
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5646636486053467,
      "learning_rate": 2.482049598997678e-06,
      "loss": 0.4239,
      "step": 24733
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6135179996490479,
      "learning_rate": 2.4813781013126747e-06,
      "loss": 0.4015,
      "step": 24734
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5951963663101196,
      "learning_rate": 2.480706681607804e-06,
      "loss": 0.3986,
      "step": 24735
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6381440162658691,
      "learning_rate": 2.4800353398900325e-06,
      "loss": 0.3881,
      "step": 24736
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4996534585952759,
      "learning_rate": 2.4793640761663162e-06,
      "loss": 0.3912,
      "step": 24737
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4893250465393066,
      "learning_rate": 2.4786928904436237e-06,
      "loss": 0.375,
      "step": 24738
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5621230602264404,
      "learning_rate": 2.47802178272891e-06,
      "loss": 0.459,
      "step": 24739
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.557519555091858,
      "learning_rate": 2.477350753029142e-06,
      "loss": 0.3693,
      "step": 24740
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6788750886917114,
      "learning_rate": 2.4766798013512716e-06,
      "loss": 0.436,
      "step": 24741
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.483516812324524,
      "learning_rate": 2.4760089277022636e-06,
      "loss": 0.3628,
      "step": 24742
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5713998079299927,
      "learning_rate": 2.4753381320890766e-06,
      "loss": 0.4586,
      "step": 24743
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.481818675994873,
      "learning_rate": 2.4746674145186666e-06,
      "loss": 0.3791,
      "step": 24744
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5459249019622803,
      "learning_rate": 2.4739967749979855e-06,
      "loss": 0.3927,
      "step": 24745
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5116685628890991,
      "learning_rate": 2.473326213533992e-06,
      "loss": 0.3792,
      "step": 24746
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6036654710769653,
      "learning_rate": 2.472655730133643e-06,
      "loss": 0.3932,
      "step": 24747
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5161559581756592,
      "learning_rate": 2.471985324803895e-06,
      "loss": 0.3727,
      "step": 24748
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6350979804992676,
      "learning_rate": 2.4713149975516925e-06,
      "loss": 0.4433,
      "step": 24749
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5518404245376587,
      "learning_rate": 2.4706447483839925e-06,
      "loss": 0.378,
      "step": 24750
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6221131086349487,
      "learning_rate": 2.4699745773077453e-06,
      "loss": 0.9614,
      "step": 24751
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4228227138519287,
      "learning_rate": 2.469304484329906e-06,
      "loss": 0.9927,
      "step": 24752
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.507959246635437,
      "learning_rate": 2.4686344694574227e-06,
      "loss": 0.4363,
      "step": 24753
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5236703157424927,
      "learning_rate": 2.467964532697239e-06,
      "loss": 0.402,
      "step": 24754
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.583701729774475,
      "learning_rate": 2.4672946740563087e-06,
      "loss": 0.4052,
      "step": 24755
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6321042776107788,
      "learning_rate": 2.4666248935415804e-06,
      "loss": 0.405,
      "step": 24756
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4740420579910278,
      "learning_rate": 2.4659551911599943e-06,
      "loss": 0.4206,
      "step": 24757
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5127872228622437,
      "learning_rate": 2.465285566918505e-06,
      "loss": 0.3768,
      "step": 24758
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5681766271591187,
      "learning_rate": 2.464616020824049e-06,
      "loss": 0.4347,
      "step": 24759
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.692262649536133,
      "learning_rate": 2.4639465528835783e-06,
      "loss": 0.3942,
      "step": 24760
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6383178234100342,
      "learning_rate": 2.463277163104029e-06,
      "loss": 0.453,
      "step": 24761
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5519638061523438,
      "learning_rate": 2.4626078514923478e-06,
      "loss": 0.3582,
      "step": 24762
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.545147180557251,
      "learning_rate": 2.4619386180554783e-06,
      "loss": 0.4353,
      "step": 24763
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5677434206008911,
      "learning_rate": 2.4612694628003598e-06,
      "loss": 0.3922,
      "step": 24764
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6367123126983643,
      "learning_rate": 2.46060038573393e-06,
      "loss": 0.4363,
      "step": 24765
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5047377347946167,
      "learning_rate": 2.459931386863129e-06,
      "loss": 0.3889,
      "step": 24766
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6292377710342407,
      "learning_rate": 2.459262466194897e-06,
      "loss": 0.4155,
      "step": 24767
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.501265287399292,
      "learning_rate": 2.4585936237361753e-06,
      "loss": 0.379,
      "step": 24768
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4839577674865723,
      "learning_rate": 2.4579248594938955e-06,
      "loss": 0.3771,
      "step": 24769
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5148732662200928,
      "learning_rate": 2.4572561734749935e-06,
      "loss": 0.3814,
      "step": 24770
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5802403688430786,
      "learning_rate": 2.456587565686406e-06,
      "loss": 0.4705,
      "step": 24771
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6288527250289917,
      "learning_rate": 2.4559190361350705e-06,
      "loss": 0.383,
      "step": 24772
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5595988035202026,
      "learning_rate": 2.455250584827916e-06,
      "loss": 0.4149,
      "step": 24773
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5177992582321167,
      "learning_rate": 2.4545822117718797e-06,
      "loss": 0.3929,
      "step": 24774
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6164053678512573,
      "learning_rate": 2.45391391697389e-06,
      "loss": 0.4071,
      "step": 24775
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.598807454109192,
      "learning_rate": 2.4532457004408818e-06,
      "loss": 0.3881,
      "step": 24776
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6174949407577515,
      "learning_rate": 2.452577562179781e-06,
      "loss": 0.4373,
      "step": 24777
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.485494613647461,
      "learning_rate": 2.4519095021975193e-06,
      "loss": 0.3696,
      "step": 24778
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5296592712402344,
      "learning_rate": 2.45124152050103e-06,
      "loss": 0.4171,
      "step": 24779
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5347905158996582,
      "learning_rate": 2.450573617097236e-06,
      "loss": 0.3884,
      "step": 24780
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.505136489868164,
      "learning_rate": 2.4499057919930635e-06,
      "loss": 0.427,
      "step": 24781
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4835889339447021,
      "learning_rate": 2.449238045195441e-06,
      "loss": 0.3735,
      "step": 24782
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.589246153831482,
      "learning_rate": 2.4485703767112932e-06,
      "loss": 0.4219,
      "step": 24783
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.527675986289978,
      "learning_rate": 2.4479027865475535e-06,
      "loss": 0.3787,
      "step": 24784
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5430747270584106,
      "learning_rate": 2.4472352747111315e-06,
      "loss": 0.3889,
      "step": 24785
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5341708660125732,
      "learning_rate": 2.4465678412089577e-06,
      "loss": 0.3781,
      "step": 24786
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.471285104751587,
      "learning_rate": 2.445900486047954e-06,
      "loss": 0.4111,
      "step": 24787
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5024051666259766,
      "learning_rate": 2.445233209235044e-06,
      "loss": 0.3842,
      "step": 24788
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4999195337295532,
      "learning_rate": 2.4445660107771472e-06,
      "loss": 0.3954,
      "step": 24789
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.473863959312439,
      "learning_rate": 2.443898890681179e-06,
      "loss": 0.3846,
      "step": 24790
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4891798496246338,
      "learning_rate": 2.443231848954063e-06,
      "loss": 0.3984,
      "step": 24791
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6432183980941772,
      "learning_rate": 2.4425648856027183e-06,
      "loss": 0.4175,
      "step": 24792
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.588700771331787,
      "learning_rate": 2.4418980006340586e-06,
      "loss": 0.4288,
      "step": 24793
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5025367736816406,
      "learning_rate": 2.4412311940550058e-06,
      "loss": 0.3639,
      "step": 24794
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.586975336074829,
      "learning_rate": 2.440564465872469e-06,
      "loss": 0.4344,
      "step": 24795
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.689611554145813,
      "learning_rate": 2.4398978160933695e-06,
      "loss": 0.4117,
      "step": 24796
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4988696575164795,
      "learning_rate": 2.439231244724617e-06,
      "loss": 0.4332,
      "step": 24797
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.452269434928894,
      "learning_rate": 2.438564751773126e-06,
      "loss": 0.3607,
      "step": 24798
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5601969957351685,
      "learning_rate": 2.437898337245813e-06,
      "loss": 0.4547,
      "step": 24799
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6348750591278076,
      "learning_rate": 2.4372320011495865e-06,
      "loss": 0.3932,
      "step": 24800
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5990233421325684,
      "learning_rate": 2.436565743491355e-06,
      "loss": 0.4413,
      "step": 24801
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5502700805664062,
      "learning_rate": 2.4358995642780313e-06,
      "loss": 0.3627,
      "step": 24802
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.639196753501892,
      "learning_rate": 2.4352334635165253e-06,
      "loss": 0.4424,
      "step": 24803
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.533475399017334,
      "learning_rate": 2.4345674412137465e-06,
      "loss": 0.3473,
      "step": 24804
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6729028224945068,
      "learning_rate": 2.4339014973766018e-06,
      "loss": 0.4189,
      "step": 24805
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5632107257843018,
      "learning_rate": 2.4332356320119943e-06,
      "loss": 0.3565,
      "step": 24806
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5526357889175415,
      "learning_rate": 2.432569845126833e-06,
      "loss": 0.4372,
      "step": 24807
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5703970193862915,
      "learning_rate": 2.431904136728027e-06,
      "loss": 0.408,
      "step": 24808
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5104039907455444,
      "learning_rate": 2.4312385068224743e-06,
      "loss": 0.3916,
      "step": 24809
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5920485258102417,
      "learning_rate": 2.430572955417083e-06,
      "loss": 0.3854,
      "step": 24810
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5675106048583984,
      "learning_rate": 2.429907482518752e-06,
      "loss": 0.4017,
      "step": 24811
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5502042770385742,
      "learning_rate": 2.429242088134389e-06,
      "loss": 0.3704,
      "step": 24812
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5563445091247559,
      "learning_rate": 2.4285767722708887e-06,
      "loss": 0.3787,
      "step": 24813
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6177140474319458,
      "learning_rate": 2.4279115349351546e-06,
      "loss": 0.387,
      "step": 24814
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4886443614959717,
      "learning_rate": 2.427246376134088e-06,
      "loss": 0.4219,
      "step": 24815
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5032768249511719,
      "learning_rate": 2.4265812958745838e-06,
      "loss": 0.3906,
      "step": 24816
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4910368919372559,
      "learning_rate": 2.425916294163546e-06,
      "loss": 0.4146,
      "step": 24817
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5535622835159302,
      "learning_rate": 2.425251371007864e-06,
      "loss": 0.3715,
      "step": 24818
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.481075644493103,
      "learning_rate": 2.4245865264144376e-06,
      "loss": 0.3859,
      "step": 24819
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5002992153167725,
      "learning_rate": 2.4239217603901653e-06,
      "loss": 0.3574,
      "step": 24820
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5055702924728394,
      "learning_rate": 2.4232570729419403e-06,
      "loss": 0.3848,
      "step": 24821
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.482575535774231,
      "learning_rate": 2.422592464076652e-06,
      "loss": 0.3707,
      "step": 24822
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4877119064331055,
      "learning_rate": 2.421927933801196e-06,
      "loss": 0.4105,
      "step": 24823
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5545388460159302,
      "learning_rate": 2.4212634821224656e-06,
      "loss": 0.3374,
      "step": 24824
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7540998458862305,
      "learning_rate": 2.420599109047358e-06,
      "loss": 1.0274,
      "step": 24825
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.436747670173645,
      "learning_rate": 2.4199348145827517e-06,
      "loss": 0.9683,
      "step": 24826
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5096880197525024,
      "learning_rate": 2.4192705987355424e-06,
      "loss": 0.4389,
      "step": 24827
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5594552755355835,
      "learning_rate": 2.41860646151262e-06,
      "loss": 0.3349,
      "step": 24828
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7476286888122559,
      "learning_rate": 2.4179424029208744e-06,
      "loss": 0.421,
      "step": 24829
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6104137897491455,
      "learning_rate": 2.4172784229671897e-06,
      "loss": 0.3681,
      "step": 24830
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4912582635879517,
      "learning_rate": 2.416614521658451e-06,
      "loss": 0.4028,
      "step": 24831
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5115230083465576,
      "learning_rate": 2.4159506990015456e-06,
      "loss": 0.3615,
      "step": 24832
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5717347860336304,
      "learning_rate": 2.415286955003362e-06,
      "loss": 0.4183,
      "step": 24833
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5244431495666504,
      "learning_rate": 2.4146232896707788e-06,
      "loss": 0.3668,
      "step": 24834
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4973652362823486,
      "learning_rate": 2.413959703010684e-06,
      "loss": 0.4067,
      "step": 24835
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5781304836273193,
      "learning_rate": 2.413296195029955e-06,
      "loss": 0.3963,
      "step": 24836
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4943879842758179,
      "learning_rate": 2.4126327657354797e-06,
      "loss": 0.4078,
      "step": 24837
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5177955627441406,
      "learning_rate": 2.4119694151341324e-06,
      "loss": 0.3786,
      "step": 24838
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5539422035217285,
      "learning_rate": 2.4113061432327956e-06,
      "loss": 0.4338,
      "step": 24839
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5776721239089966,
      "learning_rate": 2.410642950038352e-06,
      "loss": 0.4002,
      "step": 24840
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5392590761184692,
      "learning_rate": 2.4099798355576776e-06,
      "loss": 0.4047,
      "step": 24841
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5246798992156982,
      "learning_rate": 2.409316799797645e-06,
      "loss": 0.3664,
      "step": 24842
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6421399116516113,
      "learning_rate": 2.4086538427651375e-06,
      "loss": 0.4338,
      "step": 24843
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4983373880386353,
      "learning_rate": 2.4079909644670276e-06,
      "loss": 0.3556,
      "step": 24844
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6083396673202515,
      "learning_rate": 2.4073281649101944e-06,
      "loss": 0.5013,
      "step": 24845
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5700443983078003,
      "learning_rate": 2.406665444101509e-06,
      "loss": 0.4066,
      "step": 24846
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.571529507637024,
      "learning_rate": 2.406002802047843e-06,
      "loss": 0.4151,
      "step": 24847
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5492000579833984,
      "learning_rate": 2.405340238756072e-06,
      "loss": 0.3744,
      "step": 24848
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6276326179504395,
      "learning_rate": 2.404677754233069e-06,
      "loss": 0.4052,
      "step": 24849
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6333564519882202,
      "learning_rate": 2.404015348485701e-06,
      "loss": 0.4229,
      "step": 24850
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6348451375961304,
      "learning_rate": 2.4033530215208434e-06,
      "loss": 0.415,
      "step": 24851
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.778613805770874,
      "learning_rate": 2.4026907733453597e-06,
      "loss": 0.3743,
      "step": 24852
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5598419904708862,
      "learning_rate": 2.4020286039661234e-06,
      "loss": 0.4103,
      "step": 24853
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6925396919250488,
      "learning_rate": 2.4013665133899977e-06,
      "loss": 0.3791,
      "step": 24854
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.535401463508606,
      "learning_rate": 2.4007045016238528e-06,
      "loss": 0.4184,
      "step": 24855
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4435375928878784,
      "learning_rate": 2.4000425686745564e-06,
      "loss": 0.3728,
      "step": 24856
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5038912296295166,
      "learning_rate": 2.399380714548972e-06,
      "loss": 0.4342,
      "step": 24857
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5087095499038696,
      "learning_rate": 2.39871893925396e-06,
      "loss": 0.4083,
      "step": 24858
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6509426832199097,
      "learning_rate": 2.398057242796389e-06,
      "loss": 0.9361,
      "step": 24859
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.449346661567688,
      "learning_rate": 2.3973956251831186e-06,
      "loss": 0.9703,
      "step": 24860
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5802960395812988,
      "learning_rate": 2.39673408642102e-06,
      "loss": 0.4247,
      "step": 24861
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4863598346710205,
      "learning_rate": 2.396072626516941e-06,
      "loss": 0.3503,
      "step": 24862
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6895503997802734,
      "learning_rate": 2.3954112454777478e-06,
      "loss": 0.478,
      "step": 24863
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6890952587127686,
      "learning_rate": 2.3947499433103005e-06,
      "loss": 0.3581,
      "step": 24864
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4809215068817139,
      "learning_rate": 2.3940887200214615e-06,
      "loss": 0.3838,
      "step": 24865
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.534379243850708,
      "learning_rate": 2.3934275756180838e-06,
      "loss": 0.3719,
      "step": 24866
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5830042362213135,
      "learning_rate": 2.392766510107023e-06,
      "loss": 0.4061,
      "step": 24867
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.510844349861145,
      "learning_rate": 2.3921055234951384e-06,
      "loss": 0.3696,
      "step": 24868
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5680674314498901,
      "learning_rate": 2.391444615789288e-06,
      "loss": 0.4545,
      "step": 24869
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5637619495391846,
      "learning_rate": 2.3907837869963203e-06,
      "loss": 0.396,
      "step": 24870
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7309205532073975,
      "learning_rate": 2.3901230371230953e-06,
      "loss": 0.4573,
      "step": 24871
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4570120573043823,
      "learning_rate": 2.3894623661764605e-06,
      "loss": 0.3571,
      "step": 24872
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.63923180103302,
      "learning_rate": 2.3888017741632743e-06,
      "loss": 0.4282,
      "step": 24873
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.560018539428711,
      "learning_rate": 2.388141261090381e-06,
      "loss": 0.3682,
      "step": 24874
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5260480642318726,
      "learning_rate": 2.3874808269646356e-06,
      "loss": 0.433,
      "step": 24875
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.546421766281128,
      "learning_rate": 2.386820471792889e-06,
      "loss": 0.3778,
      "step": 24876
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5463825464248657,
      "learning_rate": 2.386160195581989e-06,
      "loss": 0.4078,
      "step": 24877
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6312450170516968,
      "learning_rate": 2.3854999983387794e-06,
      "loss": 0.4345,
      "step": 24878
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7469557523727417,
      "learning_rate": 2.3848398800701113e-06,
      "loss": 1.0388,
      "step": 24879
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4806852340698242,
      "learning_rate": 2.384179840782831e-06,
      "loss": 0.9267,
      "step": 24880
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.627098560333252,
      "learning_rate": 2.3835198804837865e-06,
      "loss": 0.4714,
      "step": 24881
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7558132410049438,
      "learning_rate": 2.3828599991798216e-06,
      "loss": 0.368,
      "step": 24882
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5359752178192139,
      "learning_rate": 2.382200196877774e-06,
      "loss": 0.4786,
      "step": 24883
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5566850900650024,
      "learning_rate": 2.381540473584494e-06,
      "loss": 0.4277,
      "step": 24884
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9160534143447876,
      "learning_rate": 2.380880829306823e-06,
      "loss": 0.3936,
      "step": 24885
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5500987768173218,
      "learning_rate": 2.380221264051599e-06,
      "loss": 0.381,
      "step": 24886
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6942003965377808,
      "learning_rate": 2.3795617778256684e-06,
      "loss": 0.4923,
      "step": 24887
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6022230386734009,
      "learning_rate": 2.378902370635865e-06,
      "loss": 0.3837,
      "step": 24888
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6277917623519897,
      "learning_rate": 2.3782430424890336e-06,
      "loss": 0.458,
      "step": 24889
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5050534009933472,
      "learning_rate": 2.3775837933920055e-06,
      "loss": 0.3719,
      "step": 24890
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5309507846832275,
      "learning_rate": 2.3769246233516243e-06,
      "loss": 0.4317,
      "step": 24891
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.543507695198059,
      "learning_rate": 2.3762655323747265e-06,
      "loss": 0.4109,
      "step": 24892
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6771842241287231,
      "learning_rate": 2.3756065204681455e-06,
      "loss": 0.4403,
      "step": 24893
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5167146921157837,
      "learning_rate": 2.374947587638715e-06,
      "loss": 0.3529,
      "step": 24894
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.475227952003479,
      "learning_rate": 2.3742887338932708e-06,
      "loss": 0.4338,
      "step": 24895
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5225372314453125,
      "learning_rate": 2.373629959238647e-06,
      "loss": 0.3903,
      "step": 24896
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5244219303131104,
      "learning_rate": 2.372971263681677e-06,
      "loss": 0.4326,
      "step": 24897
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6056532859802246,
      "learning_rate": 2.3723126472291935e-06,
      "loss": 0.3931,
      "step": 24898
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5126161575317383,
      "learning_rate": 2.371654109888021e-06,
      "loss": 0.4113,
      "step": 24899
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0799334049224854,
      "learning_rate": 2.3709956516649935e-06,
      "loss": 0.3956,
      "step": 24900
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5610699653625488,
      "learning_rate": 2.370337272566944e-06,
      "loss": 0.431,
      "step": 24901
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5548303127288818,
      "learning_rate": 2.369678972600696e-06,
      "loss": 0.3863,
      "step": 24902
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5256810188293457,
      "learning_rate": 2.3690207517730778e-06,
      "loss": 0.4338,
      "step": 24903
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5305445194244385,
      "learning_rate": 2.3683626100909164e-06,
      "loss": 0.4013,
      "step": 24904
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.549971103668213,
      "learning_rate": 2.3677045475610406e-06,
      "loss": 0.4154,
      "step": 24905
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.532515525817871,
      "learning_rate": 2.3670465641902706e-06,
      "loss": 0.3832,
      "step": 24906
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6017669439315796,
      "learning_rate": 2.366388659985436e-06,
      "loss": 0.4456,
      "step": 24907
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4948582649230957,
      "learning_rate": 2.365730834953356e-06,
      "loss": 0.3939,
      "step": 24908
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7244386672973633,
      "learning_rate": 2.3650730891008568e-06,
      "loss": 0.4067,
      "step": 24909
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5186325311660767,
      "learning_rate": 2.3644154224347558e-06,
      "loss": 0.3641,
      "step": 24910
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4889527559280396,
      "learning_rate": 2.3637578349618773e-06,
      "loss": 0.4089,
      "step": 24911
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6820484399795532,
      "learning_rate": 2.363100326689043e-06,
      "loss": 0.3906,
      "step": 24912
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6573991775512695,
      "learning_rate": 2.3624428976230716e-06,
      "loss": 0.4405,
      "step": 24913
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.532516360282898,
      "learning_rate": 2.3617855477707764e-06,
      "loss": 0.3695,
      "step": 24914
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6388801336288452,
      "learning_rate": 2.36112827713898e-06,
      "loss": 0.449,
      "step": 24915
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4971064329147339,
      "learning_rate": 2.360471085734498e-06,
      "loss": 0.3729,
      "step": 24916
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5728600025177002,
      "learning_rate": 2.3598139735641502e-06,
      "loss": 0.4296,
      "step": 24917
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.568373203277588,
      "learning_rate": 2.359156940634749e-06,
      "loss": 0.3794,
      "step": 24918
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9024983644485474,
      "learning_rate": 2.3584999869531057e-06,
      "loss": 1.0013,
      "step": 24919
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4817692041397095,
      "learning_rate": 2.357843112526036e-06,
      "loss": 1.0067,
      "step": 24920
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6469104290008545,
      "learning_rate": 2.357186317360357e-06,
      "loss": 0.4437,
      "step": 24921
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5797710418701172,
      "learning_rate": 2.356529601462876e-06,
      "loss": 0.3926,
      "step": 24922
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6583821773529053,
      "learning_rate": 2.3558729648404065e-06,
      "loss": 0.3943,
      "step": 24923
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6292167901992798,
      "learning_rate": 2.3552164074997565e-06,
      "loss": 0.415,
      "step": 24924
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6106034517288208,
      "learning_rate": 2.3545599294477383e-06,
      "loss": 0.4252,
      "step": 24925
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5050309896469116,
      "learning_rate": 2.353903530691157e-06,
      "loss": 0.3816,
      "step": 24926
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.558252215385437,
      "learning_rate": 2.3532472112368232e-06,
      "loss": 0.4562,
      "step": 24927
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.55803644657135,
      "learning_rate": 2.352590971091546e-06,
      "loss": 0.3933,
      "step": 24928
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5822373628616333,
      "learning_rate": 2.3519348102621297e-06,
      "loss": 0.4912,
      "step": 24929
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5396605730056763,
      "learning_rate": 2.351278728755376e-06,
      "loss": 0.4026,
      "step": 24930
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5912258625030518,
      "learning_rate": 2.3506227265780933e-06,
      "loss": 0.4415,
      "step": 24931
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.529489517211914,
      "learning_rate": 2.349966803737085e-06,
      "loss": 0.3979,
      "step": 24932
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7940828800201416,
      "learning_rate": 2.3493109602391573e-06,
      "loss": 0.983,
      "step": 24933
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4287141561508179,
      "learning_rate": 2.3486551960911075e-06,
      "loss": 0.9947,
      "step": 24934
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.657427191734314,
      "learning_rate": 2.347999511299737e-06,
      "loss": 0.4142,
      "step": 24935
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4919172525405884,
      "learning_rate": 2.3473439058718462e-06,
      "loss": 0.3857,
      "step": 24936
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5963910818099976,
      "learning_rate": 2.3466883798142403e-06,
      "loss": 0.4716,
      "step": 24937
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5779105424880981,
      "learning_rate": 2.3460329331337118e-06,
      "loss": 0.4132,
      "step": 24938
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5785424709320068,
      "learning_rate": 2.3453775658370624e-06,
      "loss": 0.4355,
      "step": 24939
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4589900970458984,
      "learning_rate": 2.344722277931086e-06,
      "loss": 0.3529,
      "step": 24940
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5643926858901978,
      "learning_rate": 2.3440670694225843e-06,
      "loss": 0.4798,
      "step": 24941
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.529595971107483,
      "learning_rate": 2.3434119403183463e-06,
      "loss": 0.3952,
      "step": 24942
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5563160181045532,
      "learning_rate": 2.3427568906251717e-06,
      "loss": 0.4335,
      "step": 24943
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4782735109329224,
      "learning_rate": 2.342101920349852e-06,
      "loss": 0.3488,
      "step": 24944
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5421985387802124,
      "learning_rate": 2.341447029499182e-06,
      "loss": 0.4397,
      "step": 24945
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.495193362236023,
      "learning_rate": 2.3407922180799515e-06,
      "loss": 0.3703,
      "step": 24946
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5443997383117676,
      "learning_rate": 2.340137486098952e-06,
      "loss": 0.3997,
      "step": 24947
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4721800088882446,
      "learning_rate": 2.3394828335629793e-06,
      "loss": 0.3701,
      "step": 24948
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8838908672332764,
      "learning_rate": 2.3388282604788192e-06,
      "loss": 0.4628,
      "step": 24949
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5249848365783691,
      "learning_rate": 2.3381737668532576e-06,
      "loss": 0.3784,
      "step": 24950
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4961800575256348,
      "learning_rate": 2.3375193526930864e-06,
      "loss": 0.4063,
      "step": 24951
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6110262870788574,
      "learning_rate": 2.3368650180050923e-06,
      "loss": 0.3985,
      "step": 24952
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.55745267868042,
      "learning_rate": 2.336210762796065e-06,
      "loss": 0.4439,
      "step": 24953
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4824860095977783,
      "learning_rate": 2.335556587072787e-06,
      "loss": 0.3788,
      "step": 24954
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5756351947784424,
      "learning_rate": 2.3349024908420403e-06,
      "loss": 0.4551,
      "step": 24955
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5097119808197021,
      "learning_rate": 2.334248474110612e-06,
      "loss": 0.3839,
      "step": 24956
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4900569915771484,
      "learning_rate": 2.3335945368852876e-06,
      "loss": 0.4353,
      "step": 24957
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4779773950576782,
      "learning_rate": 2.3329406791728447e-06,
      "loss": 0.3688,
      "step": 24958
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4593900442123413,
      "learning_rate": 2.3322869009800696e-06,
      "loss": 0.404,
      "step": 24959
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4580634832382202,
      "learning_rate": 2.331633202313739e-06,
      "loss": 0.3687,
      "step": 24960
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.625638484954834,
      "learning_rate": 2.330979583180636e-06,
      "loss": 0.4672,
      "step": 24961
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.526127815246582,
      "learning_rate": 2.330326043587535e-06,
      "loss": 0.3612,
      "step": 24962
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5955644845962524,
      "learning_rate": 2.329672583541218e-06,
      "loss": 0.4208,
      "step": 24963
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.539847731590271,
      "learning_rate": 2.329019203048465e-06,
      "loss": 0.375,
      "step": 24964
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6335047483444214,
      "learning_rate": 2.3283659021160496e-06,
      "loss": 0.4798,
      "step": 24965
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4857722520828247,
      "learning_rate": 2.327712680750743e-06,
      "loss": 0.369,
      "step": 24966
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5714199542999268,
      "learning_rate": 2.3270595389593263e-06,
      "loss": 0.428,
      "step": 24967
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4696325063705444,
      "learning_rate": 2.3264064767485705e-06,
      "loss": 0.3781,
      "step": 24968
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5742334127426147,
      "learning_rate": 2.3257534941252536e-06,
      "loss": 0.4943,
      "step": 24969
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6082866191864014,
      "learning_rate": 2.3251005910961456e-06,
      "loss": 0.3996,
      "step": 24970
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5160526037216187,
      "learning_rate": 2.3244477676680133e-06,
      "loss": 0.4209,
      "step": 24971
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4857993125915527,
      "learning_rate": 2.3237950238476313e-06,
      "loss": 0.3533,
      "step": 24972
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6860289573669434,
      "learning_rate": 2.3231423596417726e-06,
      "loss": 0.4168,
      "step": 24973
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5734102725982666,
      "learning_rate": 2.322489775057202e-06,
      "loss": 0.3834,
      "step": 24974
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6016273498535156,
      "learning_rate": 2.321837270100691e-06,
      "loss": 0.4458,
      "step": 24975
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5214753150939941,
      "learning_rate": 2.321184844779003e-06,
      "loss": 0.3743,
      "step": 24976
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.532125473022461,
      "learning_rate": 2.3205324990989096e-06,
      "loss": 0.4101,
      "step": 24977
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6305955648422241,
      "learning_rate": 2.319880233067172e-06,
      "loss": 0.3722,
      "step": 24978
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.499281406402588,
      "learning_rate": 2.31922804669056e-06,
      "loss": 0.3857,
      "step": 24979
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4562499523162842,
      "learning_rate": 2.3185759399758334e-06,
      "loss": 0.3471,
      "step": 24980
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4758009910583496,
      "learning_rate": 2.3179239129297595e-06,
      "loss": 0.3797,
      "step": 24981
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5314887762069702,
      "learning_rate": 2.317271965559096e-06,
      "loss": 0.3936,
      "step": 24982
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6215181350708008,
      "learning_rate": 2.3166200978706075e-06,
      "loss": 0.3848,
      "step": 24983
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5217701196670532,
      "learning_rate": 2.3159683098710583e-06,
      "loss": 0.3719,
      "step": 24984
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7533605098724365,
      "learning_rate": 2.3153166015672047e-06,
      "loss": 1.0073,
      "step": 24985
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4227603673934937,
      "learning_rate": 2.314664972965803e-06,
      "loss": 1.0811,
      "step": 24986
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5391004085540771,
      "learning_rate": 2.3140134240736167e-06,
      "loss": 0.4154,
      "step": 24987
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.655264973640442,
      "learning_rate": 2.3133619548974005e-06,
      "loss": 0.4004,
      "step": 24988
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5858405828475952,
      "learning_rate": 2.312710565443915e-06,
      "loss": 0.3861,
      "step": 24989
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4995887279510498,
      "learning_rate": 2.312059255719914e-06,
      "loss": 0.3845,
      "step": 24990
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5920840501785278,
      "learning_rate": 2.3114080257321492e-06,
      "loss": 0.4215,
      "step": 24991
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5823215246200562,
      "learning_rate": 2.3107568754873776e-06,
      "loss": 0.3949,
      "step": 24992
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.328005313873291,
      "learning_rate": 2.3101058049923563e-06,
      "loss": 0.4413,
      "step": 24993
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.558935284614563,
      "learning_rate": 2.309454814253832e-06,
      "loss": 0.3855,
      "step": 24994
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5757646560668945,
      "learning_rate": 2.3088039032785613e-06,
      "loss": 0.3975,
      "step": 24995
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6054447889328003,
      "learning_rate": 2.30815307207329e-06,
      "loss": 0.3737,
      "step": 24996
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7828996181488037,
      "learning_rate": 2.3075023206447745e-06,
      "loss": 0.4537,
      "step": 24997
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6052477359771729,
      "learning_rate": 2.3068516489997583e-06,
      "loss": 0.4195,
      "step": 24998
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4977343082427979,
      "learning_rate": 2.306201057144992e-06,
      "loss": 0.3974,
      "step": 24999
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5599312782287598,
      "learning_rate": 2.3055505450872262e-06,
      "loss": 0.3704,
      "step": 25000
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6812703609466553,
      "learning_rate": 2.304900112833206e-06,
      "loss": 0.4403,
      "step": 25001
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4962314367294312,
      "learning_rate": 2.304249760389673e-06,
      "loss": 0.3666,
      "step": 25002
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5176005363464355,
      "learning_rate": 2.3035994877633772e-06,
      "loss": 0.4028,
      "step": 25003
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6337281465530396,
      "learning_rate": 2.302949294961061e-06,
      "loss": 0.4066,
      "step": 25004
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.584035873413086,
      "learning_rate": 2.3022991819894724e-06,
      "loss": 0.4433,
      "step": 25005
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5393849611282349,
      "learning_rate": 2.3016491488553505e-06,
      "loss": 0.3746,
      "step": 25006
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4994184970855713,
      "learning_rate": 2.3009991955654333e-06,
      "loss": 0.4154,
      "step": 25007
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5061616897583008,
      "learning_rate": 2.3003493221264673e-06,
      "loss": 0.3836,
      "step": 25008
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4973938465118408,
      "learning_rate": 2.299699528545193e-06,
      "loss": 0.3988,
      "step": 25009
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4937853813171387,
      "learning_rate": 2.299049814828345e-06,
      "loss": 0.3776,
      "step": 25010
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6560590267181396,
      "learning_rate": 2.2984001809826684e-06,
      "loss": 0.4428,
      "step": 25011
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6660603284835815,
      "learning_rate": 2.2977506270148942e-06,
      "loss": 0.3702,
      "step": 25012
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.487997055053711,
      "learning_rate": 2.2971011529317654e-06,
      "loss": 0.4059,
      "step": 25013
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4779707193374634,
      "learning_rate": 2.2964517587400125e-06,
      "loss": 0.3741,
      "step": 25014
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.581915259361267,
      "learning_rate": 2.2958024444463733e-06,
      "loss": 0.4396,
      "step": 25015
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5550483465194702,
      "learning_rate": 2.2951532100575846e-06,
      "loss": 0.373,
      "step": 25016
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.753298044204712,
      "learning_rate": 2.2945040555803787e-06,
      "loss": 0.4294,
      "step": 25017
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5032713413238525,
      "learning_rate": 2.293854981021485e-06,
      "loss": 0.371,
      "step": 25018
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.452240228652954,
      "learning_rate": 2.2932059863876364e-06,
      "loss": 0.4273,
      "step": 25019
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5061070919036865,
      "learning_rate": 2.2925570716855695e-06,
      "loss": 0.3781,
      "step": 25020
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5785878896713257,
      "learning_rate": 2.2919082369220104e-06,
      "loss": 0.4947,
      "step": 25021
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.104931354522705,
      "learning_rate": 2.291259482103686e-06,
      "loss": 0.3701,
      "step": 25022
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5731953382492065,
      "learning_rate": 2.2906108072373277e-06,
      "loss": 0.4413,
      "step": 25023
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5083287954330444,
      "learning_rate": 2.289962212329663e-06,
      "loss": 0.3718,
      "step": 25024
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4792486429214478,
      "learning_rate": 2.2893136973874217e-06,
      "loss": 0.4192,
      "step": 25025
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4825632572174072,
      "learning_rate": 2.288665262417328e-06,
      "loss": 0.3916,
      "step": 25026
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4978069067001343,
      "learning_rate": 2.288016907426103e-06,
      "loss": 0.385,
      "step": 25027
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5797131061553955,
      "learning_rate": 2.2873686324204757e-06,
      "loss": 0.4578,
      "step": 25028
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5790388584136963,
      "learning_rate": 2.2867204374071704e-06,
      "loss": 0.4411,
      "step": 25029
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6017811298370361,
      "learning_rate": 2.286072322392907e-06,
      "loss": 0.3593,
      "step": 25030
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5053001642227173,
      "learning_rate": 2.28542428738441e-06,
      "loss": 0.3654,
      "step": 25031
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5276087522506714,
      "learning_rate": 2.2847763323883975e-06,
      "loss": 0.3853,
      "step": 25032
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5178688764572144,
      "learning_rate": 2.284128457411595e-06,
      "loss": 0.423,
      "step": 25033
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4856253862380981,
      "learning_rate": 2.2834806624607154e-06,
      "loss": 0.3586,
      "step": 25034
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5821635723114014,
      "learning_rate": 2.2828329475424803e-06,
      "loss": 0.4259,
      "step": 25035
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5300977230072021,
      "learning_rate": 2.2821853126636105e-06,
      "loss": 0.3867,
      "step": 25036
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7293317317962646,
      "learning_rate": 2.2815377578308217e-06,
      "loss": 0.4389,
      "step": 25037
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4950822591781616,
      "learning_rate": 2.280890283050825e-06,
      "loss": 0.3769,
      "step": 25038
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5934146642684937,
      "learning_rate": 2.2802428883303394e-06,
      "loss": 0.4453,
      "step": 25039
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4633303880691528,
      "learning_rate": 2.2795955736760798e-06,
      "loss": 0.3875,
      "step": 25040
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6405529975891113,
      "learning_rate": 2.278948339094763e-06,
      "loss": 0.46,
      "step": 25041
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5207490921020508,
      "learning_rate": 2.278301184593098e-06,
      "loss": 0.3629,
      "step": 25042
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5463999509811401,
      "learning_rate": 2.2776541101777948e-06,
      "loss": 0.4136,
      "step": 25043
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5133293867111206,
      "learning_rate": 2.2770071158555673e-06,
      "loss": 0.3963,
      "step": 25044
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.652951717376709,
      "learning_rate": 2.2763602016331278e-06,
      "loss": 0.4265,
      "step": 25045
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.442665934562683,
      "learning_rate": 2.2757133675171818e-06,
      "loss": 0.3744,
      "step": 25046
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4786230325698853,
      "learning_rate": 2.2750666135144416e-06,
      "loss": 0.3984,
      "step": 25047
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5798758268356323,
      "learning_rate": 2.2744199396316115e-06,
      "loss": 0.402,
      "step": 25048
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.613516926765442,
      "learning_rate": 2.273773345875403e-06,
      "loss": 0.4794,
      "step": 25049
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5166374444961548,
      "learning_rate": 2.273126832252517e-06,
      "loss": 0.3681,
      "step": 25050
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8297139406204224,
      "learning_rate": 2.272480398769662e-06,
      "loss": 0.3946,
      "step": 25051
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5543323755264282,
      "learning_rate": 2.2718340454335442e-06,
      "loss": 0.3769,
      "step": 25052
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.5289463996887207,
      "learning_rate": 2.271187772250866e-06,
      "loss": 0.4351,
      "step": 25053
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.528185486793518,
      "learning_rate": 2.270541579228326e-06,
      "loss": 0.3691,
      "step": 25054
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9939333200454712,
      "learning_rate": 2.26989546637263e-06,
      "loss": 0.4798,
      "step": 25055
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.524598479270935,
      "learning_rate": 2.2692494336904804e-06,
      "loss": 0.379,
      "step": 25056
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6253604888916016,
      "learning_rate": 2.2686034811885783e-06,
      "loss": 0.4067,
      "step": 25057
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5449823141098022,
      "learning_rate": 2.26795760887362e-06,
      "loss": 0.3683,
      "step": 25058
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6842859983444214,
      "learning_rate": 2.2673118167523046e-06,
      "loss": 0.4645,
      "step": 25059
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4936275482177734,
      "learning_rate": 2.26666610483133e-06,
      "loss": 0.3672,
      "step": 25060
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5426721572875977,
      "learning_rate": 2.2660204731173963e-06,
      "loss": 0.4624,
      "step": 25061
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4959747791290283,
      "learning_rate": 2.265374921617198e-06,
      "loss": 0.3903,
      "step": 25062
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5544334650039673,
      "learning_rate": 2.2647294503374274e-06,
      "loss": 0.4125,
      "step": 25063
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5054951906204224,
      "learning_rate": 2.264084059284782e-06,
      "loss": 0.3633,
      "step": 25064
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4838461875915527,
      "learning_rate": 2.2634387484659547e-06,
      "loss": 0.4005,
      "step": 25065
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.480554223060608,
      "learning_rate": 2.2627935178876425e-06,
      "loss": 0.364,
      "step": 25066
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5445185899734497,
      "learning_rate": 2.262148367556534e-06,
      "loss": 0.4171,
      "step": 25067
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4849269390106201,
      "learning_rate": 2.261503297479316e-06,
      "loss": 0.3618,
      "step": 25068
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.633561372756958,
      "learning_rate": 2.2608583076626846e-06,
      "loss": 0.4335,
      "step": 25069
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4909512996673584,
      "learning_rate": 2.260213398113331e-06,
      "loss": 0.3658,
      "step": 25070
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.525296926498413,
      "learning_rate": 2.2595685688379378e-06,
      "loss": 0.4241,
      "step": 25071
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.558256983757019,
      "learning_rate": 2.258923819843198e-06,
      "loss": 0.3893,
      "step": 25072
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9278018474578857,
      "learning_rate": 2.258279151135796e-06,
      "loss": 0.4234,
      "step": 25073
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4981718063354492,
      "learning_rate": 2.2576345627224204e-06,
      "loss": 0.3867,
      "step": 25074
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4838285446166992,
      "learning_rate": 2.256990054609752e-06,
      "loss": 0.3934,
      "step": 25075
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5764782428741455,
      "learning_rate": 2.2563456268044793e-06,
      "loss": 0.374,
      "step": 25076
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.515071153640747,
      "learning_rate": 2.255701279313287e-06,
      "loss": 0.4042,
      "step": 25077
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5234050750732422,
      "learning_rate": 2.2550570121428573e-06,
      "loss": 0.3775,
      "step": 25078
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5098901987075806,
      "learning_rate": 2.2544128252998677e-06,
      "loss": 0.4011,
      "step": 25079
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.559339165687561,
      "learning_rate": 2.2537687187910027e-06,
      "loss": 0.3656,
      "step": 25080
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4913427829742432,
      "learning_rate": 2.2531246926229435e-06,
      "loss": 0.3876,
      "step": 25081
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6379761695861816,
      "learning_rate": 2.2524807468023713e-06,
      "loss": 0.4136,
      "step": 25082
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6121326684951782,
      "learning_rate": 2.251836881335964e-06,
      "loss": 0.4411,
      "step": 25083
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5534048080444336,
      "learning_rate": 2.251193096230394e-06,
      "loss": 0.4204,
      "step": 25084
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5096080303192139,
      "learning_rate": 2.2505493914923426e-06,
      "loss": 0.3919,
      "step": 25085
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5618414878845215,
      "learning_rate": 2.2499057671284895e-06,
      "loss": 0.3778,
      "step": 25086
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.568957805633545,
      "learning_rate": 2.2492622231455042e-06,
      "loss": 0.434,
      "step": 25087
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5530468225479126,
      "learning_rate": 2.2486187595500652e-06,
      "loss": 0.3763,
      "step": 25088
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5798074007034302,
      "learning_rate": 2.2479753763488433e-06,
      "loss": 0.3829,
      "step": 25089
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4926326274871826,
      "learning_rate": 2.247332073548515e-06,
      "loss": 0.3724,
      "step": 25090
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6166647672653198,
      "learning_rate": 2.2466888511557473e-06,
      "loss": 0.439,
      "step": 25091
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5591988563537598,
      "learning_rate": 2.2460457091772136e-06,
      "loss": 0.3957,
      "step": 25092
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5479203462600708,
      "learning_rate": 2.245402647619589e-06,
      "loss": 0.4067,
      "step": 25093
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5269705057144165,
      "learning_rate": 2.244759666489539e-06,
      "loss": 0.3748,
      "step": 25094
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5397065877914429,
      "learning_rate": 2.24411676579373e-06,
      "loss": 0.4206,
      "step": 25095
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5950653553009033,
      "learning_rate": 2.243473945538832e-06,
      "loss": 0.3832,
      "step": 25096
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7644320726394653,
      "learning_rate": 2.242831205731515e-06,
      "loss": 0.9586,
      "step": 25097
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4878443479537964,
      "learning_rate": 2.2421885463784434e-06,
      "loss": 0.9647,
      "step": 25098
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5036661624908447,
      "learning_rate": 2.2415459674862795e-06,
      "loss": 0.4459,
      "step": 25099
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4960001707077026,
      "learning_rate": 2.240903469061689e-06,
      "loss": 0.3652,
      "step": 25100
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6726523637771606,
      "learning_rate": 2.2402610511113376e-06,
      "loss": 0.4245,
      "step": 25101
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.520870566368103,
      "learning_rate": 2.23961871364189e-06,
      "loss": 0.3863,
      "step": 25102
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5823395252227783,
      "learning_rate": 2.238976456660007e-06,
      "loss": 0.3956,
      "step": 25103
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5883069038391113,
      "learning_rate": 2.238334280172345e-06,
      "loss": 0.3864,
      "step": 25104
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.1825973987579346,
      "learning_rate": 2.237692184185568e-06,
      "loss": 0.4197,
      "step": 25105
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.716342806816101,
      "learning_rate": 2.2370501687063385e-06,
      "loss": 0.4114,
      "step": 25106
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.614357829093933,
      "learning_rate": 2.2364082337413097e-06,
      "loss": 0.4465,
      "step": 25107
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5964943170547485,
      "learning_rate": 2.235766379297145e-06,
      "loss": 0.4084,
      "step": 25108
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6797467470169067,
      "learning_rate": 2.2351246053804955e-06,
      "loss": 0.4373,
      "step": 25109
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6136066913604736,
      "learning_rate": 2.234482911998023e-06,
      "loss": 0.3868,
      "step": 25110
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5368949174880981,
      "learning_rate": 2.2338412991563785e-06,
      "loss": 0.4006,
      "step": 25111
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.537672996520996,
      "learning_rate": 2.2331997668622176e-06,
      "loss": 0.3678,
      "step": 25112
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5458406209945679,
      "learning_rate": 2.2325583151221973e-06,
      "loss": 0.4579,
      "step": 25113
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5381553173065186,
      "learning_rate": 2.231916943942969e-06,
      "loss": 0.3679,
      "step": 25114
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.54112708568573,
      "learning_rate": 2.231275653331181e-06,
      "loss": 0.4255,
      "step": 25115
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6191000938415527,
      "learning_rate": 2.2306344432934867e-06,
      "loss": 0.4146,
      "step": 25116
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5458248853683472,
      "learning_rate": 2.2299933138365382e-06,
      "loss": 0.4811,
      "step": 25117
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5875120162963867,
      "learning_rate": 2.2293522649669854e-06,
      "loss": 0.3902,
      "step": 25118
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5101133584976196,
      "learning_rate": 2.228711296691476e-06,
      "loss": 0.3934,
      "step": 25119
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4937776327133179,
      "learning_rate": 2.228070409016655e-06,
      "loss": 0.4302,
      "step": 25120
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5064899921417236,
      "learning_rate": 2.2274296019491705e-06,
      "loss": 0.4104,
      "step": 25121
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5332354307174683,
      "learning_rate": 2.2267888754956748e-06,
      "loss": 0.3889,
      "step": 25122
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4866938591003418,
      "learning_rate": 2.2261482296628034e-06,
      "loss": 0.3813,
      "step": 25123
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.45365571975708,
      "learning_rate": 2.2255076644572106e-06,
      "loss": 0.3612,
      "step": 25124
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5078891515731812,
      "learning_rate": 2.2248671798855316e-06,
      "loss": 0.4122,
      "step": 25125
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5261441469192505,
      "learning_rate": 2.2242267759544156e-06,
      "loss": 0.3809,
      "step": 25126
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4677667617797852,
      "learning_rate": 2.2235864526704996e-06,
      "loss": 0.3886,
      "step": 25127
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5341529846191406,
      "learning_rate": 2.2229462100404277e-06,
      "loss": 0.3949,
      "step": 25128
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9461770057678223,
      "learning_rate": 2.2223060480708414e-06,
      "loss": 0.3808,
      "step": 25129
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.505373477935791,
      "learning_rate": 2.2216659667683794e-06,
      "loss": 0.3602,
      "step": 25130
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.512168526649475,
      "learning_rate": 2.221025966139676e-06,
      "loss": 0.4091,
      "step": 25131
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5085902214050293,
      "learning_rate": 2.220386046191374e-06,
      "loss": 0.3788,
      "step": 25132
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5825846195220947,
      "learning_rate": 2.219746206930108e-06,
      "loss": 0.4437,
      "step": 25133
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4509283304214478,
      "learning_rate": 2.21910644836252e-06,
      "loss": 0.3746,
      "step": 25134
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.622987985610962,
      "learning_rate": 2.2184667704952356e-06,
      "loss": 0.4421,
      "step": 25135
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5389333963394165,
      "learning_rate": 2.2178271733348943e-06,
      "loss": 0.3608,
      "step": 25136
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6252449750900269,
      "learning_rate": 2.217187656888129e-06,
      "loss": 0.4444,
      "step": 25137
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5282689332962036,
      "learning_rate": 2.2165482211615764e-06,
      "loss": 0.3859,
      "step": 25138
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.554234266281128,
      "learning_rate": 2.215908866161864e-06,
      "loss": 0.4403,
      "step": 25139
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4934704303741455,
      "learning_rate": 2.2152695918956223e-06,
      "loss": 0.3765,
      "step": 25140
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6082817316055298,
      "learning_rate": 2.214630398369483e-06,
      "loss": 0.4138,
      "step": 25141
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5671446323394775,
      "learning_rate": 2.213991285590079e-06,
      "loss": 0.3936,
      "step": 25142
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6665501594543457,
      "learning_rate": 2.2133522535640326e-06,
      "loss": 0.4551,
      "step": 25143
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5988513231277466,
      "learning_rate": 2.2127133022979775e-06,
      "loss": 0.377,
      "step": 25144
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5507628917694092,
      "learning_rate": 2.2120744317985353e-06,
      "loss": 0.4648,
      "step": 25145
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4662426710128784,
      "learning_rate": 2.2114356420723383e-06,
      "loss": 0.372,
      "step": 25146
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5243726968765259,
      "learning_rate": 2.210796933126005e-06,
      "loss": 0.3782,
      "step": 25147
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5199670791625977,
      "learning_rate": 2.210158304966162e-06,
      "loss": 0.3707,
      "step": 25148
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5094305276870728,
      "learning_rate": 2.2095197575994386e-06,
      "loss": 0.4638,
      "step": 25149
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5045222043991089,
      "learning_rate": 2.208881291032452e-06,
      "loss": 0.3631,
      "step": 25150
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5501649379730225,
      "learning_rate": 2.2082429052718226e-06,
      "loss": 0.4147,
      "step": 25151
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5989233255386353,
      "learning_rate": 2.2076046003241726e-06,
      "loss": 0.4086,
      "step": 25152
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5625786781311035,
      "learning_rate": 2.2069663761961248e-06,
      "loss": 0.4425,
      "step": 25153
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4527437686920166,
      "learning_rate": 2.2063282328943002e-06,
      "loss": 0.3525,
      "step": 25154
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5924350023269653,
      "learning_rate": 2.2056901704253143e-06,
      "loss": 0.4638,
      "step": 25155
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5168893337249756,
      "learning_rate": 2.205052188795782e-06,
      "loss": 0.3749,
      "step": 25156
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.514062762260437,
      "learning_rate": 2.2044142880123232e-06,
      "loss": 0.4331,
      "step": 25157
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4828147888183594,
      "learning_rate": 2.2037764680815576e-06,
      "loss": 0.3663,
      "step": 25158
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.639505386352539,
      "learning_rate": 2.2031387290100936e-06,
      "loss": 0.4999,
      "step": 25159
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6495630741119385,
      "learning_rate": 2.2025010708045512e-06,
      "loss": 0.4245,
      "step": 25160
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4964390993118286,
      "learning_rate": 2.2018634934715387e-06,
      "loss": 0.4153,
      "step": 25161
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5692658424377441,
      "learning_rate": 2.2012259970176754e-06,
      "loss": 0.3759,
      "step": 25162
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4971357583999634,
      "learning_rate": 2.2005885814495654e-06,
      "loss": 0.4089,
      "step": 25163
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5037635564804077,
      "learning_rate": 2.199951246773825e-06,
      "loss": 0.3725,
      "step": 25164
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.552837610244751,
      "learning_rate": 2.1993139929970643e-06,
      "loss": 0.3985,
      "step": 25165
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.502881646156311,
      "learning_rate": 2.1986768201258924e-06,
      "loss": 0.3694,
      "step": 25166
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6426975727081299,
      "learning_rate": 2.1980397281669143e-06,
      "loss": 0.4346,
      "step": 25167
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5817835330963135,
      "learning_rate": 2.1974027171267398e-06,
      "loss": 0.3732,
      "step": 25168
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.52914559841156,
      "learning_rate": 2.1967657870119753e-06,
      "loss": 0.4162,
      "step": 25169
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5222958326339722,
      "learning_rate": 2.1961289378292337e-06,
      "loss": 0.3877,
      "step": 25170
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5837996006011963,
      "learning_rate": 2.1954921695851095e-06,
      "loss": 0.3998,
      "step": 25171
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5823314189910889,
      "learning_rate": 2.19485548228621e-06,
      "loss": 0.3912,
      "step": 25172
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6018141508102417,
      "learning_rate": 2.1942188759391415e-06,
      "loss": 0.4265,
      "step": 25173
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.559478998184204,
      "learning_rate": 2.1935823505505072e-06,
      "loss": 0.3736,
      "step": 25174
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8148103952407837,
      "learning_rate": 2.1929459061269076e-06,
      "loss": 0.9811,
      "step": 25175
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5202065706253052,
      "learning_rate": 2.19230954267494e-06,
      "loss": 0.9511,
      "step": 25176
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.574891448020935,
      "learning_rate": 2.191673260201208e-06,
      "loss": 0.4444,
      "step": 25177
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4848319292068481,
      "learning_rate": 2.191037058712313e-06,
      "loss": 0.3843,
      "step": 25178
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5532305240631104,
      "learning_rate": 2.1904009382148474e-06,
      "loss": 0.4085,
      "step": 25179
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5196142196655273,
      "learning_rate": 2.189764898715415e-06,
      "loss": 0.3605,
      "step": 25180
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7695701122283936,
      "learning_rate": 2.1891289402206083e-06,
      "loss": 0.907,
      "step": 25181
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4256515502929688,
      "learning_rate": 2.188493062737027e-06,
      "loss": 1.0131,
      "step": 25182
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6210919618606567,
      "learning_rate": 2.1878572662712606e-06,
      "loss": 0.414,
      "step": 25183
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5118781328201294,
      "learning_rate": 2.187221550829907e-06,
      "loss": 0.3638,
      "step": 25184
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5719660520553589,
      "learning_rate": 2.1865859164195614e-06,
      "loss": 0.4194,
      "step": 25185
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6373753547668457,
      "learning_rate": 2.185950363046815e-06,
      "loss": 0.408,
      "step": 25186
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7751071453094482,
      "learning_rate": 2.1853148907182552e-06,
      "loss": 0.4691,
      "step": 25187
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4978410005569458,
      "learning_rate": 2.1846794994404765e-06,
      "loss": 0.3864,
      "step": 25188
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.599502682685852,
      "learning_rate": 2.184044189220069e-06,
      "loss": 0.4054,
      "step": 25189
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4922033548355103,
      "learning_rate": 2.1834089600636245e-06,
      "loss": 0.3758,
      "step": 25190
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.531930685043335,
      "learning_rate": 2.1827738119777277e-06,
      "loss": 0.4081,
      "step": 25191
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.495384931564331,
      "learning_rate": 2.182138744968965e-06,
      "loss": 0.394,
      "step": 25192
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5110403299331665,
      "learning_rate": 2.1815037590439237e-06,
      "loss": 0.4351,
      "step": 25193
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5066057443618774,
      "learning_rate": 2.180868854209194e-06,
      "loss": 0.3578,
      "step": 25194
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.603763222694397,
      "learning_rate": 2.1802340304713555e-06,
      "loss": 0.4272,
      "step": 25195
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4742112159729004,
      "learning_rate": 2.1795992878369964e-06,
      "loss": 0.3561,
      "step": 25196
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5826178789138794,
      "learning_rate": 2.1789646263126962e-06,
      "loss": 0.4565,
      "step": 25197
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.524794340133667,
      "learning_rate": 2.1783300459050416e-06,
      "loss": 0.3674,
      "step": 25198
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5042988061904907,
      "learning_rate": 2.1776955466206097e-06,
      "loss": 0.3807,
      "step": 25199
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.510753870010376,
      "learning_rate": 2.177061128465983e-06,
      "loss": 0.3852,
      "step": 25200
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5010175704956055,
      "learning_rate": 2.1764267914477453e-06,
      "loss": 0.4058,
      "step": 25201
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4920687675476074,
      "learning_rate": 2.175792535572472e-06,
      "loss": 0.3858,
      "step": 25202
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5983762741088867,
      "learning_rate": 2.1751583608467387e-06,
      "loss": 0.3993,
      "step": 25203
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4991159439086914,
      "learning_rate": 2.1745242672771248e-06,
      "loss": 0.3853,
      "step": 25204
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8951176404953003,
      "learning_rate": 2.173890254870209e-06,
      "loss": 0.4341,
      "step": 25205
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5244063138961792,
      "learning_rate": 2.173256323632569e-06,
      "loss": 0.3853,
      "step": 25206
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8346308469772339,
      "learning_rate": 2.1726224735707755e-06,
      "loss": 1.0078,
      "step": 25207
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5101475715637207,
      "learning_rate": 2.171988704691401e-06,
      "loss": 0.9744,
      "step": 25208
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4961509704589844,
      "learning_rate": 2.171355017001021e-06,
      "loss": 0.4344,
      "step": 25209
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4505163431167603,
      "learning_rate": 2.1707214105062125e-06,
      "loss": 0.3546,
      "step": 25210
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.647261381149292,
      "learning_rate": 2.170087885213541e-06,
      "loss": 0.4703,
      "step": 25211
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4945193529129028,
      "learning_rate": 2.169454441129576e-06,
      "loss": 0.369,
      "step": 25212
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5794200897216797,
      "learning_rate": 2.168821078260891e-06,
      "loss": 0.438,
      "step": 25213
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5504690408706665,
      "learning_rate": 2.168187796614056e-06,
      "loss": 0.3733,
      "step": 25214
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4959237575531006,
      "learning_rate": 2.1675545961956357e-06,
      "loss": 0.4254,
      "step": 25215
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4771537780761719,
      "learning_rate": 2.1669214770122005e-06,
      "loss": 0.364,
      "step": 25216
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6595779657363892,
      "learning_rate": 2.166288439070313e-06,
      "loss": 0.4492,
      "step": 25217
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5735580921173096,
      "learning_rate": 2.165655482376544e-06,
      "loss": 0.4048,
      "step": 25218
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.777130365371704,
      "learning_rate": 2.1650226069374525e-06,
      "loss": 0.4494,
      "step": 25219
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5961042642593384,
      "learning_rate": 2.164389812759605e-06,
      "loss": 0.3748,
      "step": 25220
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5747032165527344,
      "learning_rate": 2.1637570998495684e-06,
      "loss": 0.4169,
      "step": 25221
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.676159381866455,
      "learning_rate": 2.163124468213902e-06,
      "loss": 0.4056,
      "step": 25222
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5558174848556519,
      "learning_rate": 2.162491917859163e-06,
      "loss": 0.42,
      "step": 25223
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.495963454246521,
      "learning_rate": 2.1618594487919155e-06,
      "loss": 0.3791,
      "step": 25224
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5513615608215332,
      "learning_rate": 2.161227061018719e-06,
      "loss": 0.3891,
      "step": 25225
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5809944868087769,
      "learning_rate": 2.160594754546136e-06,
      "loss": 0.3931,
      "step": 25226
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5491300821304321,
      "learning_rate": 2.1599625293807214e-06,
      "loss": 0.415,
      "step": 25227
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5311895608901978,
      "learning_rate": 2.1593303855290295e-06,
      "loss": 0.3733,
      "step": 25228
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.532620906829834,
      "learning_rate": 2.1586983229976187e-06,
      "loss": 0.43,
      "step": 25229
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4759896993637085,
      "learning_rate": 2.158066341793048e-06,
      "loss": 0.3761,
      "step": 25230
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6266252994537354,
      "learning_rate": 2.1574344419218672e-06,
      "loss": 0.431,
      "step": 25231
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5320425033569336,
      "learning_rate": 2.156802623390635e-06,
      "loss": 0.3784,
      "step": 25232
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6059714555740356,
      "learning_rate": 2.1561708862058983e-06,
      "loss": 0.4397,
      "step": 25233
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5873016119003296,
      "learning_rate": 2.155539230374215e-06,
      "loss": 0.3847,
      "step": 25234
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5694470405578613,
      "learning_rate": 2.154907655902131e-06,
      "loss": 0.4527,
      "step": 25235
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4779304265975952,
      "learning_rate": 2.1542761627961995e-06,
      "loss": 0.3643,
      "step": 25236
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5658397674560547,
      "learning_rate": 2.1536447510629733e-06,
      "loss": 0.3998,
      "step": 25237
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5675837993621826,
      "learning_rate": 2.1530134207089983e-06,
      "loss": 0.3651,
      "step": 25238
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6190446615219116,
      "learning_rate": 2.152382171740819e-06,
      "loss": 0.4057,
      "step": 25239
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5852906703948975,
      "learning_rate": 2.1517510041649847e-06,
      "loss": 0.3758,
      "step": 25240
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5661959648132324,
      "learning_rate": 2.1511199179880438e-06,
      "loss": 0.4375,
      "step": 25241
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4882302284240723,
      "learning_rate": 2.1504889132165418e-06,
      "loss": 0.3683,
      "step": 25242
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.327636957168579,
      "learning_rate": 2.1498579898570228e-06,
      "loss": 0.5074,
      "step": 25243
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4682799577713013,
      "learning_rate": 2.1492271479160263e-06,
      "loss": 0.3512,
      "step": 25244
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5234549045562744,
      "learning_rate": 2.1485963874000983e-06,
      "loss": 0.4382,
      "step": 25245
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.60704505443573,
      "learning_rate": 2.1479657083157833e-06,
      "loss": 0.3803,
      "step": 25246
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4894124269485474,
      "learning_rate": 2.1473351106696205e-06,
      "loss": 0.3932,
      "step": 25247
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5416090488433838,
      "learning_rate": 2.1467045944681463e-06,
      "loss": 0.3663,
      "step": 25248
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5059324502944946,
      "learning_rate": 2.1460741597179045e-06,
      "loss": 0.4093,
      "step": 25249
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.581142783164978,
      "learning_rate": 2.1454438064254347e-06,
      "loss": 0.4005,
      "step": 25250
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5597425699234009,
      "learning_rate": 2.14481353459727e-06,
      "loss": 0.4421,
      "step": 25251
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5455706119537354,
      "learning_rate": 2.144183344239953e-06,
      "loss": 0.4095,
      "step": 25252
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5574055910110474,
      "learning_rate": 2.143553235360014e-06,
      "loss": 0.4502,
      "step": 25253
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5160259008407593,
      "learning_rate": 2.142923207963994e-06,
      "loss": 0.3766,
      "step": 25254
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5949532985687256,
      "learning_rate": 2.1422932620584213e-06,
      "loss": 0.4228,
      "step": 25255
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.531243920326233,
      "learning_rate": 2.1416633976498325e-06,
      "loss": 0.3991,
      "step": 25256
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5543478727340698,
      "learning_rate": 2.141033614744763e-06,
      "loss": 0.4564,
      "step": 25257
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.529421091079712,
      "learning_rate": 2.1404039133497424e-06,
      "loss": 0.3903,
      "step": 25258
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6294872760772705,
      "learning_rate": 2.1397742934712985e-06,
      "loss": 0.3985,
      "step": 25259
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4819605350494385,
      "learning_rate": 2.139144755115965e-06,
      "loss": 0.3638,
      "step": 25260
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.562494158744812,
      "learning_rate": 2.1385152982902703e-06,
      "loss": 0.4745,
      "step": 25261
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6107732057571411,
      "learning_rate": 2.1378859230007456e-06,
      "loss": 0.4146,
      "step": 25262
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5506291389465332,
      "learning_rate": 2.1372566292539155e-06,
      "loss": 0.4041,
      "step": 25263
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.483809232711792,
      "learning_rate": 2.136627417056305e-06,
      "loss": 0.37,
      "step": 25264
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6231623888015747,
      "learning_rate": 2.1359982864144425e-06,
      "loss": 0.4227,
      "step": 25265
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6007699966430664,
      "learning_rate": 2.1353692373348554e-06,
      "loss": 0.4107,
      "step": 25266
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.447909951210022,
      "learning_rate": 2.1347402698240626e-06,
      "loss": 0.3728,
      "step": 25267
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5794306993484497,
      "learning_rate": 2.134111383888593e-06,
      "loss": 0.3996,
      "step": 25268
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6250793933868408,
      "learning_rate": 2.1334825795349644e-06,
      "loss": 0.4069,
      "step": 25269
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.557732105255127,
      "learning_rate": 2.1328538567697033e-06,
      "loss": 0.3791,
      "step": 25270
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.516847014427185,
      "learning_rate": 2.132225215599325e-06,
      "loss": 0.4248,
      "step": 25271
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5350068807601929,
      "learning_rate": 2.1315966560303524e-06,
      "loss": 0.3657,
      "step": 25272
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6370295286178589,
      "learning_rate": 2.130968178069308e-06,
      "loss": 0.4012,
      "step": 25273
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5383751392364502,
      "learning_rate": 2.130339781722706e-06,
      "loss": 0.3687,
      "step": 25274
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5016827583312988,
      "learning_rate": 2.129711466997062e-06,
      "loss": 0.4231,
      "step": 25275
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5193819999694824,
      "learning_rate": 2.1290832338988965e-06,
      "loss": 0.3781,
      "step": 25276
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5032767057418823,
      "learning_rate": 2.1284550824347227e-06,
      "loss": 0.4313,
      "step": 25277
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4525271654129028,
      "learning_rate": 2.12782701261106e-06,
      "loss": 0.3747,
      "step": 25278
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5448331832885742,
      "learning_rate": 2.12719902443442e-06,
      "loss": 0.3935,
      "step": 25279
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6115700006484985,
      "learning_rate": 2.1265711179113124e-06,
      "loss": 0.3541,
      "step": 25280
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.490712285041809,
      "learning_rate": 2.1259432930482517e-06,
      "loss": 0.4065,
      "step": 25281
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4585676193237305,
      "learning_rate": 2.125315549851754e-06,
      "loss": 0.3683,
      "step": 25282
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.515962839126587,
      "learning_rate": 2.124687888328323e-06,
      "loss": 0.377,
      "step": 25283
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5723752975463867,
      "learning_rate": 2.1240603084844746e-06,
      "loss": 0.4002,
      "step": 25284
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.708008050918579,
      "learning_rate": 2.1234328103267134e-06,
      "loss": 0.4315,
      "step": 25285
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.568125605583191,
      "learning_rate": 2.1228053938615513e-06,
      "loss": 0.3839,
      "step": 25286
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5984407663345337,
      "learning_rate": 2.1221780590954897e-06,
      "loss": 0.4449,
      "step": 25287
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6039891242980957,
      "learning_rate": 2.1215508060350432e-06,
      "loss": 0.4256,
      "step": 25288
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.039557456970215,
      "learning_rate": 2.1209236346867102e-06,
      "loss": 0.4233,
      "step": 25289
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.520314335823059,
      "learning_rate": 2.120296545057e-06,
      "loss": 0.3634,
      "step": 25290
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9170531034469604,
      "learning_rate": 2.1196695371524136e-06,
      "loss": 0.4578,
      "step": 25291
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5496203899383545,
      "learning_rate": 2.119042610979455e-06,
      "loss": 0.4036,
      "step": 25292
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5258033275604248,
      "learning_rate": 2.118415766544629e-06,
      "loss": 0.4117,
      "step": 25293
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5950483083724976,
      "learning_rate": 2.1177890038544346e-06,
      "loss": 0.4012,
      "step": 25294
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6974376440048218,
      "learning_rate": 2.1171623229153704e-06,
      "loss": 0.9226,
      "step": 25295
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.407949686050415,
      "learning_rate": 2.116535723733938e-06,
      "loss": 0.9807,
      "step": 25296
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5409425497055054,
      "learning_rate": 2.1159092063166362e-06,
      "loss": 0.4443,
      "step": 25297
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5150916576385498,
      "learning_rate": 2.1152827706699665e-06,
      "loss": 0.372,
      "step": 25298
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5520851612091064,
      "learning_rate": 2.1146564168004226e-06,
      "loss": 0.4053,
      "step": 25299
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5000135898590088,
      "learning_rate": 2.114030144714498e-06,
      "loss": 0.3844,
      "step": 25300
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6121726036071777,
      "learning_rate": 2.1134039544186915e-06,
      "loss": 0.453,
      "step": 25301
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4879205226898193,
      "learning_rate": 2.112777845919497e-06,
      "loss": 0.3741,
      "step": 25302
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6789700984954834,
      "learning_rate": 2.1121518192234114e-06,
      "loss": 0.4029,
      "step": 25303
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5245498418807983,
      "learning_rate": 2.111525874336925e-06,
      "loss": 0.3998,
      "step": 25304
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5517717599868774,
      "learning_rate": 2.110900011266526e-06,
      "loss": 0.4184,
      "step": 25305
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6356133222579956,
      "learning_rate": 2.1102742300187095e-06,
      "loss": 0.4032,
      "step": 25306
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5786453485488892,
      "learning_rate": 2.109648530599968e-06,
      "loss": 0.4243,
      "step": 25307
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5985397100448608,
      "learning_rate": 2.1090229130167874e-06,
      "loss": 0.3883,
      "step": 25308
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4784772396087646,
      "learning_rate": 2.108397377275658e-06,
      "loss": 0.3956,
      "step": 25309
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.496091604232788,
      "learning_rate": 2.1077719233830663e-06,
      "loss": 0.3483,
      "step": 25310
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.52187180519104,
      "learning_rate": 2.1071465513455023e-06,
      "loss": 0.406,
      "step": 25311
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.474255084991455,
      "learning_rate": 2.1065212611694464e-06,
      "loss": 0.3587,
      "step": 25312
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6095434427261353,
      "learning_rate": 2.105896052861388e-06,
      "loss": 0.4465,
      "step": 25313
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.558384656906128,
      "learning_rate": 2.1052709264278137e-06,
      "loss": 0.3869,
      "step": 25314
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6346968412399292,
      "learning_rate": 2.1046458818752036e-06,
      "loss": 0.4084,
      "step": 25315
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5644479990005493,
      "learning_rate": 2.104020919210038e-06,
      "loss": 0.3874,
      "step": 25316
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.530652642250061,
      "learning_rate": 2.103396038438803e-06,
      "loss": 0.3931,
      "step": 25317
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.567665934562683,
      "learning_rate": 2.102771239567978e-06,
      "loss": 0.3829,
      "step": 25318
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7221736907958984,
      "learning_rate": 2.102146522604046e-06,
      "loss": 0.4584,
      "step": 25319
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5174936056137085,
      "learning_rate": 2.101521887553484e-06,
      "loss": 0.3772,
      "step": 25320
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.580989122390747,
      "learning_rate": 2.1008973344227668e-06,
      "loss": 0.4204,
      "step": 25321
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5236315727233887,
      "learning_rate": 2.1002728632183765e-06,
      "loss": 0.4026,
      "step": 25322
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.556127667427063,
      "learning_rate": 2.0996484739467916e-06,
      "loss": 0.4269,
      "step": 25323
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.586608648300171,
      "learning_rate": 2.0990241666144817e-06,
      "loss": 0.374,
      "step": 25324
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5221235752105713,
      "learning_rate": 2.0983999412279287e-06,
      "loss": 0.4532,
      "step": 25325
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5385960340499878,
      "learning_rate": 2.097775797793602e-06,
      "loss": 0.3882,
      "step": 25326
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4855690002441406,
      "learning_rate": 2.097151736317978e-06,
      "loss": 0.4113,
      "step": 25327
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5907219648361206,
      "learning_rate": 2.096527756807526e-06,
      "loss": 0.3544,
      "step": 25328
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.567899465560913,
      "learning_rate": 2.095903859268721e-06,
      "loss": 0.4268,
      "step": 25329
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.598862648010254,
      "learning_rate": 2.09528004370803e-06,
      "loss": 0.374,
      "step": 25330
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6069527864456177,
      "learning_rate": 2.0946563101319285e-06,
      "loss": 0.4558,
      "step": 25331
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.494928002357483,
      "learning_rate": 2.0940326585468795e-06,
      "loss": 0.3501,
      "step": 25332
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6397027969360352,
      "learning_rate": 2.093409088959354e-06,
      "loss": 0.4331,
      "step": 25333
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5942790508270264,
      "learning_rate": 2.092785601375822e-06,
      "loss": 0.3987,
      "step": 25334
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5650445222854614,
      "learning_rate": 2.0921621958027473e-06,
      "loss": 0.4114,
      "step": 25335
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6219875812530518,
      "learning_rate": 2.0915388722465935e-06,
      "loss": 0.401,
      "step": 25336
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5257303714752197,
      "learning_rate": 2.090915630713828e-06,
      "loss": 0.4001,
      "step": 25337
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5730235576629639,
      "learning_rate": 2.090292471210915e-06,
      "loss": 0.3617,
      "step": 25338
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.519804835319519,
      "learning_rate": 2.089669393744319e-06,
      "loss": 0.3898,
      "step": 25339
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5577372312545776,
      "learning_rate": 2.089046398320501e-06,
      "loss": 0.3726,
      "step": 25340
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.569036602973938,
      "learning_rate": 2.0884234849459205e-06,
      "loss": 0.4315,
      "step": 25341
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4766405820846558,
      "learning_rate": 2.087800653627039e-06,
      "loss": 0.3617,
      "step": 25342
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6383012533187866,
      "learning_rate": 2.08717790437032e-06,
      "loss": 0.4451,
      "step": 25343
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4372711181640625,
      "learning_rate": 2.086555237182216e-06,
      "loss": 0.3617,
      "step": 25344
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5311545133590698,
      "learning_rate": 2.0859326520691912e-06,
      "loss": 0.4667,
      "step": 25345
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.56596040725708,
      "learning_rate": 2.085310149037697e-06,
      "loss": 0.3546,
      "step": 25346
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6639803647994995,
      "learning_rate": 2.084687728094197e-06,
      "loss": 0.4608,
      "step": 25347
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4847601652145386,
      "learning_rate": 2.084065389245138e-06,
      "loss": 0.3328,
      "step": 25348
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5036537647247314,
      "learning_rate": 2.0834431324969805e-06,
      "loss": 0.4315,
      "step": 25349
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8165314197540283,
      "learning_rate": 2.0828209578561787e-06,
      "loss": 0.3995,
      "step": 25350
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.663266658782959,
      "learning_rate": 2.0821988653291847e-06,
      "loss": 0.3881,
      "step": 25351
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5048094987869263,
      "learning_rate": 2.0815768549224458e-06,
      "loss": 0.3544,
      "step": 25352
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5029515027999878,
      "learning_rate": 2.080954926642418e-06,
      "loss": 0.4051,
      "step": 25353
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.522262454032898,
      "learning_rate": 2.0803330804955502e-06,
      "loss": 0.3809,
      "step": 25354
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.558011770248413,
      "learning_rate": 2.0797113164882964e-06,
      "loss": 0.3885,
      "step": 25355
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4497230052947998,
      "learning_rate": 2.0790896346271004e-06,
      "loss": 0.3762,
      "step": 25356
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.574097990989685,
      "learning_rate": 2.078468034918408e-06,
      "loss": 0.439,
      "step": 25357
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5313868522644043,
      "learning_rate": 2.07784651736867e-06,
      "loss": 0.3756,
      "step": 25358
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5532824993133545,
      "learning_rate": 2.077225081984334e-06,
      "loss": 0.4214,
      "step": 25359
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5323092937469482,
      "learning_rate": 2.076603728771841e-06,
      "loss": 0.3884,
      "step": 25360
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5276261568069458,
      "learning_rate": 2.07598245773764e-06,
      "loss": 0.4389,
      "step": 25361
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6064375638961792,
      "learning_rate": 2.0753612688881697e-06,
      "loss": 0.4064,
      "step": 25362
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5302067995071411,
      "learning_rate": 2.0747401622298767e-06,
      "loss": 0.4504,
      "step": 25363
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.681761384010315,
      "learning_rate": 2.0741191377691996e-06,
      "loss": 0.3708,
      "step": 25364
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5078774690628052,
      "learning_rate": 2.0734981955125833e-06,
      "loss": 0.4154,
      "step": 25365
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6037873029708862,
      "learning_rate": 2.0728773354664646e-06,
      "loss": 0.4132,
      "step": 25366
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4710936546325684,
      "learning_rate": 2.072256557637284e-06,
      "loss": 0.4157,
      "step": 25367
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4750816822052002,
      "learning_rate": 2.0716358620314805e-06,
      "loss": 0.3696,
      "step": 25368
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.526639699935913,
      "learning_rate": 2.071015248655489e-06,
      "loss": 0.4257,
      "step": 25369
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5338295698165894,
      "learning_rate": 2.070394717515751e-06,
      "loss": 0.3558,
      "step": 25370
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5733367204666138,
      "learning_rate": 2.0697742686187016e-06,
      "loss": 0.3953,
      "step": 25371
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4778625965118408,
      "learning_rate": 2.069153901970771e-06,
      "loss": 0.368,
      "step": 25372
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5434679985046387,
      "learning_rate": 2.0685336175783952e-06,
      "loss": 0.3984,
      "step": 25373
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.520141363143921,
      "learning_rate": 2.0679134154480095e-06,
      "loss": 0.3838,
      "step": 25374
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.546406626701355,
      "learning_rate": 2.067293295586048e-06,
      "loss": 0.4303,
      "step": 25375
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5535598993301392,
      "learning_rate": 2.0666732579989414e-06,
      "loss": 0.3767,
      "step": 25376
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5632426738739014,
      "learning_rate": 2.066053302693115e-06,
      "loss": 0.4314,
      "step": 25377
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5821810960769653,
      "learning_rate": 2.0654334296750033e-06,
      "loss": 0.4186,
      "step": 25378
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5515459775924683,
      "learning_rate": 2.0648136389510387e-06,
      "loss": 0.4547,
      "step": 25379
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.501089334487915,
      "learning_rate": 2.0641939305276417e-06,
      "loss": 0.3526,
      "step": 25380
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4791802167892456,
      "learning_rate": 2.0635743044112468e-06,
      "loss": 0.3829,
      "step": 25381
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5482572317123413,
      "learning_rate": 2.0629547606082745e-06,
      "loss": 0.3659,
      "step": 25382
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5100897550582886,
      "learning_rate": 2.0623352991251567e-06,
      "loss": 0.4376,
      "step": 25383
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5211844444274902,
      "learning_rate": 2.061715919968311e-06,
      "loss": 0.3848,
      "step": 25384
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7495696544647217,
      "learning_rate": 2.061096623144165e-06,
      "loss": 1.0628,
      "step": 25385
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4715059995651245,
      "learning_rate": 2.060477408659146e-06,
      "loss": 1.0399,
      "step": 25386
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5279364585876465,
      "learning_rate": 2.059858276519672e-06,
      "loss": 0.4475,
      "step": 25387
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.102862596511841,
      "learning_rate": 2.0592392267321616e-06,
      "loss": 0.4048,
      "step": 25388
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4975824356079102,
      "learning_rate": 2.058620259303038e-06,
      "loss": 0.3932,
      "step": 25389
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5039604902267456,
      "learning_rate": 2.0580013742387216e-06,
      "loss": 0.3637,
      "step": 25390
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5509843826293945,
      "learning_rate": 2.057382571545633e-06,
      "loss": 0.4067,
      "step": 25391
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5865942239761353,
      "learning_rate": 2.056763851230189e-06,
      "loss": 0.4139,
      "step": 25392
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6560359001159668,
      "learning_rate": 2.0561452132988024e-06,
      "loss": 0.4199,
      "step": 25393
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5783063173294067,
      "learning_rate": 2.0555266577578926e-06,
      "loss": 0.3844,
      "step": 25394
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6185449361801147,
      "learning_rate": 2.0549081846138787e-06,
      "loss": 0.4103,
      "step": 25395
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.614833950996399,
      "learning_rate": 2.054289793873168e-06,
      "loss": 0.3521,
      "step": 25396
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5600287914276123,
      "learning_rate": 2.0536714855421813e-06,
      "loss": 0.4476,
      "step": 25397
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.473073124885559,
      "learning_rate": 2.0530532596273256e-06,
      "loss": 0.3411,
      "step": 25398
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7908838987350464,
      "learning_rate": 2.0524351161350174e-06,
      "loss": 0.4287,
      "step": 25399
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.585304856300354,
      "learning_rate": 2.0518170550716634e-06,
      "loss": 0.3717,
      "step": 25400
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6622055768966675,
      "learning_rate": 2.0511990764436775e-06,
      "loss": 0.4118,
      "step": 25401
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.45947265625,
      "learning_rate": 2.0505811802574695e-06,
      "loss": 0.3507,
      "step": 25402
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.565129041671753,
      "learning_rate": 2.049963366519446e-06,
      "loss": 0.4381,
      "step": 25403
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6333398818969727,
      "learning_rate": 2.049345635236013e-06,
      "loss": 0.4021,
      "step": 25404
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.509864330291748,
      "learning_rate": 2.04872798641358e-06,
      "loss": 0.4113,
      "step": 25405
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4896517992019653,
      "learning_rate": 2.0481104200585546e-06,
      "loss": 0.3593,
      "step": 25406
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6759806871414185,
      "learning_rate": 2.04749293617734e-06,
      "loss": 0.4192,
      "step": 25407
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5262482166290283,
      "learning_rate": 2.046875534776338e-06,
      "loss": 0.3772,
      "step": 25408
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6164617538452148,
      "learning_rate": 2.046258215861954e-06,
      "loss": 0.4543,
      "step": 25409
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5526809692382812,
      "learning_rate": 2.0456409794405906e-06,
      "loss": 0.3899,
      "step": 25410
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.498870849609375,
      "learning_rate": 2.045023825518653e-06,
      "loss": 0.4297,
      "step": 25411
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4424713850021362,
      "learning_rate": 2.0444067541025393e-06,
      "loss": 0.3677,
      "step": 25412
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5475722551345825,
      "learning_rate": 2.043789765198646e-06,
      "loss": 0.4277,
      "step": 25413
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.558678150177002,
      "learning_rate": 2.043172858813376e-06,
      "loss": 0.402,
      "step": 25414
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5355994701385498,
      "learning_rate": 2.0425560349531294e-06,
      "loss": 0.4117,
      "step": 25415
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5893964767456055,
      "learning_rate": 2.041939293624299e-06,
      "loss": 0.369,
      "step": 25416
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4985430240631104,
      "learning_rate": 2.041322634833286e-06,
      "loss": 0.3893,
      "step": 25417
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5417958498001099,
      "learning_rate": 2.0407060585864813e-06,
      "loss": 0.3854,
      "step": 25418
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5544731616973877,
      "learning_rate": 2.040089564890284e-06,
      "loss": 0.4417,
      "step": 25419
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5586111545562744,
      "learning_rate": 2.039473153751085e-06,
      "loss": 0.3756,
      "step": 25420
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5979219675064087,
      "learning_rate": 2.0388568251752795e-06,
      "loss": 0.4252,
      "step": 25421
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.512091040611267,
      "learning_rate": 2.0382405791692606e-06,
      "loss": 0.3581,
      "step": 25422
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.770568609237671,
      "learning_rate": 2.0376244157394186e-06,
      "loss": 0.9445,
      "step": 25423
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5299420356750488,
      "learning_rate": 2.037008334892142e-06,
      "loss": 0.9825,
      "step": 25424
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.779952049255371,
      "learning_rate": 2.036392336633821e-06,
      "loss": 0.4133,
      "step": 25425
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6551203727722168,
      "learning_rate": 2.0357764209708476e-06,
      "loss": 0.4272,
      "step": 25426
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7170054912567139,
      "learning_rate": 2.0351605879096103e-06,
      "loss": 0.9639,
      "step": 25427
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.378520131111145,
      "learning_rate": 2.0345448374564937e-06,
      "loss": 0.9662,
      "step": 25428
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5100678205490112,
      "learning_rate": 2.033929169617882e-06,
      "loss": 0.4207,
      "step": 25429
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.549955129623413,
      "learning_rate": 2.0333135844001627e-06,
      "loss": 0.3846,
      "step": 25430
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7315911054611206,
      "learning_rate": 2.0326980818097243e-06,
      "loss": 0.4359,
      "step": 25431
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.519053339958191,
      "learning_rate": 2.0320826618529445e-06,
      "loss": 0.378,
      "step": 25432
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5028395652770996,
      "learning_rate": 2.0314673245362117e-06,
      "loss": 0.3717,
      "step": 25433
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5537118911743164,
      "learning_rate": 2.030852069865902e-06,
      "loss": 0.3848,
      "step": 25434
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5796829462051392,
      "learning_rate": 2.030236897848402e-06,
      "loss": 0.4494,
      "step": 25435
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5111503601074219,
      "learning_rate": 2.0296218084900877e-06,
      "loss": 0.4114,
      "step": 25436
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5312758684158325,
      "learning_rate": 2.0290068017973396e-06,
      "loss": 0.4521,
      "step": 25437
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.557722806930542,
      "learning_rate": 2.028391877776541e-06,
      "loss": 0.4049,
      "step": 25438
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5706839561462402,
      "learning_rate": 2.0277770364340654e-06,
      "loss": 0.3733,
      "step": 25439
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5104833841323853,
      "learning_rate": 2.0271622777762867e-06,
      "loss": 0.3764,
      "step": 25440
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4932950735092163,
      "learning_rate": 2.0265476018095854e-06,
      "loss": 0.3823,
      "step": 25441
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.583627700805664,
      "learning_rate": 2.025933008540335e-06,
      "loss": 0.4001,
      "step": 25442
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5087231397628784,
      "learning_rate": 2.0253184979749163e-06,
      "loss": 0.3962,
      "step": 25443
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5194028615951538,
      "learning_rate": 2.0247040701196917e-06,
      "loss": 0.3605,
      "step": 25444
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6258859634399414,
      "learning_rate": 2.024089724981039e-06,
      "loss": 0.4135,
      "step": 25445
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5155515670776367,
      "learning_rate": 2.0234754625653296e-06,
      "loss": 0.3615,
      "step": 25446
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5650204420089722,
      "learning_rate": 2.0228612828789387e-06,
      "loss": 0.4757,
      "step": 25447
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5452715158462524,
      "learning_rate": 2.022247185928231e-06,
      "loss": 0.4027,
      "step": 25448
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9151508808135986,
      "learning_rate": 2.021633171719575e-06,
      "loss": 0.9282,
      "step": 25449
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5040611028671265,
      "learning_rate": 2.02101924025934e-06,
      "loss": 1.0179,
      "step": 25450
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5727806091308594,
      "learning_rate": 2.020405391553898e-06,
      "loss": 0.4364,
      "step": 25451
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6843234300613403,
      "learning_rate": 2.0197916256096093e-06,
      "loss": 0.4119,
      "step": 25452
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.646596074104309,
      "learning_rate": 2.019177942432846e-06,
      "loss": 0.4389,
      "step": 25453
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.589704990386963,
      "learning_rate": 2.018564342029966e-06,
      "loss": 0.3841,
      "step": 25454
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5335742235183716,
      "learning_rate": 2.017950824407339e-06,
      "loss": 0.4015,
      "step": 25455
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5557726621627808,
      "learning_rate": 2.0173373895713234e-06,
      "loss": 0.3913,
      "step": 25456
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5683385133743286,
      "learning_rate": 2.0167240375282847e-06,
      "loss": 0.4188,
      "step": 25457
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5408915281295776,
      "learning_rate": 2.016110768284586e-06,
      "loss": 0.3761,
      "step": 25458
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5943940877914429,
      "learning_rate": 2.015497581846585e-06,
      "loss": 0.4388,
      "step": 25459
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5405995845794678,
      "learning_rate": 2.01488447822064e-06,
      "loss": 0.4,
      "step": 25460
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5044188499450684,
      "learning_rate": 2.0142714574131107e-06,
      "loss": 0.3974,
      "step": 25461
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6058999300003052,
      "learning_rate": 2.013658519430357e-06,
      "loss": 0.4051,
      "step": 25462
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6055980920791626,
      "learning_rate": 2.013045664278738e-06,
      "loss": 0.4505,
      "step": 25463
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.561295747756958,
      "learning_rate": 2.0124328919646075e-06,
      "loss": 0.3886,
      "step": 25464
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.595017433166504,
      "learning_rate": 2.011820202494318e-06,
      "loss": 0.4525,
      "step": 25465
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5643810033798218,
      "learning_rate": 2.011207595874226e-06,
      "loss": 0.3873,
      "step": 25466
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5820207595825195,
      "learning_rate": 2.010595072110689e-06,
      "loss": 0.4139,
      "step": 25467
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6942167282104492,
      "learning_rate": 2.0099826312100536e-06,
      "loss": 0.4051,
      "step": 25468
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5413202047348022,
      "learning_rate": 2.009370273178678e-06,
      "loss": 0.432,
      "step": 25469
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5930196046829224,
      "learning_rate": 2.008757998022909e-06,
      "loss": 0.3888,
      "step": 25470
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7623165845870972,
      "learning_rate": 2.008145805749099e-06,
      "loss": 0.4554,
      "step": 25471
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5230695009231567,
      "learning_rate": 2.007533696363595e-06,
      "loss": 0.3781,
      "step": 25472
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.539764404296875,
      "learning_rate": 2.0069216698727467e-06,
      "loss": 0.4221,
      "step": 25473
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5315219163894653,
      "learning_rate": 2.0063097262829044e-06,
      "loss": 0.3873,
      "step": 25474
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6207023859024048,
      "learning_rate": 2.0056978656004145e-06,
      "loss": 0.463,
      "step": 25475
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5903273820877075,
      "learning_rate": 2.005086087831617e-06,
      "loss": 0.3799,
      "step": 25476
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5766783952713013,
      "learning_rate": 2.0044743929828626e-06,
      "loss": 0.4263,
      "step": 25477
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5917295217514038,
      "learning_rate": 2.0038627810604925e-06,
      "loss": 0.3804,
      "step": 25478
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4606066942214966,
      "learning_rate": 2.003251252070859e-06,
      "loss": 0.4031,
      "step": 25479
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5086311101913452,
      "learning_rate": 2.002639806020291e-06,
      "loss": 0.3966,
      "step": 25480
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7356778383255005,
      "learning_rate": 2.002028442915137e-06,
      "loss": 0.4198,
      "step": 25481
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.475631833076477,
      "learning_rate": 2.0014171627617374e-06,
      "loss": 0.348,
      "step": 25482
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6558996438980103,
      "learning_rate": 2.0008059655664346e-06,
      "loss": 0.4687,
      "step": 25483
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5249090194702148,
      "learning_rate": 2.000194851335566e-06,
      "loss": 0.3606,
      "step": 25484
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4976369142532349,
      "learning_rate": 1.9995838200754656e-06,
      "loss": 0.3875,
      "step": 25485
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5494563579559326,
      "learning_rate": 1.998972871792475e-06,
      "loss": 0.3703,
      "step": 25486
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.565566897392273,
      "learning_rate": 1.998362006492932e-06,
      "loss": 0.4005,
      "step": 25487
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.464869737625122,
      "learning_rate": 1.9977512241831686e-06,
      "loss": 0.3726,
      "step": 25488
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5459054708480835,
      "learning_rate": 1.997140524869523e-06,
      "loss": 0.4413,
      "step": 25489
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5394128561019897,
      "learning_rate": 1.9965299085583247e-06,
      "loss": 0.3676,
      "step": 25490
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5427294969558716,
      "learning_rate": 1.9959193752559127e-06,
      "loss": 0.4692,
      "step": 25491
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5359596014022827,
      "learning_rate": 1.995308924968613e-06,
      "loss": 0.3644,
      "step": 25492
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5195964574813843,
      "learning_rate": 1.99469855770276e-06,
      "loss": 0.4831,
      "step": 25493
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.53513503074646,
      "learning_rate": 1.994088273464687e-06,
      "loss": 0.3901,
      "step": 25494
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6088262796401978,
      "learning_rate": 1.9934780722607207e-06,
      "loss": 0.4477,
      "step": 25495
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5355585813522339,
      "learning_rate": 1.9928679540971865e-06,
      "loss": 0.3902,
      "step": 25496
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.581174373626709,
      "learning_rate": 1.992257918980416e-06,
      "loss": 0.4311,
      "step": 25497
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5530414581298828,
      "learning_rate": 1.9916479669167356e-06,
      "loss": 0.376,
      "step": 25498
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5471165180206299,
      "learning_rate": 1.9910380979124757e-06,
      "loss": 0.4241,
      "step": 25499
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5057685375213623,
      "learning_rate": 1.9904283119739565e-06,
      "loss": 0.3677,
      "step": 25500
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.970416784286499,
      "learning_rate": 1.9898186091075012e-06,
      "loss": 0.4465,
      "step": 25501
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4527064561843872,
      "learning_rate": 1.9892089893194354e-06,
      "loss": 0.3517,
      "step": 25502
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.555298924446106,
      "learning_rate": 1.9885994526160856e-06,
      "loss": 0.4186,
      "step": 25503
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4334088563919067,
      "learning_rate": 1.9879899990037677e-06,
      "loss": 0.3653,
      "step": 25504
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6757423877716064,
      "learning_rate": 1.9873806284888077e-06,
      "loss": 0.4592,
      "step": 25505
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.508116364479065,
      "learning_rate": 1.986771341077521e-06,
      "loss": 0.4015,
      "step": 25506
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5889339447021484,
      "learning_rate": 1.986162136776232e-06,
      "loss": 0.4279,
      "step": 25507
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.252827167510986,
      "learning_rate": 1.985553015591254e-06,
      "loss": 0.4139,
      "step": 25508
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.535221815109253,
      "learning_rate": 1.984943977528907e-06,
      "loss": 0.473,
      "step": 25509
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4961142539978027,
      "learning_rate": 1.9843350225955095e-06,
      "loss": 0.3804,
      "step": 25510
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5789363384246826,
      "learning_rate": 1.983726150797376e-06,
      "loss": 0.4286,
      "step": 25511
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5609383583068848,
      "learning_rate": 1.9831173621408186e-06,
      "loss": 0.382,
      "step": 25512
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4705815315246582,
      "learning_rate": 1.982508656632154e-06,
      "loss": 0.4282,
      "step": 25513
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5333272218704224,
      "learning_rate": 1.9819000342776938e-06,
      "loss": 0.3795,
      "step": 25514
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5721901655197144,
      "learning_rate": 1.981291495083756e-06,
      "loss": 0.3993,
      "step": 25515
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5713998079299927,
      "learning_rate": 1.9806830390566465e-06,
      "loss": 0.3808,
      "step": 25516
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6711318492889404,
      "learning_rate": 1.980074666202676e-06,
      "loss": 0.4356,
      "step": 25517
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5104402303695679,
      "learning_rate": 1.9794663765281562e-06,
      "loss": 0.3644,
      "step": 25518
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5140700340270996,
      "learning_rate": 1.978858170039397e-06,
      "loss": 0.4332,
      "step": 25519
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4973264932632446,
      "learning_rate": 1.9782500467427056e-06,
      "loss": 0.3868,
      "step": 25520
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8773223161697388,
      "learning_rate": 1.977642006644386e-06,
      "loss": 1.0438,
      "step": 25521
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4490478038787842,
      "learning_rate": 1.977034049750747e-06,
      "loss": 1.0241,
      "step": 25522
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4993144273757935,
      "learning_rate": 1.976426176068097e-06,
      "loss": 0.3897,
      "step": 25523
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5507678985595703,
      "learning_rate": 1.975818385602736e-06,
      "loss": 0.3933,
      "step": 25524
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5917627811431885,
      "learning_rate": 1.9752106783609704e-06,
      "loss": 0.418,
      "step": 25525
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.566053032875061,
      "learning_rate": 1.9746030543491012e-06,
      "loss": 0.3683,
      "step": 25526
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5667665004730225,
      "learning_rate": 1.9739955135734335e-06,
      "loss": 0.4054,
      "step": 25527
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.541332483291626,
      "learning_rate": 1.9733880560402653e-06,
      "loss": 0.4042,
      "step": 25528
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5502381324768066,
      "learning_rate": 1.972780681755897e-06,
      "loss": 0.4609,
      "step": 25529
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0681726932525635,
      "learning_rate": 1.9721733907266316e-06,
      "loss": 0.3797,
      "step": 25530
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5471540689468384,
      "learning_rate": 1.9715661829587653e-06,
      "loss": 0.4417,
      "step": 25531
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.530734658241272,
      "learning_rate": 1.9709590584585936e-06,
      "loss": 0.3884,
      "step": 25532
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.491283893585205,
      "learning_rate": 1.9703520172324154e-06,
      "loss": 0.3741,
      "step": 25533
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5738354921340942,
      "learning_rate": 1.9697450592865276e-06,
      "loss": 0.4186,
      "step": 25534
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4729267358779907,
      "learning_rate": 1.969138184627226e-06,
      "loss": 0.4054,
      "step": 25535
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4960647821426392,
      "learning_rate": 1.968531393260804e-06,
      "loss": 0.3929,
      "step": 25536
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4551596641540527,
      "learning_rate": 1.967924685193552e-06,
      "loss": 0.3924,
      "step": 25537
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5735864639282227,
      "learning_rate": 1.9673180604317645e-06,
      "loss": 0.3852,
      "step": 25538
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5311594009399414,
      "learning_rate": 1.9667115189817343e-06,
      "loss": 0.3902,
      "step": 25539
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4918248653411865,
      "learning_rate": 1.9661050608497535e-06,
      "loss": 0.378,
      "step": 25540
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5261014699935913,
      "learning_rate": 1.96549868604211e-06,
      "loss": 0.3966,
      "step": 25541
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6205142736434937,
      "learning_rate": 1.964892394565091e-06,
      "loss": 0.4083,
      "step": 25542
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6805338859558105,
      "learning_rate": 1.964286186424986e-06,
      "loss": 0.4804,
      "step": 25543
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6505552530288696,
      "learning_rate": 1.963680061628086e-06,
      "loss": 0.3906,
      "step": 25544
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5785243511199951,
      "learning_rate": 1.963074020180672e-06,
      "loss": 0.4161,
      "step": 25545
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5798183679580688,
      "learning_rate": 1.9624680620890336e-06,
      "loss": 0.3791,
      "step": 25546
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5889424085617065,
      "learning_rate": 1.9618621873594524e-06,
      "loss": 0.4038,
      "step": 25547
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5202916860580444,
      "learning_rate": 1.9612563959982165e-06,
      "loss": 0.3757,
      "step": 25548
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.481682300567627,
      "learning_rate": 1.960650688011604e-06,
      "loss": 0.3871,
      "step": 25549
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4984855651855469,
      "learning_rate": 1.9600450634058986e-06,
      "loss": 0.3664,
      "step": 25550
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.739585041999817,
      "learning_rate": 1.9594395221873864e-06,
      "loss": 0.4436,
      "step": 25551
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.525938868522644,
      "learning_rate": 1.9588340643623426e-06,
      "loss": 0.3646,
      "step": 25552
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.610190749168396,
      "learning_rate": 1.9582286899370462e-06,
      "loss": 0.4712,
      "step": 25553
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5758237838745117,
      "learning_rate": 1.9576233989177772e-06,
      "loss": 0.3828,
      "step": 25554
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5375477075576782,
      "learning_rate": 1.9570181913108146e-06,
      "loss": 0.4105,
      "step": 25555
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5090129375457764,
      "learning_rate": 1.956413067122439e-06,
      "loss": 0.3662,
      "step": 25556
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5796664953231812,
      "learning_rate": 1.9558080263589176e-06,
      "loss": 0.4586,
      "step": 25557
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4294719696044922,
      "learning_rate": 1.95520306902653e-06,
      "loss": 0.3518,
      "step": 25558
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.480443000793457,
      "learning_rate": 1.9545981951315516e-06,
      "loss": 0.413,
      "step": 25559
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6420843601226807,
      "learning_rate": 1.953993404680257e-06,
      "loss": 0.4203,
      "step": 25560
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.488937258720398,
      "learning_rate": 1.9533886976789174e-06,
      "loss": 0.3967,
      "step": 25561
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.80708646774292,
      "learning_rate": 1.952784074133801e-06,
      "loss": 0.3519,
      "step": 25562
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6665458679199219,
      "learning_rate": 1.952179534051183e-06,
      "loss": 0.4021,
      "step": 25563
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5928465127944946,
      "learning_rate": 1.9515750774373343e-06,
      "loss": 0.3633,
      "step": 25564
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.622298240661621,
      "learning_rate": 1.9509707042985195e-06,
      "loss": 0.4724,
      "step": 25565
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.480741262435913,
      "learning_rate": 1.9503664146410116e-06,
      "loss": 0.3713,
      "step": 25566
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.618708848953247,
      "learning_rate": 1.9497622084710745e-06,
      "loss": 0.408,
      "step": 25567
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5154088735580444,
      "learning_rate": 1.9491580857949775e-06,
      "loss": 0.3839,
      "step": 25568
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6509881019592285,
      "learning_rate": 1.9485540466189844e-06,
      "loss": 0.4489,
      "step": 25569
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5293365716934204,
      "learning_rate": 1.94795009094936e-06,
      "loss": 0.3858,
      "step": 25570
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6405973434448242,
      "learning_rate": 1.9473462187923707e-06,
      "loss": 0.4495,
      "step": 25571
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.492830753326416,
      "learning_rate": 1.946742430154278e-06,
      "loss": 0.3643,
      "step": 25572
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5982383489608765,
      "learning_rate": 1.9461387250413433e-06,
      "loss": 0.9323,
      "step": 25573
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3753411769866943,
      "learning_rate": 1.945535103459827e-06,
      "loss": 0.9506,
      "step": 25574
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5831563472747803,
      "learning_rate": 1.9449315654159917e-06,
      "loss": 0.4569,
      "step": 25575
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6131322383880615,
      "learning_rate": 1.9443281109161007e-06,
      "loss": 0.3896,
      "step": 25576
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.550431728363037,
      "learning_rate": 1.943724739966407e-06,
      "loss": 0.4444,
      "step": 25577
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5244688987731934,
      "learning_rate": 1.943121452573169e-06,
      "loss": 0.3816,
      "step": 25578
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.538324236869812,
      "learning_rate": 1.9425182487426443e-06,
      "loss": 0.4115,
      "step": 25579
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5470889806747437,
      "learning_rate": 1.9419151284810933e-06,
      "loss": 0.3969,
      "step": 25580
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5849838256835938,
      "learning_rate": 1.941312091794765e-06,
      "loss": 0.4259,
      "step": 25581
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5140491724014282,
      "learning_rate": 1.94070913868992e-06,
      "loss": 0.3536,
      "step": 25582
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5465927124023438,
      "learning_rate": 1.9401062691728058e-06,
      "loss": 0.4091,
      "step": 25583
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6985989809036255,
      "learning_rate": 1.9395034832496793e-06,
      "loss": 0.3933,
      "step": 25584
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.545669436454773,
      "learning_rate": 1.93890078092679e-06,
      "loss": 0.4103,
      "step": 25585
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6077486276626587,
      "learning_rate": 1.938298162210389e-06,
      "loss": 0.3729,
      "step": 25586
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4988553524017334,
      "learning_rate": 1.9376956271067304e-06,
      "loss": 0.3947,
      "step": 25587
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.62175452709198,
      "learning_rate": 1.9370931756220614e-06,
      "loss": 0.3854,
      "step": 25588
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6557564735412598,
      "learning_rate": 1.9364908077626255e-06,
      "loss": 0.4156,
      "step": 25589
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.467783808708191,
      "learning_rate": 1.9358885235346747e-06,
      "loss": 0.3498,
      "step": 25590
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.757917046546936,
      "learning_rate": 1.9352863229444554e-06,
      "loss": 0.4141,
      "step": 25591
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.508196473121643,
      "learning_rate": 1.934684205998214e-06,
      "loss": 0.3636,
      "step": 25592
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.511880874633789,
      "learning_rate": 1.934082172702196e-06,
      "loss": 0.3963,
      "step": 25593
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5108894109725952,
      "learning_rate": 1.9334802230626416e-06,
      "loss": 0.3766,
      "step": 25594
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7120362520217896,
      "learning_rate": 1.9328783570857954e-06,
      "loss": 0.4369,
      "step": 25595
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4768431186676025,
      "learning_rate": 1.9322765747779037e-06,
      "loss": 0.349,
      "step": 25596
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.565604567527771,
      "learning_rate": 1.9316748761452043e-06,
      "loss": 0.4444,
      "step": 25597
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.023738145828247,
      "learning_rate": 1.9310732611939366e-06,
      "loss": 0.3695,
      "step": 25598
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4919240474700928,
      "learning_rate": 1.930471729930341e-06,
      "loss": 0.3948,
      "step": 25599
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4744946956634521,
      "learning_rate": 1.9298702823606606e-06,
      "loss": 0.3697,
      "step": 25600
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5817749500274658,
      "learning_rate": 1.9292689184911275e-06,
      "loss": 0.4119,
      "step": 25601
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.62461256980896,
      "learning_rate": 1.9286676383279847e-06,
      "loss": 0.3468,
      "step": 25602
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4944740533828735,
      "learning_rate": 1.928066441877461e-06,
      "loss": 0.4104,
      "step": 25603
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5462868213653564,
      "learning_rate": 1.9274653291458003e-06,
      "loss": 0.3862,
      "step": 25604
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6568403244018555,
      "learning_rate": 1.9268643001392283e-06,
      "loss": 0.4436,
      "step": 25605
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4950189590454102,
      "learning_rate": 1.9262633548639854e-06,
      "loss": 0.367,
      "step": 25606
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6551045179367065,
      "learning_rate": 1.9256624933263036e-06,
      "loss": 0.463,
      "step": 25607
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5192713737487793,
      "learning_rate": 1.9250617155324124e-06,
      "loss": 0.3868,
      "step": 25608
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7000160217285156,
      "learning_rate": 1.924461021488542e-06,
      "loss": 0.4477,
      "step": 25609
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5063306093215942,
      "learning_rate": 1.9238604112009242e-06,
      "loss": 0.3585,
      "step": 25610
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4911693334579468,
      "learning_rate": 1.923259884675788e-06,
      "loss": 0.3853,
      "step": 25611
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.665272831916809,
      "learning_rate": 1.9226594419193657e-06,
      "loss": 0.3946,
      "step": 25612
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5560930967330933,
      "learning_rate": 1.9220590829378804e-06,
      "loss": 0.3845,
      "step": 25613
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4372138977050781,
      "learning_rate": 1.9214588077375572e-06,
      "loss": 0.3507,
      "step": 25614
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.493947982788086,
      "learning_rate": 1.9208586163246244e-06,
      "loss": 0.4055,
      "step": 25615
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5751292705535889,
      "learning_rate": 1.9202585087053094e-06,
      "loss": 0.3793,
      "step": 25616
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6854910850524902,
      "learning_rate": 1.9196584848858315e-06,
      "loss": 0.4878,
      "step": 25617
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5612609386444092,
      "learning_rate": 1.9190585448724185e-06,
      "loss": 0.3685,
      "step": 25618
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5607120990753174,
      "learning_rate": 1.9184586886712884e-06,
      "loss": 0.4051,
      "step": 25619
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5638868808746338,
      "learning_rate": 1.9178589162886675e-06,
      "loss": 0.3843,
      "step": 25620
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.639705777168274,
      "learning_rate": 1.91725922773077e-06,
      "loss": 0.4258,
      "step": 25621
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5298995971679688,
      "learning_rate": 1.9166596230038205e-06,
      "loss": 0.3733,
      "step": 25622
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.583116054534912,
      "learning_rate": 1.916060102114039e-06,
      "loss": 0.4081,
      "step": 25623
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4904603958129883,
      "learning_rate": 1.91546066506764e-06,
      "loss": 0.3748,
      "step": 25624
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5331501960754395,
      "learning_rate": 1.9148613118708405e-06,
      "loss": 0.4503,
      "step": 25625
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.468837857246399,
      "learning_rate": 1.9142620425298574e-06,
      "loss": 0.3691,
      "step": 25626
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5996938943862915,
      "learning_rate": 1.9136628570509065e-06,
      "loss": 0.4347,
      "step": 25627
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5217024087905884,
      "learning_rate": 1.9130637554402054e-06,
      "loss": 0.394,
      "step": 25628
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5500550270080566,
      "learning_rate": 1.912464737703964e-06,
      "loss": 0.4199,
      "step": 25629
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6533771753311157,
      "learning_rate": 1.911865803848394e-06,
      "loss": 0.4123,
      "step": 25630
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.613789439201355,
      "learning_rate": 1.911266953879709e-06,
      "loss": 0.4344,
      "step": 25631
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5822967290878296,
      "learning_rate": 1.910668187804123e-06,
      "loss": 0.3399,
      "step": 25632
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4733675718307495,
      "learning_rate": 1.910069505627844e-06,
      "loss": 0.3922,
      "step": 25633
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6008057594299316,
      "learning_rate": 1.909470907357077e-06,
      "loss": 0.3875,
      "step": 25634
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4809321165084839,
      "learning_rate": 1.9088723929980346e-06,
      "loss": 0.38,
      "step": 25635
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5074397325515747,
      "learning_rate": 1.908273962556927e-06,
      "loss": 0.4189,
      "step": 25636
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6211220026016235,
      "learning_rate": 1.9076756160399546e-06,
      "loss": 0.4577,
      "step": 25637
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.51205575466156,
      "learning_rate": 1.9070773534533283e-06,
      "loss": 0.3974,
      "step": 25638
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.614595890045166,
      "learning_rate": 1.9064791748032495e-06,
      "loss": 0.5235,
      "step": 25639
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5631111860275269,
      "learning_rate": 1.905881080095926e-06,
      "loss": 0.3847,
      "step": 25640
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6474958658218384,
      "learning_rate": 1.9052830693375556e-06,
      "loss": 0.4012,
      "step": 25641
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5932189226150513,
      "learning_rate": 1.9046851425343449e-06,
      "loss": 0.3938,
      "step": 25642
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.800486445426941,
      "learning_rate": 1.9040872996924963e-06,
      "loss": 0.9464,
      "step": 25643
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4385634660720825,
      "learning_rate": 1.9034895408182087e-06,
      "loss": 0.909,
      "step": 25644
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5315592288970947,
      "learning_rate": 1.9028918659176787e-06,
      "loss": 0.3812,
      "step": 25645
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5611974000930786,
      "learning_rate": 1.9022942749971085e-06,
      "loss": 0.386,
      "step": 25646
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6040267944335938,
      "learning_rate": 1.9016967680626951e-06,
      "loss": 0.466,
      "step": 25647
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.533858299255371,
      "learning_rate": 1.9010993451206395e-06,
      "loss": 0.3965,
      "step": 25648
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6202502250671387,
      "learning_rate": 1.900502006177134e-06,
      "loss": 0.4305,
      "step": 25649
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.528368353843689,
      "learning_rate": 1.8999047512383718e-06,
      "loss": 0.3606,
      "step": 25650
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.5173988342285156,
      "learning_rate": 1.8993075803105497e-06,
      "loss": 0.4566,
      "step": 25651
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5257164239883423,
      "learning_rate": 1.8987104933998645e-06,
      "loss": 0.3718,
      "step": 25652
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5126067399978638,
      "learning_rate": 1.8981134905125031e-06,
      "loss": 0.4425,
      "step": 25653
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6431941986083984,
      "learning_rate": 1.8975165716546628e-06,
      "loss": 0.3967,
      "step": 25654
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.580947756767273,
      "learning_rate": 1.896919736832531e-06,
      "loss": 0.9701,
      "step": 25655
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5054066181182861,
      "learning_rate": 1.8963229860523002e-06,
      "loss": 0.9799,
      "step": 25656
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5723015069961548,
      "learning_rate": 1.8957263193201568e-06,
      "loss": 0.4404,
      "step": 25657
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5253669023513794,
      "learning_rate": 1.8951297366422904e-06,
      "loss": 0.364,
      "step": 25658
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5165587663650513,
      "learning_rate": 1.8945332380248914e-06,
      "loss": 0.4113,
      "step": 25659
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.496710181236267,
      "learning_rate": 1.8939368234741439e-06,
      "loss": 0.3592,
      "step": 25660
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4659044742584229,
      "learning_rate": 1.8933404929962317e-06,
      "loss": 0.3961,
      "step": 25661
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6073877811431885,
      "learning_rate": 1.8927442465973423e-06,
      "loss": 0.3749,
      "step": 25662
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5281332731246948,
      "learning_rate": 1.8921480842836582e-06,
      "loss": 0.3938,
      "step": 25663
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5550788640975952,
      "learning_rate": 1.8915520060613669e-06,
      "loss": 0.3866,
      "step": 25664
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5717566013336182,
      "learning_rate": 1.8909560119366467e-06,
      "loss": 0.425,
      "step": 25665
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5679726600646973,
      "learning_rate": 1.890360101915677e-06,
      "loss": 0.3972,
      "step": 25666
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5311542749404907,
      "learning_rate": 1.889764276004642e-06,
      "loss": 0.3988,
      "step": 25667
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.519381046295166,
      "learning_rate": 1.8891685342097222e-06,
      "loss": 0.3848,
      "step": 25668
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5492388010025024,
      "learning_rate": 1.8885728765370925e-06,
      "loss": 0.4005,
      "step": 25669
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.544330358505249,
      "learning_rate": 1.887977302992935e-06,
      "loss": 0.3967,
      "step": 25670
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5250732898712158,
      "learning_rate": 1.887381813583422e-06,
      "loss": 0.4676,
      "step": 25671
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4738837480545044,
      "learning_rate": 1.8867864083147358e-06,
      "loss": 0.3642,
      "step": 25672
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6052217483520508,
      "learning_rate": 1.8861910871930445e-06,
      "loss": 0.4259,
      "step": 25673
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5570095777511597,
      "learning_rate": 1.8855958502245298e-06,
      "loss": 0.3798,
      "step": 25674
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5561683177947998,
      "learning_rate": 1.885000697415358e-06,
      "loss": 0.4348,
      "step": 25675
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5059794187545776,
      "learning_rate": 1.8844056287717095e-06,
      "loss": 0.3536,
      "step": 25676
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.533616065979004,
      "learning_rate": 1.8838106442997484e-06,
      "loss": 0.3999,
      "step": 25677
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4986789226531982,
      "learning_rate": 1.8832157440056497e-06,
      "loss": 0.37,
      "step": 25678
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5917326211929321,
      "learning_rate": 1.8826209278955853e-06,
      "loss": 0.4288,
      "step": 25679
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5245944261550903,
      "learning_rate": 1.8820261959757235e-06,
      "loss": 0.3697,
      "step": 25680
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.568915843963623,
      "learning_rate": 1.8814315482522271e-06,
      "loss": 0.3992,
      "step": 25681
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4847404956817627,
      "learning_rate": 1.8808369847312691e-06,
      "loss": 0.3598,
      "step": 25682
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.599366307258606,
      "learning_rate": 1.8802425054190142e-06,
      "loss": 0.4202,
      "step": 25683
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.519139289855957,
      "learning_rate": 1.8796481103216313e-06,
      "loss": 0.3652,
      "step": 25684
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5274003744125366,
      "learning_rate": 1.8790537994452828e-06,
      "loss": 0.4575,
      "step": 25685
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5287857055664062,
      "learning_rate": 1.8784595727961296e-06,
      "loss": 0.3788,
      "step": 25686
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.600180745124817,
      "learning_rate": 1.877865430380338e-06,
      "loss": 0.4346,
      "step": 25687
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5396233797073364,
      "learning_rate": 1.8772713722040726e-06,
      "loss": 0.3928,
      "step": 25688
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5265438556671143,
      "learning_rate": 1.8766773982734887e-06,
      "loss": 0.4295,
      "step": 25689
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6340817213058472,
      "learning_rate": 1.8760835085947527e-06,
      "loss": 0.3888,
      "step": 25690
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5459500551223755,
      "learning_rate": 1.8754897031740194e-06,
      "loss": 0.4456,
      "step": 25691
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5311532020568848,
      "learning_rate": 1.8748959820174517e-06,
      "loss": 0.3836,
      "step": 25692
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.554507851600647,
      "learning_rate": 1.8743023451312026e-06,
      "loss": 0.4205,
      "step": 25693
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5051697492599487,
      "learning_rate": 1.8737087925214325e-06,
      "loss": 0.3722,
      "step": 25694
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.535856008529663,
      "learning_rate": 1.873115324194299e-06,
      "loss": 0.3805,
      "step": 25695
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5169260501861572,
      "learning_rate": 1.8725219401559557e-06,
      "loss": 0.38,
      "step": 25696
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5812801122665405,
      "learning_rate": 1.8719286404125537e-06,
      "loss": 0.4567,
      "step": 25697
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4625765085220337,
      "learning_rate": 1.8713354249702487e-06,
      "loss": 0.3617,
      "step": 25698
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7727552652359009,
      "learning_rate": 1.8707422938351937e-06,
      "loss": 0.4577,
      "step": 25699
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4941152334213257,
      "learning_rate": 1.870149247013543e-06,
      "loss": 0.3376,
      "step": 25700
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5647646188735962,
      "learning_rate": 1.8695562845114456e-06,
      "loss": 0.4067,
      "step": 25701
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5579594373703003,
      "learning_rate": 1.8689634063350471e-06,
      "loss": 0.3885,
      "step": 25702
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7887731790542603,
      "learning_rate": 1.868370612490501e-06,
      "loss": 1.0133,
      "step": 25703
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4626554250717163,
      "learning_rate": 1.8677779029839571e-06,
      "loss": 1.0107,
      "step": 25704
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.58745539188385,
      "learning_rate": 1.8671852778215582e-06,
      "loss": 0.447,
      "step": 25705
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5220903158187866,
      "learning_rate": 1.866592737009455e-06,
      "loss": 0.3988,
      "step": 25706
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5587657690048218,
      "learning_rate": 1.866000280553789e-06,
      "loss": 0.4585,
      "step": 25707
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.538223147392273,
      "learning_rate": 1.8654079084607091e-06,
      "loss": 0.397,
      "step": 25708
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5418294668197632,
      "learning_rate": 1.8648156207363554e-06,
      "loss": 0.4181,
      "step": 25709
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.484139084815979,
      "learning_rate": 1.8642234173868724e-06,
      "loss": 0.3574,
      "step": 25710
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.535617470741272,
      "learning_rate": 1.8636312984184034e-06,
      "loss": 0.4286,
      "step": 25711
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.745819330215454,
      "learning_rate": 1.8630392638370899e-06,
      "loss": 0.4097,
      "step": 25712
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5773727893829346,
      "learning_rate": 1.8624473136490684e-06,
      "loss": 0.4522,
      "step": 25713
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5228888988494873,
      "learning_rate": 1.86185544786048e-06,
      "loss": 0.363,
      "step": 25714
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5499614477157593,
      "learning_rate": 1.8612636664774674e-06,
      "loss": 0.4182,
      "step": 25715
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5338679552078247,
      "learning_rate": 1.8606719695061648e-06,
      "loss": 0.3575,
      "step": 25716
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.569822907447815,
      "learning_rate": 1.860080356952707e-06,
      "loss": 0.4069,
      "step": 25717
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4699158668518066,
      "learning_rate": 1.8594888288232327e-06,
      "loss": 0.3478,
      "step": 25718
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5496711730957031,
      "learning_rate": 1.8588973851238756e-06,
      "loss": 0.4027,
      "step": 25719
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.569794774055481,
      "learning_rate": 1.8583060258607744e-06,
      "loss": 0.382,
      "step": 25720
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4679639339447021,
      "learning_rate": 1.8577147510400584e-06,
      "loss": 0.3804,
      "step": 25721
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5853917598724365,
      "learning_rate": 1.8571235606678572e-06,
      "loss": 0.3926,
      "step": 25722
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6137268543243408,
      "learning_rate": 1.8565324547503072e-06,
      "loss": 0.4663,
      "step": 25723
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5543233156204224,
      "learning_rate": 1.8559414332935398e-06,
      "loss": 0.3671,
      "step": 25724
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5536613464355469,
      "learning_rate": 1.8553504963036806e-06,
      "loss": 0.4353,
      "step": 25725
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5148687362670898,
      "learning_rate": 1.854759643786862e-06,
      "loss": 0.3665,
      "step": 25726
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.519033432006836,
      "learning_rate": 1.8541688757492094e-06,
      "loss": 0.3824,
      "step": 25727
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.614367961883545,
      "learning_rate": 1.8535781921968532e-06,
      "loss": 0.386,
      "step": 25728
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5187954902648926,
      "learning_rate": 1.8529875931359165e-06,
      "loss": 0.3951,
      "step": 25729
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.546182632446289,
      "learning_rate": 1.8523970785725253e-06,
      "loss": 0.397,
      "step": 25730
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.570699691772461,
      "learning_rate": 1.8518066485128073e-06,
      "loss": 0.3978,
      "step": 25731
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.519637107849121,
      "learning_rate": 1.8512163029628848e-06,
      "loss": 0.3833,
      "step": 25732
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4763559103012085,
      "learning_rate": 1.850626041928877e-06,
      "loss": 0.3671,
      "step": 25733
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5141524076461792,
      "learning_rate": 1.8500358654169082e-06,
      "loss": 0.3602,
      "step": 25734
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5309902429580688,
      "learning_rate": 1.8494457734331006e-06,
      "loss": 0.4533,
      "step": 25735
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6043510437011719,
      "learning_rate": 1.8488557659835759e-06,
      "loss": 0.4139,
      "step": 25736
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6131542921066284,
      "learning_rate": 1.8482658430744515e-06,
      "loss": 0.4101,
      "step": 25737
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5011807680130005,
      "learning_rate": 1.8476760047118426e-06,
      "loss": 0.3572,
      "step": 25738
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4971976280212402,
      "learning_rate": 1.8470862509018704e-06,
      "loss": 0.4181,
      "step": 25739
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.608954906463623,
      "learning_rate": 1.846496581650653e-06,
      "loss": 0.4024,
      "step": 25740
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0446701049804688,
      "learning_rate": 1.8459069969643019e-06,
      "loss": 0.4168,
      "step": 25741
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5015156269073486,
      "learning_rate": 1.8453174968489363e-06,
      "loss": 0.3772,
      "step": 25742
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5522878170013428,
      "learning_rate": 1.844728081310666e-06,
      "loss": 0.4298,
      "step": 25743
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4433894157409668,
      "learning_rate": 1.8441387503556096e-06,
      "loss": 0.3613,
      "step": 25744
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.594362735748291,
      "learning_rate": 1.843549503989873e-06,
      "loss": 0.4075,
      "step": 25745
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.556965947151184,
      "learning_rate": 1.8429603422195707e-06,
      "loss": 0.3789,
      "step": 25746
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5890331268310547,
      "learning_rate": 1.8423712650508162e-06,
      "loss": 0.4504,
      "step": 25747
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.534519076347351,
      "learning_rate": 1.8417822724897171e-06,
      "loss": 0.4098,
      "step": 25748
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4902225732803345,
      "learning_rate": 1.8411933645423773e-06,
      "loss": 0.3984,
      "step": 25749
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.535470962524414,
      "learning_rate": 1.840604541214911e-06,
      "loss": 0.3834,
      "step": 25750
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5710941553115845,
      "learning_rate": 1.840015802513424e-06,
      "loss": 0.4704,
      "step": 25751
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5191657543182373,
      "learning_rate": 1.8394271484440229e-06,
      "loss": 0.3691,
      "step": 25752
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6699119806289673,
      "learning_rate": 1.8388385790128084e-06,
      "loss": 0.9376,
      "step": 25753
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.438055396080017,
      "learning_rate": 1.8382500942258874e-06,
      "loss": 1.0345,
      "step": 25754
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.561628818511963,
      "learning_rate": 1.8376616940893655e-06,
      "loss": 0.4369,
      "step": 25755
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4488774538040161,
      "learning_rate": 1.8370733786093463e-06,
      "loss": 0.3655,
      "step": 25756
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5762354135513306,
      "learning_rate": 1.8364851477919287e-06,
      "loss": 0.4256,
      "step": 25757
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4879885911941528,
      "learning_rate": 1.8358970016432132e-06,
      "loss": 0.399,
      "step": 25758
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.606248378753662,
      "learning_rate": 1.835308940169299e-06,
      "loss": 0.4769,
      "step": 25759
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5046993494033813,
      "learning_rate": 1.834720963376291e-06,
      "loss": 0.3734,
      "step": 25760
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4679216146469116,
      "learning_rate": 1.834133071270281e-06,
      "loss": 0.4032,
      "step": 25761
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4982166290283203,
      "learning_rate": 1.8335452638573702e-06,
      "loss": 0.352,
      "step": 25762
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6874576807022095,
      "learning_rate": 1.8329575411436529e-06,
      "loss": 0.4651,
      "step": 25763
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.520689845085144,
      "learning_rate": 1.8323699031352271e-06,
      "loss": 0.3843,
      "step": 25764
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.614291787147522,
      "learning_rate": 1.8317823498381838e-06,
      "loss": 0.4604,
      "step": 25765
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6095571517944336,
      "learning_rate": 1.831194881258619e-06,
      "loss": 0.4075,
      "step": 25766
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5773909091949463,
      "learning_rate": 1.830607497402629e-06,
      "loss": 0.4094,
      "step": 25767
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.589663028717041,
      "learning_rate": 1.8300201982763022e-06,
      "loss": 0.4142,
      "step": 25768
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6653116941452026,
      "learning_rate": 1.8294329838857283e-06,
      "loss": 0.9661,
      "step": 25769
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.447889804840088,
      "learning_rate": 1.828845854236999e-06,
      "loss": 0.958,
      "step": 25770
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6924872398376465,
      "learning_rate": 1.8282588093362053e-06,
      "loss": 0.9895,
      "step": 25771
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3919856548309326,
      "learning_rate": 1.8276718491894374e-06,
      "loss": 0.9671,
      "step": 25772
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.546684980392456,
      "learning_rate": 1.8270849738027806e-06,
      "loss": 0.3917,
      "step": 25773
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6540547609329224,
      "learning_rate": 1.8264981831823193e-06,
      "loss": 0.395,
      "step": 25774
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5827990770339966,
      "learning_rate": 1.8259114773341413e-06,
      "loss": 0.4375,
      "step": 25775
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5952199697494507,
      "learning_rate": 1.8253248562643356e-06,
      "loss": 0.3826,
      "step": 25776
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.563598394393921,
      "learning_rate": 1.8247383199789804e-06,
      "loss": 0.4066,
      "step": 25777
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5038011074066162,
      "learning_rate": 1.8241518684841642e-06,
      "loss": 0.3751,
      "step": 25778
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6545807123184204,
      "learning_rate": 1.8235655017859643e-06,
      "loss": 0.422,
      "step": 25779
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5296324491500854,
      "learning_rate": 1.822979219890466e-06,
      "loss": 0.3662,
      "step": 25780
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5502796173095703,
      "learning_rate": 1.82239302280375e-06,
      "loss": 0.4216,
      "step": 25781
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5297824144363403,
      "learning_rate": 1.8218069105318936e-06,
      "loss": 0.3646,
      "step": 25782
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4739350080490112,
      "learning_rate": 1.8212208830809785e-06,
      "loss": 0.384,
      "step": 25783
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5827728509902954,
      "learning_rate": 1.82063494045708e-06,
      "loss": 0.3649,
      "step": 25784
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5123176574707031,
      "learning_rate": 1.8200490826662788e-06,
      "loss": 0.3908,
      "step": 25785
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5556554794311523,
      "learning_rate": 1.8194633097146452e-06,
      "loss": 0.3812,
      "step": 25786
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.556474208831787,
      "learning_rate": 1.8188776216082604e-06,
      "loss": 0.4079,
      "step": 25787
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5128421783447266,
      "learning_rate": 1.8182920183531982e-06,
      "loss": 0.3636,
      "step": 25788
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6529394388198853,
      "learning_rate": 1.8177064999555317e-06,
      "loss": 0.454,
      "step": 25789
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6149564981460571,
      "learning_rate": 1.817121066421329e-06,
      "loss": 0.3808,
      "step": 25790
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5371228456497192,
      "learning_rate": 1.8165357177566678e-06,
      "loss": 0.4577,
      "step": 25791
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5119903087615967,
      "learning_rate": 1.8159504539676187e-06,
      "loss": 0.3659,
      "step": 25792
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.611159086227417,
      "learning_rate": 1.8153652750602502e-06,
      "loss": 0.9078,
      "step": 25793
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4003366231918335,
      "learning_rate": 1.8147801810406296e-06,
      "loss": 1.004,
      "step": 25794
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5628048181533813,
      "learning_rate": 1.8141951719148276e-06,
      "loss": 0.4609,
      "step": 25795
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.52053701877594,
      "learning_rate": 1.8136102476889117e-06,
      "loss": 0.3902,
      "step": 25796
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.599419355392456,
      "learning_rate": 1.8130254083689503e-06,
      "loss": 0.4418,
      "step": 25797
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5750701427459717,
      "learning_rate": 1.8124406539610063e-06,
      "loss": 0.3847,
      "step": 25798
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.567489504814148,
      "learning_rate": 1.8118559844711436e-06,
      "loss": 0.4265,
      "step": 25799
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.482774257659912,
      "learning_rate": 1.8112713999054276e-06,
      "loss": 0.3666,
      "step": 25800
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5010454654693604,
      "learning_rate": 1.8106869002699246e-06,
      "loss": 0.4249,
      "step": 25801
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5686744451522827,
      "learning_rate": 1.8101024855706906e-06,
      "loss": 0.3657,
      "step": 25802
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5519204139709473,
      "learning_rate": 1.8095181558137931e-06,
      "loss": 0.3882,
      "step": 25803
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4869458675384521,
      "learning_rate": 1.8089339110052862e-06,
      "loss": 0.3677,
      "step": 25804
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5196093320846558,
      "learning_rate": 1.808349751151236e-06,
      "loss": 0.4152,
      "step": 25805
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4766430854797363,
      "learning_rate": 1.8077656762576957e-06,
      "loss": 0.3666,
      "step": 25806
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6298435926437378,
      "learning_rate": 1.8071816863307256e-06,
      "loss": 0.3952,
      "step": 25807
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6182745695114136,
      "learning_rate": 1.8065977813763847e-06,
      "loss": 0.3985,
      "step": 25808
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5439600944519043,
      "learning_rate": 1.8060139614007265e-06,
      "loss": 0.4181,
      "step": 25809
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5900408029556274,
      "learning_rate": 1.8054302264098033e-06,
      "loss": 0.3821,
      "step": 25810
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5971651077270508,
      "learning_rate": 1.8048465764096735e-06,
      "loss": 0.4313,
      "step": 25811
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5658659934997559,
      "learning_rate": 1.8042630114063886e-06,
      "loss": 0.3667,
      "step": 25812
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5196126699447632,
      "learning_rate": 1.803679531406005e-06,
      "loss": 0.4155,
      "step": 25813
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4358506202697754,
      "learning_rate": 1.8030961364145715e-06,
      "loss": 0.3563,
      "step": 25814
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5678160190582275,
      "learning_rate": 1.8025128264381352e-06,
      "loss": 0.4685,
      "step": 25815
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5737817287445068,
      "learning_rate": 1.8019296014827492e-06,
      "loss": 0.3653,
      "step": 25816
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6339718103408813,
      "learning_rate": 1.8013464615544662e-06,
      "loss": 0.4997,
      "step": 25817
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5502541065216064,
      "learning_rate": 1.800763406659327e-06,
      "loss": 0.3783,
      "step": 25818
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5931897163391113,
      "learning_rate": 1.8001804368033859e-06,
      "loss": 0.4117,
      "step": 25819
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5050485134124756,
      "learning_rate": 1.7995975519926834e-06,
      "loss": 0.4056,
      "step": 25820
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.534524917602539,
      "learning_rate": 1.7990147522332701e-06,
      "loss": 0.4114,
      "step": 25821
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5546678304672241,
      "learning_rate": 1.7984320375311848e-06,
      "loss": 0.3756,
      "step": 25822
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5293922424316406,
      "learning_rate": 1.7978494078924734e-06,
      "loss": 0.4172,
      "step": 25823
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5543984174728394,
      "learning_rate": 1.7972668633231838e-06,
      "loss": 0.3748,
      "step": 25824
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.549131155014038,
      "learning_rate": 1.796684403829353e-06,
      "loss": 0.4325,
      "step": 25825
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.429604172706604,
      "learning_rate": 1.7961020294170194e-06,
      "loss": 0.3469,
      "step": 25826
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8752621412277222,
      "learning_rate": 1.7955197400922275e-06,
      "loss": 0.9858,
      "step": 25827
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4154345989227295,
      "learning_rate": 1.7949375358610143e-06,
      "loss": 0.9532,
      "step": 25828
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5250194072723389,
      "learning_rate": 1.7943554167294254e-06,
      "loss": 0.4265,
      "step": 25829
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.482879638671875,
      "learning_rate": 1.7937733827034876e-06,
      "loss": 0.3604,
      "step": 25830
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5153605937957764,
      "learning_rate": 1.793191433789242e-06,
      "loss": 0.3835,
      "step": 25831
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6135941743850708,
      "learning_rate": 1.7926095699927238e-06,
      "loss": 0.3902,
      "step": 25832
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.61076819896698,
      "learning_rate": 1.7920277913199712e-06,
      "loss": 0.4583,
      "step": 25833
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6389358043670654,
      "learning_rate": 1.7914460977770165e-06,
      "loss": 0.3993,
      "step": 25834
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5192862749099731,
      "learning_rate": 1.790864489369889e-06,
      "loss": 0.4245,
      "step": 25835
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5712695121765137,
      "learning_rate": 1.790282966104624e-06,
      "loss": 0.3807,
      "step": 25836
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6179832220077515,
      "learning_rate": 1.7897015279872554e-06,
      "loss": 0.4467,
      "step": 25837
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4752254486083984,
      "learning_rate": 1.7891201750238073e-06,
      "loss": 0.3585,
      "step": 25838
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6003715991973877,
      "learning_rate": 1.7885389072203163e-06,
      "loss": 0.4286,
      "step": 25839
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5280694961547852,
      "learning_rate": 1.787957724582805e-06,
      "loss": 0.379,
      "step": 25840
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5474909543991089,
      "learning_rate": 1.7873766271173054e-06,
      "loss": 0.4534,
      "step": 25841
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.472765564918518,
      "learning_rate": 1.7867956148298415e-06,
      "loss": 0.3461,
      "step": 25842
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.475342869758606,
      "learning_rate": 1.7862146877264408e-06,
      "loss": 0.4056,
      "step": 25843
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5085334777832031,
      "learning_rate": 1.7856338458131296e-06,
      "loss": 0.387,
      "step": 25844
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.519521713256836,
      "learning_rate": 1.7850530890959316e-06,
      "loss": 0.4123,
      "step": 25845
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5545644760131836,
      "learning_rate": 1.784472417580867e-06,
      "loss": 0.4199,
      "step": 25846
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.604406476020813,
      "learning_rate": 1.7838918312739606e-06,
      "loss": 0.4414,
      "step": 25847
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4255915880203247,
      "learning_rate": 1.7833113301812343e-06,
      "loss": 0.365,
      "step": 25848
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5141716003417969,
      "learning_rate": 1.78273091430871e-06,
      "loss": 0.3949,
      "step": 25849
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5167075395584106,
      "learning_rate": 1.7821505836624076e-06,
      "loss": 0.3806,
      "step": 25850
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5743305683135986,
      "learning_rate": 1.7815703382483418e-06,
      "loss": 0.4466,
      "step": 25851
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5641745328903198,
      "learning_rate": 1.7809901780725337e-06,
      "loss": 0.3585,
      "step": 25852
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6187762022018433,
      "learning_rate": 1.7804101031410027e-06,
      "loss": 0.4285,
      "step": 25853
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.59128999710083,
      "learning_rate": 1.7798301134597595e-06,
      "loss": 0.4009,
      "step": 25854
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5934014320373535,
      "learning_rate": 1.7792502090348264e-06,
      "loss": 0.4186,
      "step": 25855
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5343661308288574,
      "learning_rate": 1.7786703898722102e-06,
      "loss": 0.3994,
      "step": 25856
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.167586326599121,
      "learning_rate": 1.778090655977932e-06,
      "loss": 0.4696,
      "step": 25857
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5379323959350586,
      "learning_rate": 1.7775110073579993e-06,
      "loss": 0.3813,
      "step": 25858
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5651600360870361,
      "learning_rate": 1.7769314440184249e-06,
      "loss": 0.4143,
      "step": 25859
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4435933828353882,
      "learning_rate": 1.7763519659652239e-06,
      "loss": 0.3431,
      "step": 25860
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5866143703460693,
      "learning_rate": 1.775772573204403e-06,
      "loss": 0.3913,
      "step": 25861
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5585592985153198,
      "learning_rate": 1.7751932657419701e-06,
      "loss": 0.3761,
      "step": 25862
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5737481117248535,
      "learning_rate": 1.7746140435839343e-06,
      "loss": 0.4521,
      "step": 25863
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4750269651412964,
      "learning_rate": 1.7740349067363038e-06,
      "loss": 0.365,
      "step": 25864
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6187094449996948,
      "learning_rate": 1.7734558552050907e-06,
      "loss": 0.4311,
      "step": 25865
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5863748788833618,
      "learning_rate": 1.77287688899629e-06,
      "loss": 0.3931,
      "step": 25866
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5654231309890747,
      "learning_rate": 1.7722980081159124e-06,
      "loss": 0.4238,
      "step": 25867
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5325642824172974,
      "learning_rate": 1.7717192125699601e-06,
      "loss": 0.3643,
      "step": 25868
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4849482774734497,
      "learning_rate": 1.7711405023644402e-06,
      "loss": 0.3943,
      "step": 25869
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4638538360595703,
      "learning_rate": 1.7705618775053523e-06,
      "loss": 0.375,
      "step": 25870
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6084250211715698,
      "learning_rate": 1.7699833379986953e-06,
      "loss": 1.0325,
      "step": 25871
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5113654136657715,
      "learning_rate": 1.7694048838504706e-06,
      "loss": 1.0058,
      "step": 25872
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.599647045135498,
      "learning_rate": 1.7688265150666817e-06,
      "loss": 0.4599,
      "step": 25873
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5295029878616333,
      "learning_rate": 1.7682482316533212e-06,
      "loss": 0.373,
      "step": 25874
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4991168975830078,
      "learning_rate": 1.767670033616392e-06,
      "loss": 0.4573,
      "step": 25875
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5721749067306519,
      "learning_rate": 1.7670919209618865e-06,
      "loss": 0.3988,
      "step": 25876
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5070340633392334,
      "learning_rate": 1.7665138936958058e-06,
      "loss": 0.4391,
      "step": 25877
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5027838945388794,
      "learning_rate": 1.7659359518241393e-06,
      "loss": 0.3564,
      "step": 25878
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5642927885055542,
      "learning_rate": 1.7653580953528837e-06,
      "loss": 0.4251,
      "step": 25879
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5143195390701294,
      "learning_rate": 1.7647803242880346e-06,
      "loss": 0.3875,
      "step": 25880
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5206440687179565,
      "learning_rate": 1.7642026386355837e-06,
      "loss": 0.3959,
      "step": 25881
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5122871398925781,
      "learning_rate": 1.7636250384015175e-06,
      "loss": 0.3804,
      "step": 25882
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6287930011749268,
      "learning_rate": 1.763047523591831e-06,
      "loss": 0.409,
      "step": 25883
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5903717279434204,
      "learning_rate": 1.7624700942125127e-06,
      "loss": 0.375,
      "step": 25884
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6635613441467285,
      "learning_rate": 1.7618927502695538e-06,
      "loss": 0.447,
      "step": 25885
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4900158643722534,
      "learning_rate": 1.7613154917689424e-06,
      "loss": 0.3644,
      "step": 25886
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.561524748802185,
      "learning_rate": 1.7607383187166593e-06,
      "loss": 0.4061,
      "step": 25887
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5140748023986816,
      "learning_rate": 1.7601612311186955e-06,
      "loss": 0.3719,
      "step": 25888
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5833728313446045,
      "learning_rate": 1.7595842289810395e-06,
      "loss": 0.4511,
      "step": 25889
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6488591432571411,
      "learning_rate": 1.7590073123096685e-06,
      "loss": 0.3875,
      "step": 25890
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6131411790847778,
      "learning_rate": 1.7584304811105724e-06,
      "loss": 0.4074,
      "step": 25891
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7515792846679688,
      "learning_rate": 1.7578537353897295e-06,
      "loss": 0.4001,
      "step": 25892
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7474349737167358,
      "learning_rate": 1.7572770751531254e-06,
      "loss": 0.4399,
      "step": 25893
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5674593448638916,
      "learning_rate": 1.756700500406735e-06,
      "loss": 0.3696,
      "step": 25894
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5815074443817139,
      "learning_rate": 1.7561240111565437e-06,
      "loss": 0.4238,
      "step": 25895
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5644007921218872,
      "learning_rate": 1.755547607408531e-06,
      "loss": 0.3794,
      "step": 25896
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5529106855392456,
      "learning_rate": 1.7549712891686731e-06,
      "loss": 0.4394,
      "step": 25897
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4936883449554443,
      "learning_rate": 1.7543950564429446e-06,
      "loss": 0.4014,
      "step": 25898
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.846151351928711,
      "learning_rate": 1.7538189092373248e-06,
      "loss": 0.4385,
      "step": 25899
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5201646089553833,
      "learning_rate": 1.7532428475577901e-06,
      "loss": 0.3813,
      "step": 25900
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6487065553665161,
      "learning_rate": 1.7526668714103156e-06,
      "loss": 0.4488,
      "step": 25901
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5112438201904297,
      "learning_rate": 1.7520909808008745e-06,
      "loss": 0.3756,
      "step": 25902
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6325870752334595,
      "learning_rate": 1.7515151757354365e-06,
      "loss": 0.4122,
      "step": 25903
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6119887828826904,
      "learning_rate": 1.7509394562199755e-06,
      "loss": 0.384,
      "step": 25904
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5117264986038208,
      "learning_rate": 1.750363822260467e-06,
      "loss": 0.4221,
      "step": 25905
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5363441705703735,
      "learning_rate": 1.7497882738628768e-06,
      "loss": 0.3505,
      "step": 25906
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4654064178466797,
      "learning_rate": 1.7492128110331718e-06,
      "loss": 0.4,
      "step": 25907
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5591516494750977,
      "learning_rate": 1.748637433777325e-06,
      "loss": 0.3709,
      "step": 25908
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5227317810058594,
      "learning_rate": 1.7480621421013044e-06,
      "loss": 0.4415,
      "step": 25909
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4754735231399536,
      "learning_rate": 1.7474869360110714e-06,
      "loss": 0.3652,
      "step": 25910
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5402448177337646,
      "learning_rate": 1.7469118155125997e-06,
      "loss": 0.4177,
      "step": 25911
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6551384925842285,
      "learning_rate": 1.746336780611846e-06,
      "loss": 0.3903,
      "step": 25912
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5585423707962036,
      "learning_rate": 1.7457618313147805e-06,
      "loss": 0.4043,
      "step": 25913
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5181063413619995,
      "learning_rate": 1.7451869676273625e-06,
      "loss": 0.3931,
      "step": 25914
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.51300847530365,
      "learning_rate": 1.7446121895555556e-06,
      "loss": 0.4115,
      "step": 25915
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.58294677734375,
      "learning_rate": 1.7440374971053232e-06,
      "loss": 0.3797,
      "step": 25916
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5014622211456299,
      "learning_rate": 1.7434628902826246e-06,
      "loss": 0.4824,
      "step": 25917
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5313336849212646,
      "learning_rate": 1.7428883690934152e-06,
      "loss": 0.3835,
      "step": 25918
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.972420692443848,
      "learning_rate": 1.7423139335436578e-06,
      "loss": 0.4405,
      "step": 25919
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.547115445137024,
      "learning_rate": 1.7417395836393103e-06,
      "loss": 0.3907,
      "step": 25920
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5943927764892578,
      "learning_rate": 1.741165319386331e-06,
      "loss": 0.4449,
      "step": 25921
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4134708642959595,
      "learning_rate": 1.7405911407906728e-06,
      "loss": 0.3553,
      "step": 25922
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.606211543083191,
      "learning_rate": 1.7400170478582913e-06,
      "loss": 0.4222,
      "step": 25923
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5327589511871338,
      "learning_rate": 1.73944304059514e-06,
      "loss": 0.371,
      "step": 25924
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5688939094543457,
      "learning_rate": 1.738869119007176e-06,
      "loss": 0.4406,
      "step": 25925
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6387947797775269,
      "learning_rate": 1.7382952831003485e-06,
      "loss": 0.3467,
      "step": 25926
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.090580701828003,
      "learning_rate": 1.7377215328806107e-06,
      "loss": 0.42,
      "step": 25927
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4436074495315552,
      "learning_rate": 1.7371478683539112e-06,
      "loss": 0.3767,
      "step": 25928
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.694077730178833,
      "learning_rate": 1.7365742895262038e-06,
      "loss": 0.5035,
      "step": 25929
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5451722145080566,
      "learning_rate": 1.736000796403432e-06,
      "loss": 0.398,
      "step": 25930
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6275087594985962,
      "learning_rate": 1.7354273889915473e-06,
      "loss": 0.4508,
      "step": 25931
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5354609489440918,
      "learning_rate": 1.7348540672964986e-06,
      "loss": 0.3582,
      "step": 25932
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4700502157211304,
      "learning_rate": 1.73428083132423e-06,
      "loss": 0.3939,
      "step": 25933
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5332680940628052,
      "learning_rate": 1.7337076810806842e-06,
      "loss": 0.3761,
      "step": 25934
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5479192733764648,
      "learning_rate": 1.7331346165718078e-06,
      "loss": 0.4032,
      "step": 25935
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.598672866821289,
      "learning_rate": 1.7325616378035448e-06,
      "loss": 0.3957,
      "step": 25936
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7606536149978638,
      "learning_rate": 1.7319887447818406e-06,
      "loss": 1.0077,
      "step": 25937
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4390367269515991,
      "learning_rate": 1.7314159375126338e-06,
      "loss": 0.921,
      "step": 25938
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5528982877731323,
      "learning_rate": 1.730843216001864e-06,
      "loss": 0.4305,
      "step": 25939
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4980353116989136,
      "learning_rate": 1.7302705802554732e-06,
      "loss": 0.3609,
      "step": 25940
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.534362554550171,
      "learning_rate": 1.7296980302794031e-06,
      "loss": 0.4312,
      "step": 25941
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5968501567840576,
      "learning_rate": 1.7291255660795892e-06,
      "loss": 0.3663,
      "step": 25942
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9407296180725098,
      "learning_rate": 1.7285531876619666e-06,
      "loss": 0.4468,
      "step": 25943
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5193036794662476,
      "learning_rate": 1.7279808950324728e-06,
      "loss": 0.3923,
      "step": 25944
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5958914756774902,
      "learning_rate": 1.7274086881970486e-06,
      "loss": 0.437,
      "step": 25945
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5612105131149292,
      "learning_rate": 1.7268365671616216e-06,
      "loss": 0.433,
      "step": 25946
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7005770206451416,
      "learning_rate": 1.7262645319321325e-06,
      "loss": 0.4584,
      "step": 25947
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4998880624771118,
      "learning_rate": 1.7256925825145078e-06,
      "loss": 0.3745,
      "step": 25948
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.439277172088623,
      "learning_rate": 1.7251207189146836e-06,
      "loss": 0.405,
      "step": 25949
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6427348852157593,
      "learning_rate": 1.7245489411385885e-06,
      "loss": 0.3924,
      "step": 25950
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.584486722946167,
      "learning_rate": 1.7239772491921535e-06,
      "loss": 0.4404,
      "step": 25951
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5257917642593384,
      "learning_rate": 1.7234056430813106e-06,
      "loss": 0.3838,
      "step": 25952
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6276625394821167,
      "learning_rate": 1.722834122811986e-06,
      "loss": 0.4603,
      "step": 25953
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.595842957496643,
      "learning_rate": 1.722262688390106e-06,
      "loss": 0.3707,
      "step": 25954
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.689244270324707,
      "learning_rate": 1.7216913398215984e-06,
      "loss": 0.4705,
      "step": 25955
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4754955768585205,
      "learning_rate": 1.7211200771123882e-06,
      "loss": 0.366,
      "step": 25956
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6295883655548096,
      "learning_rate": 1.720548900268405e-06,
      "loss": 0.4222,
      "step": 25957
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5274723768234253,
      "learning_rate": 1.7199778092955689e-06,
      "loss": 0.3541,
      "step": 25958
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.581386685371399,
      "learning_rate": 1.7194068041998014e-06,
      "loss": 0.464,
      "step": 25959
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5124709606170654,
      "learning_rate": 1.7188358849870258e-06,
      "loss": 0.3593,
      "step": 25960
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.734060525894165,
      "learning_rate": 1.7182650516631672e-06,
      "loss": 0.925,
      "step": 25961
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4257124662399292,
      "learning_rate": 1.717694304234141e-06,
      "loss": 0.963,
      "step": 25962
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6554213762283325,
      "learning_rate": 1.71712364270587e-06,
      "loss": 0.455,
      "step": 25963
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5566473007202148,
      "learning_rate": 1.7165530670842712e-06,
      "loss": 0.3701,
      "step": 25964
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.495300531387329,
      "learning_rate": 1.7159825773752648e-06,
      "loss": 0.3815,
      "step": 25965
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5638118982315063,
      "learning_rate": 1.7154121735847629e-06,
      "loss": 0.4144,
      "step": 25966
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5085567235946655,
      "learning_rate": 1.7148418557186852e-06,
      "loss": 0.4275,
      "step": 25967
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7889513969421387,
      "learning_rate": 1.7142716237829483e-06,
      "loss": 0.42,
      "step": 25968
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5456374883651733,
      "learning_rate": 1.7137014777834637e-06,
      "loss": 0.435,
      "step": 25969
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4208004474639893,
      "learning_rate": 1.7131314177261439e-06,
      "loss": 0.3789,
      "step": 25970
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5600059032440186,
      "learning_rate": 1.7125614436169014e-06,
      "loss": 0.4596,
      "step": 25971
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5209723711013794,
      "learning_rate": 1.7119915554616495e-06,
      "loss": 0.3858,
      "step": 25972
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5297784805297852,
      "learning_rate": 1.7114217532663013e-06,
      "loss": 0.4329,
      "step": 25973
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4610265493392944,
      "learning_rate": 1.7108520370367631e-06,
      "loss": 0.3762,
      "step": 25974
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6320794820785522,
      "learning_rate": 1.7102824067789413e-06,
      "loss": 0.413,
      "step": 25975
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.695961594581604,
      "learning_rate": 1.709712862498748e-06,
      "loss": 0.4221,
      "step": 25976
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5714969635009766,
      "learning_rate": 1.7091434042020917e-06,
      "loss": 0.4096,
      "step": 25977
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5474140644073486,
      "learning_rate": 1.708574031894873e-06,
      "loss": 0.3883,
      "step": 25978
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5237433910369873,
      "learning_rate": 1.708004745583003e-06,
      "loss": 0.4416,
      "step": 25979
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5156081914901733,
      "learning_rate": 1.70743554527238e-06,
      "loss": 0.3699,
      "step": 25980
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5817641019821167,
      "learning_rate": 1.7068664309689143e-06,
      "loss": 0.4042,
      "step": 25981
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4911304712295532,
      "learning_rate": 1.7062974026785017e-06,
      "loss": 0.354,
      "step": 25982
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.2481765747070312,
      "learning_rate": 1.7057284604070511e-06,
      "loss": 0.451,
      "step": 25983
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.445295810699463,
      "learning_rate": 1.7051596041604568e-06,
      "loss": 0.3595,
      "step": 25984
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5040836334228516,
      "learning_rate": 1.7045908339446237e-06,
      "loss": 0.3961,
      "step": 25985
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5672612190246582,
      "learning_rate": 1.7040221497654464e-06,
      "loss": 0.4608,
      "step": 25986
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.489655613899231,
      "learning_rate": 1.7034535516288242e-06,
      "loss": 0.4185,
      "step": 25987
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5312831401824951,
      "learning_rate": 1.7028850395406593e-06,
      "loss": 0.3679,
      "step": 25988
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.551505446434021,
      "learning_rate": 1.7023166135068436e-06,
      "loss": 0.4385,
      "step": 25989
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5733718872070312,
      "learning_rate": 1.7017482735332712e-06,
      "loss": 0.3926,
      "step": 25990
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.593696117401123,
      "learning_rate": 1.7011800196258387e-06,
      "loss": 0.4467,
      "step": 25991
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4736244678497314,
      "learning_rate": 1.70061185179044e-06,
      "loss": 0.367,
      "step": 25992
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5759586095809937,
      "learning_rate": 1.7000437700329707e-06,
      "loss": 0.3664,
      "step": 25993
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.780108094215393,
      "learning_rate": 1.6994757743593194e-06,
      "loss": 0.3964,
      "step": 25994
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6680833101272583,
      "learning_rate": 1.6989078647753743e-06,
      "loss": 0.4106,
      "step": 25995
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5133136510849,
      "learning_rate": 1.6983400412870298e-06,
      "loss": 0.4123,
      "step": 25996
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.676000714302063,
      "learning_rate": 1.6977723039001759e-06,
      "loss": 0.4412,
      "step": 25997
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5789872407913208,
      "learning_rate": 1.6972046526206964e-06,
      "loss": 0.3978,
      "step": 25998
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6366827487945557,
      "learning_rate": 1.6966370874544846e-06,
      "loss": 0.4208,
      "step": 25999
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5492385625839233,
      "learning_rate": 1.6960696084074225e-06,
      "loss": 0.396,
      "step": 26000
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5043269395828247,
      "learning_rate": 1.6955022154853984e-06,
      "loss": 0.4131,
      "step": 26001
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5201858282089233,
      "learning_rate": 1.6949349086942935e-06,
      "loss": 0.3892,
      "step": 26002
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5725774765014648,
      "learning_rate": 1.694367688039994e-06,
      "loss": 0.3882,
      "step": 26003
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.544866919517517,
      "learning_rate": 1.6938005535283853e-06,
      "loss": 0.377,
      "step": 26004
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.645081639289856,
      "learning_rate": 1.693233505165348e-06,
      "loss": 0.4205,
      "step": 26005
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5685433149337769,
      "learning_rate": 1.692666542956759e-06,
      "loss": 0.3747,
      "step": 26006
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5638736486434937,
      "learning_rate": 1.692099666908502e-06,
      "loss": 0.4791,
      "step": 26007
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.681546688079834,
      "learning_rate": 1.6915328770264572e-06,
      "loss": 0.3851,
      "step": 26008
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4971169233322144,
      "learning_rate": 1.690966173316505e-06,
      "loss": 0.4111,
      "step": 26009
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4992356300354004,
      "learning_rate": 1.69039955578452e-06,
      "loss": 0.3476,
      "step": 26010
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.601755142211914,
      "learning_rate": 1.689833024436377e-06,
      "loss": 0.4573,
      "step": 26011
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5231854915618896,
      "learning_rate": 1.689266579277954e-06,
      "loss": 0.3689,
      "step": 26012
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.572811484336853,
      "learning_rate": 1.6887002203151282e-06,
      "loss": 0.3738,
      "step": 26013
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.535037875175476,
      "learning_rate": 1.6881339475537694e-06,
      "loss": 0.3963,
      "step": 26014
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6311231851577759,
      "learning_rate": 1.6875677609997554e-06,
      "loss": 0.4418,
      "step": 26015
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4983069896697998,
      "learning_rate": 1.6870016606589535e-06,
      "loss": 0.3776,
      "step": 26016
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5825756788253784,
      "learning_rate": 1.68643564653724e-06,
      "loss": 0.4621,
      "step": 26017
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5427501201629639,
      "learning_rate": 1.6858697186404804e-06,
      "loss": 0.3575,
      "step": 26018
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7642453908920288,
      "learning_rate": 1.6853038769745466e-06,
      "loss": 0.9542,
      "step": 26019
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3785996437072754,
      "learning_rate": 1.6847381215453108e-06,
      "loss": 1.0114,
      "step": 26020
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6692523956298828,
      "learning_rate": 1.6841724523586345e-06,
      "loss": 0.3876,
      "step": 26021
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5562973022460938,
      "learning_rate": 1.68360686942039e-06,
      "loss": 0.3901,
      "step": 26022
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5962666273117065,
      "learning_rate": 1.6830413727364381e-06,
      "loss": 0.4145,
      "step": 26023
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5058311223983765,
      "learning_rate": 1.6824759623126496e-06,
      "loss": 0.3524,
      "step": 26024
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5530227422714233,
      "learning_rate": 1.6819106381548823e-06,
      "loss": 0.4161,
      "step": 26025
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.619312047958374,
      "learning_rate": 1.6813454002690067e-06,
      "loss": 0.3781,
      "step": 26026
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8720030784606934,
      "learning_rate": 1.6807802486608782e-06,
      "loss": 0.442,
      "step": 26027
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5959956645965576,
      "learning_rate": 1.6802151833363622e-06,
      "loss": 0.3852,
      "step": 26028
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5027406215667725,
      "learning_rate": 1.6796502043013207e-06,
      "loss": 0.3897,
      "step": 26029
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.522137999534607,
      "learning_rate": 1.6790853115616124e-06,
      "loss": 0.408,
      "step": 26030
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5906928777694702,
      "learning_rate": 1.6785205051230913e-06,
      "loss": 0.4399,
      "step": 26031
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4452242851257324,
      "learning_rate": 1.6779557849916206e-06,
      "loss": 0.3527,
      "step": 26032
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4647431373596191,
      "learning_rate": 1.6773911511730556e-06,
      "loss": 0.382,
      "step": 26033
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.511960506439209,
      "learning_rate": 1.676826603673255e-06,
      "loss": 0.3441,
      "step": 26034
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.510612964630127,
      "learning_rate": 1.676262142498072e-06,
      "loss": 0.3972,
      "step": 26035
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5158149003982544,
      "learning_rate": 1.6756977676533592e-06,
      "loss": 0.3593,
      "step": 26036
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5328048467636108,
      "learning_rate": 1.6751334791449714e-06,
      "loss": 0.4504,
      "step": 26037
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4942712783813477,
      "learning_rate": 1.6745692769787637e-06,
      "loss": 0.3743,
      "step": 26038
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5508276224136353,
      "learning_rate": 1.6740051611605834e-06,
      "loss": 0.4233,
      "step": 26039
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5875582695007324,
      "learning_rate": 1.6734411316962862e-06,
      "loss": 0.4053,
      "step": 26040
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6700838804244995,
      "learning_rate": 1.6728771885917162e-06,
      "loss": 0.4527,
      "step": 26041
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5310429334640503,
      "learning_rate": 1.6723133318527284e-06,
      "loss": 0.3614,
      "step": 26042
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5590251684188843,
      "learning_rate": 1.6717495614851654e-06,
      "loss": 0.4305,
      "step": 26043
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5021424293518066,
      "learning_rate": 1.6711858774948774e-06,
      "loss": 0.3789,
      "step": 26044
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7842626571655273,
      "learning_rate": 1.6706222798877125e-06,
      "loss": 0.9959,
      "step": 26045
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4362058639526367,
      "learning_rate": 1.6700587686695135e-06,
      "loss": 0.9947,
      "step": 26046
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6177629232406616,
      "learning_rate": 1.6694953438461237e-06,
      "loss": 0.4899,
      "step": 26047
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7641459703445435,
      "learning_rate": 1.668932005423387e-06,
      "loss": 0.4344,
      "step": 26048
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6423038244247437,
      "learning_rate": 1.668368753407149e-06,
      "loss": 0.4729,
      "step": 26049
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.468415379524231,
      "learning_rate": 1.6678055878032517e-06,
      "loss": 0.354,
      "step": 26050
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5343308448791504,
      "learning_rate": 1.6672425086175337e-06,
      "loss": 0.4021,
      "step": 26051
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4517672061920166,
      "learning_rate": 1.6666795158558347e-06,
      "loss": 0.3721,
      "step": 26052
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6861534118652344,
      "learning_rate": 1.6661166095239932e-06,
      "loss": 0.4837,
      "step": 26053
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5768579244613647,
      "learning_rate": 1.6655537896278528e-06,
      "loss": 0.374,
      "step": 26054
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5887025594711304,
      "learning_rate": 1.6649910561732442e-06,
      "loss": 0.3982,
      "step": 26055
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5094940662384033,
      "learning_rate": 1.6644284091660078e-06,
      "loss": 0.3572,
      "step": 26056
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5511528253555298,
      "learning_rate": 1.6638658486119773e-06,
      "loss": 0.4259,
      "step": 26057
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5708165168762207,
      "learning_rate": 1.6633033745169901e-06,
      "loss": 0.3777,
      "step": 26058
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6535191535949707,
      "learning_rate": 1.6627409868868748e-06,
      "loss": 0.9147,
      "step": 26059
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4231427907943726,
      "learning_rate": 1.6621786857274713e-06,
      "loss": 0.9347,
      "step": 26060
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.511415719985962,
      "learning_rate": 1.6616164710446037e-06,
      "loss": 0.4055,
      "step": 26061
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4865202903747559,
      "learning_rate": 1.6610543428441105e-06,
      "loss": 0.3898,
      "step": 26062
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4976385831832886,
      "learning_rate": 1.6604923011318152e-06,
      "loss": 0.3866,
      "step": 26063
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5189176797866821,
      "learning_rate": 1.6599303459135518e-06,
      "loss": 0.3818,
      "step": 26064
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6607630252838135,
      "learning_rate": 1.659368477195148e-06,
      "loss": 0.4744,
      "step": 26065
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.624307632446289,
      "learning_rate": 1.658806694982431e-06,
      "loss": 0.3704,
      "step": 26066
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5571714639663696,
      "learning_rate": 1.6582449992812244e-06,
      "loss": 0.4152,
      "step": 26067
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4530075788497925,
      "learning_rate": 1.6576833900973566e-06,
      "loss": 0.3606,
      "step": 26068
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6185890436172485,
      "learning_rate": 1.6571218674366518e-06,
      "loss": 0.4525,
      "step": 26069
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4957226514816284,
      "learning_rate": 1.6565604313049366e-06,
      "loss": 0.3637,
      "step": 26070
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5602601766586304,
      "learning_rate": 1.6559990817080307e-06,
      "loss": 0.4359,
      "step": 26071
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.519600749015808,
      "learning_rate": 1.655437818651756e-06,
      "loss": 0.3794,
      "step": 26072
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8398469686508179,
      "learning_rate": 1.6548766421419337e-06,
      "loss": 0.4485,
      "step": 26073
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4916881322860718,
      "learning_rate": 1.654315552184389e-06,
      "loss": 0.357,
      "step": 26074
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6129029989242554,
      "learning_rate": 1.6537545487849338e-06,
      "loss": 0.4323,
      "step": 26075
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0038414001464844,
      "learning_rate": 1.6531936319493924e-06,
      "loss": 0.3802,
      "step": 26076
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6362615823745728,
      "learning_rate": 1.652632801683579e-06,
      "loss": 0.428,
      "step": 26077
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5798765420913696,
      "learning_rate": 1.6520720579933136e-06,
      "loss": 0.3982,
      "step": 26078
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5921766757965088,
      "learning_rate": 1.6515114008844068e-06,
      "loss": 0.4665,
      "step": 26079
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4848092794418335,
      "learning_rate": 1.6509508303626776e-06,
      "loss": 0.3665,
      "step": 26080
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5107638835906982,
      "learning_rate": 1.650390346433941e-06,
      "loss": 0.4149,
      "step": 26081
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.539587140083313,
      "learning_rate": 1.6498299491040082e-06,
      "loss": 0.3593,
      "step": 26082
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6885161399841309,
      "learning_rate": 1.6492696383786887e-06,
      "loss": 0.5095,
      "step": 26083
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6095904111862183,
      "learning_rate": 1.6487094142637972e-06,
      "loss": 0.375,
      "step": 26084
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5317671298980713,
      "learning_rate": 1.648149276765144e-06,
      "loss": 0.4259,
      "step": 26085
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.567795991897583,
      "learning_rate": 1.6475892258885396e-06,
      "loss": 0.3856,
      "step": 26086
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6039570569992065,
      "learning_rate": 1.647029261639791e-06,
      "loss": 0.446,
      "step": 26087
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.527679681777954,
      "learning_rate": 1.6464693840247037e-06,
      "loss": 0.384,
      "step": 26088
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6702951192855835,
      "learning_rate": 1.6459095930490876e-06,
      "loss": 0.4617,
      "step": 26089
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5464943647384644,
      "learning_rate": 1.645349888718749e-06,
      "loss": 0.3802,
      "step": 26090
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5506471395492554,
      "learning_rate": 1.6447902710394902e-06,
      "loss": 0.418,
      "step": 26091
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6575876474380493,
      "learning_rate": 1.6442307400171198e-06,
      "loss": 0.4101,
      "step": 26092
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6131784915924072,
      "learning_rate": 1.6436712956574352e-06,
      "loss": 0.4536,
      "step": 26093
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4900599718093872,
      "learning_rate": 1.6431119379662441e-06,
      "loss": 0.3455,
      "step": 26094
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6506482362747192,
      "learning_rate": 1.642552666949343e-06,
      "loss": 0.4059,
      "step": 26095
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5307713747024536,
      "learning_rate": 1.641993482612536e-06,
      "loss": 0.4099,
      "step": 26096
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.659691572189331,
      "learning_rate": 1.6414343849616232e-06,
      "loss": 0.4979,
      "step": 26097
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5992224216461182,
      "learning_rate": 1.6408753740024018e-06,
      "loss": 0.4188,
      "step": 26098
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5725629329681396,
      "learning_rate": 1.6403164497406677e-06,
      "loss": 0.4998,
      "step": 26099
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.563242793083191,
      "learning_rate": 1.639757612182219e-06,
      "loss": 0.4296,
      "step": 26100
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5117777585983276,
      "learning_rate": 1.6391988613328557e-06,
      "loss": 0.4091,
      "step": 26101
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4755116701126099,
      "learning_rate": 1.6386401971983701e-06,
      "loss": 0.3731,
      "step": 26102
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5375467538833618,
      "learning_rate": 1.6380816197845528e-06,
      "loss": 0.4214,
      "step": 26103
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.546507716178894,
      "learning_rate": 1.6375231290972016e-06,
      "loss": 0.3716,
      "step": 26104
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.564423680305481,
      "learning_rate": 1.6369647251421071e-06,
      "loss": 0.4302,
      "step": 26105
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.493675947189331,
      "learning_rate": 1.6364064079250641e-06,
      "loss": 0.3806,
      "step": 26106
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7645039558410645,
      "learning_rate": 1.6358481774518608e-06,
      "loss": 0.3886,
      "step": 26107
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5344396829605103,
      "learning_rate": 1.635290033728284e-06,
      "loss": 0.372,
      "step": 26108
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5371522903442383,
      "learning_rate": 1.6347319767601267e-06,
      "loss": 0.3936,
      "step": 26109
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5888967514038086,
      "learning_rate": 1.6341740065531764e-06,
      "loss": 0.3846,
      "step": 26110
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6096891164779663,
      "learning_rate": 1.6336161231132174e-06,
      "loss": 0.4469,
      "step": 26111
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.520126461982727,
      "learning_rate": 1.6330583264460409e-06,
      "loss": 0.3929,
      "step": 26112
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7747490406036377,
      "learning_rate": 1.6325006165574253e-06,
      "loss": 0.9612,
      "step": 26113
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4915517568588257,
      "learning_rate": 1.6319429934531616e-06,
      "loss": 1.0136,
      "step": 26114
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.608096718788147,
      "learning_rate": 1.6313854571390286e-06,
      "loss": 0.4938,
      "step": 26115
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5055954456329346,
      "learning_rate": 1.6308280076208095e-06,
      "loss": 0.3734,
      "step": 26116
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4894535541534424,
      "learning_rate": 1.6302706449042893e-06,
      "loss": 0.4131,
      "step": 26117
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5570417642593384,
      "learning_rate": 1.629713368995247e-06,
      "loss": 0.3988,
      "step": 26118
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.560414433479309,
      "learning_rate": 1.6291561798994594e-06,
      "loss": 0.4449,
      "step": 26119
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6221950054168701,
      "learning_rate": 1.6285990776227078e-06,
      "loss": 0.3896,
      "step": 26120
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6560813188552856,
      "learning_rate": 1.6280420621707693e-06,
      "loss": 0.3983,
      "step": 26121
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4714174270629883,
      "learning_rate": 1.6274851335494256e-06,
      "loss": 0.3647,
      "step": 26122
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.603334903717041,
      "learning_rate": 1.6269282917644492e-06,
      "loss": 0.4329,
      "step": 26123
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5472660064697266,
      "learning_rate": 1.6263715368216127e-06,
      "loss": 0.3484,
      "step": 26124
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5026190280914307,
      "learning_rate": 1.6258148687266938e-06,
      "loss": 0.3986,
      "step": 26125
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5159906148910522,
      "learning_rate": 1.6252582874854684e-06,
      "loss": 0.3793,
      "step": 26126
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.520802617073059,
      "learning_rate": 1.6247017931037045e-06,
      "loss": 0.4322,
      "step": 26127
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.549291729927063,
      "learning_rate": 1.624145385587177e-06,
      "loss": 0.3753,
      "step": 26128
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5453646183013916,
      "learning_rate": 1.623589064941654e-06,
      "loss": 0.4275,
      "step": 26129
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5269546508789062,
      "learning_rate": 1.6230328311729092e-06,
      "loss": 0.3875,
      "step": 26130
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6850016117095947,
      "learning_rate": 1.6224766842867067e-06,
      "loss": 0.4849,
      "step": 26131
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5215046405792236,
      "learning_rate": 1.6219206242888174e-06,
      "loss": 0.364,
      "step": 26132
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5812114477157593,
      "learning_rate": 1.621364651185011e-06,
      "loss": 0.4224,
      "step": 26133
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4533302783966064,
      "learning_rate": 1.6208087649810512e-06,
      "loss": 0.37,
      "step": 26134
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5622167587280273,
      "learning_rate": 1.6202529656827015e-06,
      "loss": 0.4001,
      "step": 26135
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4995858669281006,
      "learning_rate": 1.6196972532957278e-06,
      "loss": 0.3623,
      "step": 26136
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4991049766540527,
      "learning_rate": 1.6191416278258965e-06,
      "loss": 0.4097,
      "step": 26137
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5867540836334229,
      "learning_rate": 1.6185860892789685e-06,
      "loss": 0.3588,
      "step": 26138
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5848731994628906,
      "learning_rate": 1.6180306376607036e-06,
      "loss": 0.4127,
      "step": 26139
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5178245306015015,
      "learning_rate": 1.6174752729768639e-06,
      "loss": 0.3763,
      "step": 26140
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4387019872665405,
      "learning_rate": 1.6169199952332103e-06,
      "loss": 0.3856,
      "step": 26141
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4741164445877075,
      "learning_rate": 1.616364804435503e-06,
      "loss": 0.3619,
      "step": 26142
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7351393699645996,
      "learning_rate": 1.6158097005895e-06,
      "loss": 1.0422,
      "step": 26143
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.393815040588379,
      "learning_rate": 1.6152546837009541e-06,
      "loss": 0.9236,
      "step": 26144
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6404855251312256,
      "learning_rate": 1.614699753775626e-06,
      "loss": 0.4271,
      "step": 26145
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5269664525985718,
      "learning_rate": 1.6141449108192719e-06,
      "loss": 0.3617,
      "step": 26146
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.675159215927124,
      "learning_rate": 1.613590154837642e-06,
      "loss": 0.4206,
      "step": 26147
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6309059858322144,
      "learning_rate": 1.613035485836496e-06,
      "loss": 0.3691,
      "step": 26148
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.608866810798645,
      "learning_rate": 1.6124809038215806e-06,
      "loss": 0.4032,
      "step": 26149
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7383079528808594,
      "learning_rate": 1.611926408798653e-06,
      "loss": 0.3537,
      "step": 26150
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5058510303497314,
      "learning_rate": 1.611372000773459e-06,
      "loss": 0.4297,
      "step": 26151
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5216165781021118,
      "learning_rate": 1.6108176797517517e-06,
      "loss": 0.3725,
      "step": 26152
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.538076400756836,
      "learning_rate": 1.6102634457392829e-06,
      "loss": 0.4097,
      "step": 26153
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6285662651062012,
      "learning_rate": 1.6097092987417972e-06,
      "loss": 0.3856,
      "step": 26154
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5071619749069214,
      "learning_rate": 1.609155238765041e-06,
      "loss": 0.4239,
      "step": 26155
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.485016942024231,
      "learning_rate": 1.608601265814763e-06,
      "loss": 0.3508,
      "step": 26156
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.599184513092041,
      "learning_rate": 1.6080473798967077e-06,
      "loss": 0.4535,
      "step": 26157
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5626057386398315,
      "learning_rate": 1.6074935810166226e-06,
      "loss": 0.4092,
      "step": 26158
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6150784492492676,
      "learning_rate": 1.6069398691802506e-06,
      "loss": 0.4657,
      "step": 26159
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4959608316421509,
      "learning_rate": 1.6063862443933298e-06,
      "loss": 0.3848,
      "step": 26160
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7284460067749023,
      "learning_rate": 1.6058327066616064e-06,
      "loss": 1.0091,
      "step": 26161
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4898931980133057,
      "learning_rate": 1.6052792559908226e-06,
      "loss": 0.9923,
      "step": 26162
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5246307849884033,
      "learning_rate": 1.6047258923867148e-06,
      "loss": 0.3932,
      "step": 26163
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4575891494750977,
      "learning_rate": 1.6041726158550263e-06,
      "loss": 0.3655,
      "step": 26164
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.4017481803894043,
      "learning_rate": 1.6036194264014914e-06,
      "loss": 0.3922,
      "step": 26165
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.719732403755188,
      "learning_rate": 1.6030663240318512e-06,
      "loss": 0.4392,
      "step": 26166
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.594996452331543,
      "learning_rate": 1.6025133087518386e-06,
      "loss": 0.402,
      "step": 26167
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5470812320709229,
      "learning_rate": 1.6019603805671912e-06,
      "loss": 0.3982,
      "step": 26168
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.614959478378296,
      "learning_rate": 1.601407539483647e-06,
      "loss": 0.4347,
      "step": 26169
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.506209373474121,
      "learning_rate": 1.6008547855069357e-06,
      "loss": 0.382,
      "step": 26170
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.530716061592102,
      "learning_rate": 1.6003021186427892e-06,
      "loss": 0.4252,
      "step": 26171
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4654040336608887,
      "learning_rate": 1.5997495388969419e-06,
      "loss": 0.3799,
      "step": 26172
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.567062258720398,
      "learning_rate": 1.5991970462751239e-06,
      "loss": 0.4438,
      "step": 26173
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5229089260101318,
      "learning_rate": 1.5986446407830703e-06,
      "loss": 0.38,
      "step": 26174
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5525970458984375,
      "learning_rate": 1.5980923224265021e-06,
      "loss": 0.4385,
      "step": 26175
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4795860052108765,
      "learning_rate": 1.5975400912111526e-06,
      "loss": 0.3638,
      "step": 26176
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7552753686904907,
      "learning_rate": 1.5969879471427474e-06,
      "loss": 0.5036,
      "step": 26177
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5255777835845947,
      "learning_rate": 1.5964358902270172e-06,
      "loss": 0.3485,
      "step": 26178
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4990381002426147,
      "learning_rate": 1.5958839204696852e-06,
      "loss": 0.439,
      "step": 26179
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4781055450439453,
      "learning_rate": 1.5953320378764725e-06,
      "loss": 0.3773,
      "step": 26180
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5250895023345947,
      "learning_rate": 1.5947802424531068e-06,
      "loss": 0.9835,
      "step": 26181
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4158767461776733,
      "learning_rate": 1.5942285342053132e-06,
      "loss": 0.9584,
      "step": 26182
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5756572484970093,
      "learning_rate": 1.5936769131388085e-06,
      "loss": 0.4435,
      "step": 26183
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.514459490776062,
      "learning_rate": 1.5931253792593192e-06,
      "loss": 0.3744,
      "step": 26184
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4913722276687622,
      "learning_rate": 1.592573932572561e-06,
      "loss": 0.3892,
      "step": 26185
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6021260023117065,
      "learning_rate": 1.5920225730842575e-06,
      "loss": 0.3966,
      "step": 26186
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5037617683410645,
      "learning_rate": 1.5914713008001215e-06,
      "loss": 0.4455,
      "step": 26187
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4564276933670044,
      "learning_rate": 1.5909201157258747e-06,
      "loss": 0.3499,
      "step": 26188
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5393768548965454,
      "learning_rate": 1.590369017867236e-06,
      "loss": 0.4122,
      "step": 26189
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.498774766921997,
      "learning_rate": 1.5898180072299174e-06,
      "loss": 0.3779,
      "step": 26190
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5407636165618896,
      "learning_rate": 1.5892670838196334e-06,
      "loss": 0.435,
      "step": 26191
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.480517864227295,
      "learning_rate": 1.588716247642098e-06,
      "loss": 0.3717,
      "step": 26192
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5883857011795044,
      "learning_rate": 1.588165498703026e-06,
      "loss": 0.4222,
      "step": 26193
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5683274269104004,
      "learning_rate": 1.5876148370081313e-06,
      "loss": 0.3768,
      "step": 26194
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5950254201889038,
      "learning_rate": 1.5870642625631227e-06,
      "loss": 0.4244,
      "step": 26195
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.487642526626587,
      "learning_rate": 1.586513775373708e-06,
      "loss": 0.3695,
      "step": 26196
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5519928932189941,
      "learning_rate": 1.5859633754456005e-06,
      "loss": 0.4314,
      "step": 26197
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5394301414489746,
      "learning_rate": 1.5854130627845088e-06,
      "loss": 0.4005,
      "step": 26198
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5587916374206543,
      "learning_rate": 1.5848628373961372e-06,
      "loss": 0.4262,
      "step": 26199
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4453134536743164,
      "learning_rate": 1.5843126992861968e-06,
      "loss": 0.373,
      "step": 26200
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5293349027633667,
      "learning_rate": 1.5837626484603885e-06,
      "loss": 0.3928,
      "step": 26201
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.446958065032959,
      "learning_rate": 1.5832126849244223e-06,
      "loss": 0.3606,
      "step": 26202
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4971261024475098,
      "learning_rate": 1.5826628086839968e-06,
      "loss": 0.4247,
      "step": 26203
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.55087411403656,
      "learning_rate": 1.5821130197448187e-06,
      "loss": 0.3884,
      "step": 26204
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5718247890472412,
      "learning_rate": 1.5815633181125922e-06,
      "loss": 0.4434,
      "step": 26205
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5269215106964111,
      "learning_rate": 1.5810137037930151e-06,
      "loss": 0.3907,
      "step": 26206
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5658540725708008,
      "learning_rate": 1.580464176791786e-06,
      "loss": 0.4176,
      "step": 26207
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5220303535461426,
      "learning_rate": 1.5799147371146073e-06,
      "loss": 0.3857,
      "step": 26208
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4975743293762207,
      "learning_rate": 1.5793653847671775e-06,
      "loss": 0.3847,
      "step": 26209
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4866406917572021,
      "learning_rate": 1.5788161197551989e-06,
      "loss": 0.3676,
      "step": 26210
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7821028232574463,
      "learning_rate": 1.5782669420843576e-06,
      "loss": 0.9026,
      "step": 26211
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5271165370941162,
      "learning_rate": 1.5777178517603552e-06,
      "loss": 1.033,
      "step": 26212
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.584977388381958,
      "learning_rate": 1.577168848788887e-06,
      "loss": 0.4435,
      "step": 26213
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8007183074951172,
      "learning_rate": 1.5766199331756493e-06,
      "loss": 0.3796,
      "step": 26214
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6336385011672974,
      "learning_rate": 1.5760711049263322e-06,
      "loss": 0.4823,
      "step": 26215
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5974546670913696,
      "learning_rate": 1.5755223640466256e-06,
      "loss": 0.3819,
      "step": 26216
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5384199619293213,
      "learning_rate": 1.5749737105422236e-06,
      "loss": 0.3923,
      "step": 26217
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4741207361221313,
      "learning_rate": 1.5744251444188197e-06,
      "loss": 0.3461,
      "step": 26218
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6260071992874146,
      "learning_rate": 1.573876665682097e-06,
      "loss": 0.4565,
      "step": 26219
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5645688772201538,
      "learning_rate": 1.5733282743377498e-06,
      "loss": 0.3913,
      "step": 26220
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.508320927619934,
      "learning_rate": 1.5727799703914615e-06,
      "loss": 0.4156,
      "step": 26221
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5323172807693481,
      "learning_rate": 1.5722317538489219e-06,
      "loss": 0.3769,
      "step": 26222
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4874241352081299,
      "learning_rate": 1.571683624715814e-06,
      "loss": 0.4322,
      "step": 26223
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5272396802902222,
      "learning_rate": 1.5711355829978247e-06,
      "loss": 0.3893,
      "step": 26224
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6488178968429565,
      "learning_rate": 1.5705876287006405e-06,
      "loss": 0.4423,
      "step": 26225
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5087695121765137,
      "learning_rate": 1.5700397618299412e-06,
      "loss": 0.4004,
      "step": 26226
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5506094694137573,
      "learning_rate": 1.5694919823914079e-06,
      "loss": 0.4082,
      "step": 26227
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5114449262619019,
      "learning_rate": 1.5689442903907237e-06,
      "loss": 0.3863,
      "step": 26228
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6077107191085815,
      "learning_rate": 1.5683966858335685e-06,
      "loss": 0.3836,
      "step": 26229
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5073667764663696,
      "learning_rate": 1.567849168725626e-06,
      "loss": 0.3826,
      "step": 26230
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6577666997909546,
      "learning_rate": 1.5673017390725709e-06,
      "loss": 0.4136,
      "step": 26231
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4845224618911743,
      "learning_rate": 1.566754396880078e-06,
      "loss": 0.369,
      "step": 26232
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6885262727737427,
      "learning_rate": 1.5662071421538284e-06,
      "loss": 0.4227,
      "step": 26233
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5274717807769775,
      "learning_rate": 1.5656599748994994e-06,
      "loss": 0.4076,
      "step": 26234
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5815564393997192,
      "learning_rate": 1.5651128951227613e-06,
      "loss": 0.4443,
      "step": 26235
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6091398000717163,
      "learning_rate": 1.5645659028292925e-06,
      "loss": 0.3766,
      "step": 26236
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.514159083366394,
      "learning_rate": 1.5640189980247622e-06,
      "loss": 0.3922,
      "step": 26237
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5664088726043701,
      "learning_rate": 1.5634721807148467e-06,
      "loss": 0.4007,
      "step": 26238
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5575863122940063,
      "learning_rate": 1.5629254509052138e-06,
      "loss": 0.3962,
      "step": 26239
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.617050290107727,
      "learning_rate": 1.5623788086015345e-06,
      "loss": 0.3683,
      "step": 26240
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.529191493988037,
      "learning_rate": 1.5618322538094832e-06,
      "loss": 0.4235,
      "step": 26241
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5008248090744019,
      "learning_rate": 1.561285786534723e-06,
      "loss": 0.3822,
      "step": 26242
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6400763988494873,
      "learning_rate": 1.5607394067829229e-06,
      "loss": 0.9074,
      "step": 26243
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3780564069747925,
      "learning_rate": 1.5601931145597483e-06,
      "loss": 1.03,
      "step": 26244
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5573668479919434,
      "learning_rate": 1.5596469098708677e-06,
      "loss": 0.4015,
      "step": 26245
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5461459159851074,
      "learning_rate": 1.5591007927219481e-06,
      "loss": 0.3823,
      "step": 26246
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5414025783538818,
      "learning_rate": 1.5585547631186514e-06,
      "loss": 0.396,
      "step": 26247
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5563116073608398,
      "learning_rate": 1.5580088210666379e-06,
      "loss": 0.3509,
      "step": 26248
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8618333339691162,
      "learning_rate": 1.5574629665715713e-06,
      "loss": 0.4555,
      "step": 26249
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5469036102294922,
      "learning_rate": 1.5569171996391174e-06,
      "loss": 0.367,
      "step": 26250
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7800556421279907,
      "learning_rate": 1.556371520274933e-06,
      "loss": 0.4706,
      "step": 26251
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4440574645996094,
      "learning_rate": 1.5558259284846755e-06,
      "loss": 0.3546,
      "step": 26252
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.501114845275879,
      "learning_rate": 1.5552804242740061e-06,
      "loss": 0.444,
      "step": 26253
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4172422885894775,
      "learning_rate": 1.5547350076485845e-06,
      "loss": 0.354,
      "step": 26254
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.580806016921997,
      "learning_rate": 1.5541896786140643e-06,
      "loss": 0.4494,
      "step": 26255
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4894636869430542,
      "learning_rate": 1.5536444371761038e-06,
      "loss": 0.3746,
      "step": 26256
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5737642049789429,
      "learning_rate": 1.5530992833403546e-06,
      "loss": 0.411,
      "step": 26257
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5897436141967773,
      "learning_rate": 1.5525542171124762e-06,
      "loss": 0.4118,
      "step": 26258
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5622563362121582,
      "learning_rate": 1.5520092384981168e-06,
      "loss": 0.4285,
      "step": 26259
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4616262912750244,
      "learning_rate": 1.551464347502929e-06,
      "loss": 0.3582,
      "step": 26260
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.506686806678772,
      "learning_rate": 1.5509195441325698e-06,
      "loss": 0.4095,
      "step": 26261
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4580830335617065,
      "learning_rate": 1.5503748283926823e-06,
      "loss": 0.3667,
      "step": 26262
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6900701522827148,
      "learning_rate": 1.5498302002889221e-06,
      "loss": 0.4832,
      "step": 26263
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5024099349975586,
      "learning_rate": 1.5492856598269323e-06,
      "loss": 0.3747,
      "step": 26264
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5135401487350464,
      "learning_rate": 1.5487412070123641e-06,
      "loss": 0.4046,
      "step": 26265
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.595206379890442,
      "learning_rate": 1.5481968418508663e-06,
      "loss": 0.376,
      "step": 26266
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6110433340072632,
      "learning_rate": 1.5476525643480822e-06,
      "loss": 0.4498,
      "step": 26267
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5281062126159668,
      "learning_rate": 1.547108374509655e-06,
      "loss": 0.3788,
      "step": 26268
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4883501529693604,
      "learning_rate": 1.54656427234123e-06,
      "loss": 0.4079,
      "step": 26269
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.511756181716919,
      "learning_rate": 1.5460202578484517e-06,
      "loss": 0.3635,
      "step": 26270
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4981937408447266,
      "learning_rate": 1.5454763310369648e-06,
      "loss": 0.413,
      "step": 26271
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5653105974197388,
      "learning_rate": 1.5449324919124076e-06,
      "loss": 0.3949,
      "step": 26272
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6169180870056152,
      "learning_rate": 1.544388740480418e-06,
      "loss": 0.425,
      "step": 26273
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.525829792022705,
      "learning_rate": 1.5438450767466383e-06,
      "loss": 0.3983,
      "step": 26274
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6234649419784546,
      "learning_rate": 1.5433015007167106e-06,
      "loss": 0.4271,
      "step": 26275
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.566154956817627,
      "learning_rate": 1.542758012396266e-06,
      "loss": 0.3924,
      "step": 26276
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4580891132354736,
      "learning_rate": 1.5422146117909464e-06,
      "loss": 0.4321,
      "step": 26277
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5766586065292358,
      "learning_rate": 1.5416712989063852e-06,
      "loss": 0.3726,
      "step": 26278
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.601068377494812,
      "learning_rate": 1.5411280737482192e-06,
      "loss": 0.4591,
      "step": 26279
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5389832258224487,
      "learning_rate": 1.5405849363220804e-06,
      "loss": 0.3734,
      "step": 26280
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4864293336868286,
      "learning_rate": 1.540041886633602e-06,
      "loss": 0.3897,
      "step": 26281
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.544785499572754,
      "learning_rate": 1.5394989246884196e-06,
      "loss": 0.382,
      "step": 26282
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.508384108543396,
      "learning_rate": 1.538956050492163e-06,
      "loss": 0.4098,
      "step": 26283
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5685651302337646,
      "learning_rate": 1.53841326405046e-06,
      "loss": 0.3751,
      "step": 26284
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5926015377044678,
      "learning_rate": 1.537870565368942e-06,
      "loss": 0.4964,
      "step": 26285
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5290025472640991,
      "learning_rate": 1.5373279544532372e-06,
      "loss": 0.3606,
      "step": 26286
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5937086343765259,
      "learning_rate": 1.536785431308977e-06,
      "loss": 0.4525,
      "step": 26287
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5005733966827393,
      "learning_rate": 1.5362429959417846e-06,
      "loss": 0.3508,
      "step": 26288
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.58814537525177,
      "learning_rate": 1.5357006483572857e-06,
      "loss": 0.4682,
      "step": 26289
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9019814729690552,
      "learning_rate": 1.5351583885611042e-06,
      "loss": 0.4237,
      "step": 26290
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.532372236251831,
      "learning_rate": 1.5346162165588696e-06,
      "loss": 0.39,
      "step": 26291
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5667656660079956,
      "learning_rate": 1.5340741323562013e-06,
      "loss": 0.3812,
      "step": 26292
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.482767105102539,
      "learning_rate": 1.5335321359587208e-06,
      "loss": 0.4025,
      "step": 26293
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5206685066223145,
      "learning_rate": 1.5329902273720486e-06,
      "loss": 0.3853,
      "step": 26294
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6199177503585815,
      "learning_rate": 1.5324484066018107e-06,
      "loss": 0.4705,
      "step": 26295
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4693026542663574,
      "learning_rate": 1.5319066736536215e-06,
      "loss": 0.3767,
      "step": 26296
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4832710027694702,
      "learning_rate": 1.5313650285331027e-06,
      "loss": 0.3947,
      "step": 26297
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5030837059020996,
      "learning_rate": 1.530823471245868e-06,
      "loss": 0.3531,
      "step": 26298
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5201482772827148,
      "learning_rate": 1.5302820017975396e-06,
      "loss": 0.4279,
      "step": 26299
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4646995067596436,
      "learning_rate": 1.5297406201937282e-06,
      "loss": 0.3746,
      "step": 26300
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7962098121643066,
      "learning_rate": 1.5291993264400506e-06,
      "loss": 0.9896,
      "step": 26301
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3960516452789307,
      "learning_rate": 1.5286581205421248e-06,
      "loss": 1.0244,
      "step": 26302
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6353397369384766,
      "learning_rate": 1.5281170025055604e-06,
      "loss": 0.4225,
      "step": 26303
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5174833536148071,
      "learning_rate": 1.5275759723359672e-06,
      "loss": 0.371,
      "step": 26304
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.504690408706665,
      "learning_rate": 1.5270350300389592e-06,
      "loss": 0.4272,
      "step": 26305
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.531646966934204,
      "learning_rate": 1.5264941756201478e-06,
      "loss": 0.3763,
      "step": 26306
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.510343313217163,
      "learning_rate": 1.5259534090851424e-06,
      "loss": 0.4033,
      "step": 26307
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5006152391433716,
      "learning_rate": 1.5254127304395528e-06,
      "loss": 0.3723,
      "step": 26308
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.529723882675171,
      "learning_rate": 1.5248721396889809e-06,
      "loss": 0.4025,
      "step": 26309
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6082321405410767,
      "learning_rate": 1.5243316368390382e-06,
      "loss": 0.3748,
      "step": 26310
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6931124925613403,
      "learning_rate": 1.523791221895331e-06,
      "loss": 0.4783,
      "step": 26311
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5649476051330566,
      "learning_rate": 1.5232508948634616e-06,
      "loss": 0.3972,
      "step": 26312
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5394216775894165,
      "learning_rate": 1.5227106557490368e-06,
      "loss": 0.4006,
      "step": 26313
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6014484167099,
      "learning_rate": 1.5221705045576574e-06,
      "loss": 0.3751,
      "step": 26314
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.573630690574646,
      "learning_rate": 1.521630441294928e-06,
      "loss": 0.4271,
      "step": 26315
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5006099939346313,
      "learning_rate": 1.5210904659664461e-06,
      "loss": 0.3752,
      "step": 26316
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.572292685508728,
      "learning_rate": 1.5205505785778151e-06,
      "loss": 0.4632,
      "step": 26317
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.493870496749878,
      "learning_rate": 1.5200107791346353e-06,
      "loss": 0.376,
      "step": 26318
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5778613090515137,
      "learning_rate": 1.519471067642505e-06,
      "loss": 0.4201,
      "step": 26319
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4882646799087524,
      "learning_rate": 1.518931444107018e-06,
      "loss": 0.3974,
      "step": 26320
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.725759506225586,
      "learning_rate": 1.518391908533774e-06,
      "loss": 0.4302,
      "step": 26321
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5149539709091187,
      "learning_rate": 1.5178524609283685e-06,
      "loss": 0.3771,
      "step": 26322
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5474255084991455,
      "learning_rate": 1.5173131012963993e-06,
      "loss": 0.3801,
      "step": 26323
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5796059370040894,
      "learning_rate": 1.5167738296434575e-06,
      "loss": 0.4429,
      "step": 26324
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.64850914478302,
      "learning_rate": 1.5162346459751331e-06,
      "loss": 0.4404,
      "step": 26325
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5462833642959595,
      "learning_rate": 1.5156955502970228e-06,
      "loss": 0.3628,
      "step": 26326
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6402487754821777,
      "learning_rate": 1.5151565426147196e-06,
      "loss": 0.4248,
      "step": 26327
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6085317134857178,
      "learning_rate": 1.5146176229338094e-06,
      "loss": 0.419,
      "step": 26328
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7103182077407837,
      "learning_rate": 1.514078791259882e-06,
      "loss": 0.4239,
      "step": 26329
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6237003803253174,
      "learning_rate": 1.5135400475985273e-06,
      "loss": 0.4091,
      "step": 26330
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5815365314483643,
      "learning_rate": 1.5130013919553355e-06,
      "loss": 0.4374,
      "step": 26331
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6291050910949707,
      "learning_rate": 1.5124628243358875e-06,
      "loss": 0.393,
      "step": 26332
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5933815240859985,
      "learning_rate": 1.5119243447457743e-06,
      "loss": 0.4106,
      "step": 26333
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.550155758857727,
      "learning_rate": 1.5113859531905771e-06,
      "loss": 0.3663,
      "step": 26334
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5379139184951782,
      "learning_rate": 1.510847649675884e-06,
      "loss": 0.4317,
      "step": 26335
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4835619926452637,
      "learning_rate": 1.5103094342072733e-06,
      "loss": 0.3926,
      "step": 26336
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5413813591003418,
      "learning_rate": 1.5097713067903286e-06,
      "loss": 0.4834,
      "step": 26337
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5036251544952393,
      "learning_rate": 1.5092332674306353e-06,
      "loss": 0.3558,
      "step": 26338
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.541678547859192,
      "learning_rate": 1.5086953161337702e-06,
      "loss": 0.4158,
      "step": 26339
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.500242829322815,
      "learning_rate": 1.50815745290531e-06,
      "loss": 0.3882,
      "step": 26340
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.610535979270935,
      "learning_rate": 1.5076196777508368e-06,
      "loss": 0.4258,
      "step": 26341
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.662101149559021,
      "learning_rate": 1.507081990675928e-06,
      "loss": 0.441,
      "step": 26342
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6066391468048096,
      "learning_rate": 1.5065443916861612e-06,
      "loss": 0.4126,
      "step": 26343
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.635361909866333,
      "learning_rate": 1.506006880787111e-06,
      "loss": 0.3888,
      "step": 26344
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6198033094406128,
      "learning_rate": 1.5054694579843498e-06,
      "loss": 0.4137,
      "step": 26345
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0942471027374268,
      "learning_rate": 1.5049321232834536e-06,
      "loss": 0.3593,
      "step": 26346
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5603679418563843,
      "learning_rate": 1.5043948766899985e-06,
      "loss": 0.3968,
      "step": 26347
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4650486707687378,
      "learning_rate": 1.5038577182095516e-06,
      "loss": 0.3594,
      "step": 26348
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.464900255203247,
      "learning_rate": 1.503320647847687e-06,
      "loss": 0.3929,
      "step": 26349
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.508508563041687,
      "learning_rate": 1.5027836656099725e-06,
      "loss": 0.3868,
      "step": 26350
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5364692211151123,
      "learning_rate": 1.5022467715019828e-06,
      "loss": 0.459,
      "step": 26351
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.558023452758789,
      "learning_rate": 1.501709965529279e-06,
      "loss": 0.3631,
      "step": 26352
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5404834747314453,
      "learning_rate": 1.5011732476974327e-06,
      "loss": 0.4283,
      "step": 26353
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5182933807373047,
      "learning_rate": 1.5006366180120123e-06,
      "loss": 0.3993,
      "step": 26354
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5222606658935547,
      "learning_rate": 1.5001000764785811e-06,
      "loss": 0.4147,
      "step": 26355
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.507604718208313,
      "learning_rate": 1.4995636231027022e-06,
      "loss": 0.3571,
      "step": 26356
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.52959144115448,
      "learning_rate": 1.4990272578899411e-06,
      "loss": 0.3883,
      "step": 26357
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5068838596343994,
      "learning_rate": 1.4984909808458613e-06,
      "loss": 0.3532,
      "step": 26358
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5303202867507935,
      "learning_rate": 1.4979547919760273e-06,
      "loss": 0.3828,
      "step": 26359
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6873401403427124,
      "learning_rate": 1.4974186912859966e-06,
      "loss": 0.4206,
      "step": 26360
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5196646451950073,
      "learning_rate": 1.4968826787813285e-06,
      "loss": 0.4002,
      "step": 26361
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6026026010513306,
      "learning_rate": 1.4963467544675848e-06,
      "loss": 0.3662,
      "step": 26362
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5543677806854248,
      "learning_rate": 1.4958109183503245e-06,
      "loss": 0.4625,
      "step": 26363
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4910404682159424,
      "learning_rate": 1.495275170435102e-06,
      "loss": 0.363,
      "step": 26364
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5524978637695312,
      "learning_rate": 1.4947395107274787e-06,
      "loss": 0.3902,
      "step": 26365
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5247218608856201,
      "learning_rate": 1.4942039392330044e-06,
      "loss": 0.3569,
      "step": 26366
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8970788717269897,
      "learning_rate": 1.493668455957239e-06,
      "loss": 0.4385,
      "step": 26367
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5164166688919067,
      "learning_rate": 1.4931330609057326e-06,
      "loss": 0.3981,
      "step": 26368
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6275008916854858,
      "learning_rate": 1.492597754084041e-06,
      "loss": 0.386,
      "step": 26369
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5543426275253296,
      "learning_rate": 1.4920625354977126e-06,
      "loss": 0.4171,
      "step": 26370
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6040971279144287,
      "learning_rate": 1.491527405152302e-06,
      "loss": 0.4445,
      "step": 26371
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6212544441223145,
      "learning_rate": 1.4909923630533562e-06,
      "loss": 0.4248,
      "step": 26372
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6643494367599487,
      "learning_rate": 1.4904574092064272e-06,
      "loss": 0.4349,
      "step": 26373
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6199181079864502,
      "learning_rate": 1.489922543617064e-06,
      "loss": 0.4169,
      "step": 26374
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5644673109054565,
      "learning_rate": 1.4893877662908118e-06,
      "loss": 0.457,
      "step": 26375
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5491080284118652,
      "learning_rate": 1.4888530772332155e-06,
      "loss": 0.365,
      "step": 26376
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.475934386253357,
      "learning_rate": 1.4883184764498227e-06,
      "loss": 0.4142,
      "step": 26377
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.513219952583313,
      "learning_rate": 1.487783963946179e-06,
      "loss": 0.3528,
      "step": 26378
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.559375524520874,
      "learning_rate": 1.4872495397278287e-06,
      "loss": 0.4242,
      "step": 26379
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5124694108963013,
      "learning_rate": 1.486715203800314e-06,
      "loss": 0.3637,
      "step": 26380
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6940174102783203,
      "learning_rate": 1.486180956169173e-06,
      "loss": 0.432,
      "step": 26381
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6201250553131104,
      "learning_rate": 1.4856467968399502e-06,
      "loss": 0.3845,
      "step": 26382
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5197467803955078,
      "learning_rate": 1.4851127258181875e-06,
      "loss": 0.4037,
      "step": 26383
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5555493831634521,
      "learning_rate": 1.4845787431094193e-06,
      "loss": 0.3869,
      "step": 26384
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6044384241104126,
      "learning_rate": 1.4840448487191894e-06,
      "loss": 0.4284,
      "step": 26385
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.512224555015564,
      "learning_rate": 1.4835110426530286e-06,
      "loss": 0.3816,
      "step": 26386
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.827144742012024,
      "learning_rate": 1.4829773249164803e-06,
      "loss": 0.938,
      "step": 26387
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4179292917251587,
      "learning_rate": 1.4824436955150733e-06,
      "loss": 1.0332,
      "step": 26388
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5777788162231445,
      "learning_rate": 1.481910154454347e-06,
      "loss": 0.4221,
      "step": 26389
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.533559799194336,
      "learning_rate": 1.481376701739834e-06,
      "loss": 0.3866,
      "step": 26390
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5050721168518066,
      "learning_rate": 1.480843337377068e-06,
      "loss": 0.4263,
      "step": 26391
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.535062551498413,
      "learning_rate": 1.480310061371576e-06,
      "loss": 0.3832,
      "step": 26392
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5635335445404053,
      "learning_rate": 1.479776873728893e-06,
      "loss": 0.394,
      "step": 26393
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5598578453063965,
      "learning_rate": 1.4792437744545485e-06,
      "loss": 0.3799,
      "step": 26394
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5575792789459229,
      "learning_rate": 1.4787107635540732e-06,
      "loss": 0.4136,
      "step": 26395
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5321695804595947,
      "learning_rate": 1.478177841032994e-06,
      "loss": 0.3832,
      "step": 26396
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.48542320728302,
      "learning_rate": 1.477645006896834e-06,
      "loss": 0.368,
      "step": 26397
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5321998596191406,
      "learning_rate": 1.4771122611511246e-06,
      "loss": 0.3792,
      "step": 26398
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6632448434829712,
      "learning_rate": 1.4765796038013925e-06,
      "loss": 0.4198,
      "step": 26399
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5265071392059326,
      "learning_rate": 1.4760470348531564e-06,
      "loss": 0.3645,
      "step": 26400
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5894895792007446,
      "learning_rate": 1.4755145543119453e-06,
      "loss": 0.452,
      "step": 26401
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9273085594177246,
      "learning_rate": 1.4749821621832771e-06,
      "loss": 0.3816,
      "step": 26402
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7259790897369385,
      "learning_rate": 1.4744498584726796e-06,
      "loss": 0.4623,
      "step": 26403
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5129446983337402,
      "learning_rate": 1.4739176431856662e-06,
      "loss": 0.3875,
      "step": 26404
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4590067863464355,
      "learning_rate": 1.473385516327762e-06,
      "loss": 0.4148,
      "step": 26405
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5611686706542969,
      "learning_rate": 1.4728534779044868e-06,
      "loss": 0.3911,
      "step": 26406
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.584293007850647,
      "learning_rate": 1.4723215279213566e-06,
      "loss": 0.4042,
      "step": 26407
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.445397138595581,
      "learning_rate": 1.471789666383886e-06,
      "loss": 0.3521,
      "step": 26408
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4998215436935425,
      "learning_rate": 1.471257893297594e-06,
      "loss": 0.4451,
      "step": 26409
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5984629392623901,
      "learning_rate": 1.4707262086679974e-06,
      "loss": 0.3978,
      "step": 26410
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6064646244049072,
      "learning_rate": 1.4701946125006095e-06,
      "loss": 0.4816,
      "step": 26411
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6793276071548462,
      "learning_rate": 1.4696631048009413e-06,
      "loss": 0.389,
      "step": 26412
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.633022427558899,
      "learning_rate": 1.4691316855745074e-06,
      "loss": 0.4791,
      "step": 26413
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.549723505973816,
      "learning_rate": 1.4686003548268202e-06,
      "loss": 0.3865,
      "step": 26414
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5402710437774658,
      "learning_rate": 1.4680691125633906e-06,
      "loss": 0.4014,
      "step": 26415
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5793068408966064,
      "learning_rate": 1.4675379587897288e-06,
      "loss": 0.4037,
      "step": 26416
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5426732301712036,
      "learning_rate": 1.467006893511339e-06,
      "loss": 0.4244,
      "step": 26417
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5592550039291382,
      "learning_rate": 1.4664759167337327e-06,
      "loss": 0.3892,
      "step": 26418
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.602445363998413,
      "learning_rate": 1.46594502846242e-06,
      "loss": 0.4585,
      "step": 26419
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5419670343399048,
      "learning_rate": 1.4654142287029015e-06,
      "loss": 0.3894,
      "step": 26420
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5953477621078491,
      "learning_rate": 1.4648835174606879e-06,
      "loss": 0.4436,
      "step": 26421
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6353013515472412,
      "learning_rate": 1.4643528947412766e-06,
      "loss": 0.3931,
      "step": 26422
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5558313131332397,
      "learning_rate": 1.463822360550179e-06,
      "loss": 0.4167,
      "step": 26423
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.527948021888733,
      "learning_rate": 1.4632919148928904e-06,
      "loss": 0.3861,
      "step": 26424
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5073344707489014,
      "learning_rate": 1.4627615577749155e-06,
      "loss": 0.3755,
      "step": 26425
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5064516067504883,
      "learning_rate": 1.4622312892017576e-06,
      "loss": 0.4029,
      "step": 26426
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5184699296951294,
      "learning_rate": 1.4617011091789135e-06,
      "loss": 0.3982,
      "step": 26427
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5631898641586304,
      "learning_rate": 1.4611710177118798e-06,
      "loss": 0.4161,
      "step": 26428
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.546438455581665,
      "learning_rate": 1.4606410148061578e-06,
      "loss": 0.5306,
      "step": 26429
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5554044246673584,
      "learning_rate": 1.460111100467243e-06,
      "loss": 0.3685,
      "step": 26430
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7301511764526367,
      "learning_rate": 1.4595812747006343e-06,
      "loss": 0.4032,
      "step": 26431
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.537368893623352,
      "learning_rate": 1.459051537511824e-06,
      "loss": 0.3699,
      "step": 26432
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5949252843856812,
      "learning_rate": 1.4585218889063056e-06,
      "loss": 0.4094,
      "step": 26433
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5132824182510376,
      "learning_rate": 1.4579923288895737e-06,
      "loss": 0.3522,
      "step": 26434
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5559468269348145,
      "learning_rate": 1.4574628574671224e-06,
      "loss": 0.3976,
      "step": 26435
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.456744909286499,
      "learning_rate": 1.4569334746444397e-06,
      "loss": 0.3552,
      "step": 26436
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4727734327316284,
      "learning_rate": 1.45640418042702e-06,
      "loss": 0.4027,
      "step": 26437
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4645521640777588,
      "learning_rate": 1.4558749748203493e-06,
      "loss": 0.3914,
      "step": 26438
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6485607624053955,
      "learning_rate": 1.4553458578299196e-06,
      "loss": 0.4878,
      "step": 26439
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5168511867523193,
      "learning_rate": 1.454816829461214e-06,
      "loss": 0.3732,
      "step": 26440
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5301698446273804,
      "learning_rate": 1.4542878897197233e-06,
      "loss": 0.3974,
      "step": 26441
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4724407196044922,
      "learning_rate": 1.4537590386109346e-06,
      "loss": 0.376,
      "step": 26442
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.544463038444519,
      "learning_rate": 1.4532302761403305e-06,
      "loss": 0.4006,
      "step": 26443
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.577321171760559,
      "learning_rate": 1.4527016023133922e-06,
      "loss": 0.4596,
      "step": 26444
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.56645667552948,
      "learning_rate": 1.4521730171356075e-06,
      "loss": 0.4245,
      "step": 26445
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4847992658615112,
      "learning_rate": 1.4516445206124596e-06,
      "loss": 0.3554,
      "step": 26446
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.513763189315796,
      "learning_rate": 1.4511161127494266e-06,
      "loss": 0.4187,
      "step": 26447
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5353822708129883,
      "learning_rate": 1.4505877935519885e-06,
      "loss": 0.3963,
      "step": 26448
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.472008466720581,
      "learning_rate": 1.4500595630256253e-06,
      "loss": 0.364,
      "step": 26449
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5993021726608276,
      "learning_rate": 1.4495314211758171e-06,
      "loss": 0.3694,
      "step": 26450
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7714406251907349,
      "learning_rate": 1.449003368008043e-06,
      "loss": 0.4831,
      "step": 26451
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.502455472946167,
      "learning_rate": 1.4484754035277782e-06,
      "loss": 0.3768,
      "step": 26452
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5162289142608643,
      "learning_rate": 1.4479475277404953e-06,
      "loss": 0.4183,
      "step": 26453
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.58359694480896,
      "learning_rate": 1.447419740651671e-06,
      "loss": 0.3936,
      "step": 26454
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5910956859588623,
      "learning_rate": 1.4468920422667843e-06,
      "loss": 0.4431,
      "step": 26455
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5707539319992065,
      "learning_rate": 1.4463644325913008e-06,
      "loss": 0.381,
      "step": 26456
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5311594009399414,
      "learning_rate": 1.4458369116306992e-06,
      "loss": 0.4118,
      "step": 26457
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.530652403831482,
      "learning_rate": 1.4453094793904442e-06,
      "loss": 0.3652,
      "step": 26458
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.570075511932373,
      "learning_rate": 1.4447821358760127e-06,
      "loss": 0.4528,
      "step": 26459
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5091191530227661,
      "learning_rate": 1.4442548810928691e-06,
      "loss": 0.3747,
      "step": 26460
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5756947994232178,
      "learning_rate": 1.4437277150464823e-06,
      "loss": 0.4269,
      "step": 26461
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5264606475830078,
      "learning_rate": 1.4432006377423247e-06,
      "loss": 0.4029,
      "step": 26462
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5986908674240112,
      "learning_rate": 1.4426736491858596e-06,
      "loss": 0.4385,
      "step": 26463
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.47990083694458,
      "learning_rate": 1.442146749382549e-06,
      "loss": 0.3764,
      "step": 26464
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6304197311401367,
      "learning_rate": 1.4416199383378626e-06,
      "loss": 0.4026,
      "step": 26465
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5085370540618896,
      "learning_rate": 1.4410932160572622e-06,
      "loss": 0.3648,
      "step": 26466
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4775062799453735,
      "learning_rate": 1.4405665825462144e-06,
      "loss": 0.4127,
      "step": 26467
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.460081934928894,
      "learning_rate": 1.4400400378101786e-06,
      "loss": 0.3752,
      "step": 26468
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5683594942092896,
      "learning_rate": 1.4395135818546124e-06,
      "loss": 0.4178,
      "step": 26469
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.492183804512024,
      "learning_rate": 1.4389872146849804e-06,
      "loss": 0.3704,
      "step": 26470
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6077696084976196,
      "learning_rate": 1.4384609363067426e-06,
      "loss": 0.4451,
      "step": 26471
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.50552499294281,
      "learning_rate": 1.4379347467253523e-06,
      "loss": 0.3799,
      "step": 26472
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5695127248764038,
      "learning_rate": 1.4374086459462734e-06,
      "loss": 0.4375,
      "step": 26473
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.686092734336853,
      "learning_rate": 1.4368826339749565e-06,
      "loss": 0.3822,
      "step": 26474
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.490457534790039,
      "learning_rate": 1.4363567108168619e-06,
      "loss": 0.3731,
      "step": 26475
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.494650959968567,
      "learning_rate": 1.435830876477441e-06,
      "loss": 0.3667,
      "step": 26476
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5150567293167114,
      "learning_rate": 1.4353051309621469e-06,
      "loss": 0.3895,
      "step": 26477
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0655407905578613,
      "learning_rate": 1.4347794742764376e-06,
      "loss": 0.4151,
      "step": 26478
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5584497451782227,
      "learning_rate": 1.4342539064257622e-06,
      "loss": 0.4253,
      "step": 26479
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5224838256835938,
      "learning_rate": 1.4337284274155684e-06,
      "loss": 0.3507,
      "step": 26480
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6985595226287842,
      "learning_rate": 1.4332030372513083e-06,
      "loss": 0.3976,
      "step": 26481
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5925410985946655,
      "learning_rate": 1.4326777359384315e-06,
      "loss": 0.3949,
      "step": 26482
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5866100788116455,
      "learning_rate": 1.432152523482392e-06,
      "loss": 0.4927,
      "step": 26483
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.578260064125061,
      "learning_rate": 1.4316273998886265e-06,
      "loss": 0.4096,
      "step": 26484
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.557691216468811,
      "learning_rate": 1.4311023651625867e-06,
      "loss": 0.3985,
      "step": 26485
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4669575691223145,
      "learning_rate": 1.4305774193097176e-06,
      "loss": 0.3638,
      "step": 26486
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5192043781280518,
      "learning_rate": 1.4300525623354654e-06,
      "loss": 0.3958,
      "step": 26487
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5396465063095093,
      "learning_rate": 1.4295277942452735e-06,
      "loss": 0.3781,
      "step": 26488
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5151166915893555,
      "learning_rate": 1.4290031150445793e-06,
      "loss": 0.4066,
      "step": 26489
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.577062964439392,
      "learning_rate": 1.42847852473883e-06,
      "loss": 0.3876,
      "step": 26490
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5411596298217773,
      "learning_rate": 1.4279540233334666e-06,
      "loss": 0.4255,
      "step": 26491
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.501283049583435,
      "learning_rate": 1.4274296108339248e-06,
      "loss": 0.3783,
      "step": 26492
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.544122576713562,
      "learning_rate": 1.426905287245649e-06,
      "loss": 0.4232,
      "step": 26493
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6064530611038208,
      "learning_rate": 1.4263810525740718e-06,
      "loss": 0.3739,
      "step": 26494
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5948760509490967,
      "learning_rate": 1.4258569068246365e-06,
      "loss": 0.4473,
      "step": 26495
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5465847253799438,
      "learning_rate": 1.4253328500027718e-06,
      "loss": 0.3872,
      "step": 26496
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4417799711227417,
      "learning_rate": 1.4248088821139182e-06,
      "loss": 0.3809,
      "step": 26497
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4953317642211914,
      "learning_rate": 1.4242850031635102e-06,
      "loss": 0.3716,
      "step": 26498
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4815068244934082,
      "learning_rate": 1.423761213156981e-06,
      "loss": 0.3928,
      "step": 26499
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5691176652908325,
      "learning_rate": 1.4232375120997588e-06,
      "loss": 0.4164,
      "step": 26500
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8180954456329346,
      "learning_rate": 1.4227138999972801e-06,
      "loss": 0.9556,
      "step": 26501
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4743520021438599,
      "learning_rate": 1.4221903768549728e-06,
      "loss": 1.0009,
      "step": 26502
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.47048819065094,
      "learning_rate": 1.4216669426782703e-06,
      "loss": 0.4262,
      "step": 26503
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4921255111694336,
      "learning_rate": 1.4211435974725996e-06,
      "loss": 0.4006,
      "step": 26504
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6727010011672974,
      "learning_rate": 1.420620341243385e-06,
      "loss": 0.4222,
      "step": 26505
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6849424839019775,
      "learning_rate": 1.4200971739960567e-06,
      "loss": 0.3991,
      "step": 26506
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5058720111846924,
      "learning_rate": 1.4195740957360416e-06,
      "loss": 0.4466,
      "step": 26507
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5906727313995361,
      "learning_rate": 1.4190511064687652e-06,
      "loss": 0.3995,
      "step": 26508
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5464838743209839,
      "learning_rate": 1.4185282061996508e-06,
      "loss": 0.4084,
      "step": 26509
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.542188048362732,
      "learning_rate": 1.4180053949341188e-06,
      "loss": 0.3807,
      "step": 26510
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7869986295700073,
      "learning_rate": 1.4174826726775948e-06,
      "loss": 0.9636,
      "step": 26511
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4940953254699707,
      "learning_rate": 1.416960039435501e-06,
      "loss": 0.9901,
      "step": 26512
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5595033168792725,
      "learning_rate": 1.4164374952132554e-06,
      "loss": 0.3965,
      "step": 26513
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5829381942749023,
      "learning_rate": 1.4159150400162802e-06,
      "loss": 0.3497,
      "step": 26514
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5334206819534302,
      "learning_rate": 1.4153926738499912e-06,
      "loss": 0.3646,
      "step": 26515
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5261015892028809,
      "learning_rate": 1.4148703967198096e-06,
      "loss": 0.3589,
      "step": 26516
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5772159099578857,
      "learning_rate": 1.4143482086311477e-06,
      "loss": 0.4222,
      "step": 26517
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6095207929611206,
      "learning_rate": 1.4138261095894234e-06,
      "loss": 0.381,
      "step": 26518
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.520835041999817,
      "learning_rate": 1.4133040996000557e-06,
      "loss": 0.3908,
      "step": 26519
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5145469903945923,
      "learning_rate": 1.412782178668456e-06,
      "loss": 0.3867,
      "step": 26520
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6713203191757202,
      "learning_rate": 1.412260346800033e-06,
      "loss": 0.4263,
      "step": 26521
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5104936361312866,
      "learning_rate": 1.4117386040002035e-06,
      "loss": 0.3582,
      "step": 26522
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8248720169067383,
      "learning_rate": 1.41121695027438e-06,
      "loss": 0.471,
      "step": 26523
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5610435009002686,
      "learning_rate": 1.4106953856279703e-06,
      "loss": 0.3692,
      "step": 26524
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.4096031188964844,
      "learning_rate": 1.4101739100663825e-06,
      "loss": 0.4174,
      "step": 26525
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5580462217330933,
      "learning_rate": 1.4096525235950276e-06,
      "loss": 0.4303,
      "step": 26526
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7161099910736084,
      "learning_rate": 1.4091312262193125e-06,
      "loss": 0.4023,
      "step": 26527
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4407775402069092,
      "learning_rate": 1.408610017944646e-06,
      "loss": 0.3667,
      "step": 26528
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5312219858169556,
      "learning_rate": 1.408088898776433e-06,
      "loss": 0.4275,
      "step": 26529
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.648169755935669,
      "learning_rate": 1.4075678687200733e-06,
      "loss": 0.3799,
      "step": 26530
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7998740673065186,
      "learning_rate": 1.4070469277809761e-06,
      "loss": 0.5112,
      "step": 26531
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.584183931350708,
      "learning_rate": 1.4065260759645448e-06,
      "loss": 0.3982,
      "step": 26532
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9217817783355713,
      "learning_rate": 1.4060053132761775e-06,
      "loss": 0.9775,
      "step": 26533
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3816925287246704,
      "learning_rate": 1.4054846397212807e-06,
      "loss": 1.1011,
      "step": 26534
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.593621850013733,
      "learning_rate": 1.4049640553052479e-06,
      "loss": 0.4149,
      "step": 26535
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5257142782211304,
      "learning_rate": 1.404443560033485e-06,
      "loss": 0.3792,
      "step": 26536
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5468463897705078,
      "learning_rate": 1.403923153911385e-06,
      "loss": 0.411,
      "step": 26537
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6600497961044312,
      "learning_rate": 1.4034028369443487e-06,
      "loss": 0.3678,
      "step": 26538
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4745404720306396,
      "learning_rate": 1.4028826091377723e-06,
      "loss": 0.4068,
      "step": 26539
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5604192018508911,
      "learning_rate": 1.4023624704970518e-06,
      "loss": 0.3782,
      "step": 26540
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7874149084091187,
      "learning_rate": 1.4018424210275782e-06,
      "loss": 0.4366,
      "step": 26541
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5474063158035278,
      "learning_rate": 1.4013224607347475e-06,
      "loss": 0.4002,
      "step": 26542
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7423306703567505,
      "learning_rate": 1.400802589623953e-06,
      "loss": 0.9673,
      "step": 26543
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.356557011604309,
      "learning_rate": 1.4002828077005892e-06,
      "loss": 0.917,
      "step": 26544
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7086210250854492,
      "learning_rate": 1.399763114970044e-06,
      "loss": 0.4719,
      "step": 26545
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4708001613616943,
      "learning_rate": 1.3992435114377056e-06,
      "loss": 0.3559,
      "step": 26546
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6851603984832764,
      "learning_rate": 1.398723997108965e-06,
      "loss": 0.4148,
      "step": 26547
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5958619117736816,
      "learning_rate": 1.3982045719892135e-06,
      "loss": 0.4129,
      "step": 26548
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5065598487854004,
      "learning_rate": 1.3976852360838321e-06,
      "loss": 0.4153,
      "step": 26549
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5390292406082153,
      "learning_rate": 1.3971659893982136e-06,
      "loss": 0.3562,
      "step": 26550
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5690315961837769,
      "learning_rate": 1.396646831937739e-06,
      "loss": 0.4288,
      "step": 26551
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5530216693878174,
      "learning_rate": 1.3961277637077952e-06,
      "loss": 0.3606,
      "step": 26552
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6396750211715698,
      "learning_rate": 1.3956087847137635e-06,
      "loss": 0.4956,
      "step": 26553
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.492571473121643,
      "learning_rate": 1.395089894961027e-06,
      "loss": 0.352,
      "step": 26554
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6212177276611328,
      "learning_rate": 1.3945710944549705e-06,
      "loss": 0.4422,
      "step": 26555
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4912663698196411,
      "learning_rate": 1.3940523832009733e-06,
      "loss": 0.3702,
      "step": 26556
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.618787169456482,
      "learning_rate": 1.393533761204412e-06,
      "loss": 0.4385,
      "step": 26557
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5415579080581665,
      "learning_rate": 1.3930152284706676e-06,
      "loss": 0.3695,
      "step": 26558
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4717971086502075,
      "learning_rate": 1.3924967850051196e-06,
      "loss": 0.3699,
      "step": 26559
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5102002620697021,
      "learning_rate": 1.3919784308131468e-06,
      "loss": 0.3629,
      "step": 26560
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5633728504180908,
      "learning_rate": 1.3914601659001192e-06,
      "loss": 0.3987,
      "step": 26561
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5414931774139404,
      "learning_rate": 1.390941990271416e-06,
      "loss": 0.4167,
      "step": 26562
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8048110008239746,
      "learning_rate": 1.3904239039324097e-06,
      "loss": 0.4072,
      "step": 26563
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5170626640319824,
      "learning_rate": 1.389905906888478e-06,
      "loss": 0.3763,
      "step": 26564
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5492640733718872,
      "learning_rate": 1.38938799914499e-06,
      "loss": 0.4395,
      "step": 26565
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7108242511749268,
      "learning_rate": 1.3888701807073146e-06,
      "loss": 0.375,
      "step": 26566
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5423887968063354,
      "learning_rate": 1.3883524515808267e-06,
      "loss": 0.4238,
      "step": 26567
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4844927787780762,
      "learning_rate": 1.3878348117708962e-06,
      "loss": 0.3763,
      "step": 26568
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5564600229263306,
      "learning_rate": 1.3873172612828877e-06,
      "loss": 0.3968,
      "step": 26569
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5281939506530762,
      "learning_rate": 1.3867998001221738e-06,
      "loss": 0.3529,
      "step": 26570
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6346689462661743,
      "learning_rate": 1.3862824282941178e-06,
      "loss": 0.4814,
      "step": 26571
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5433768033981323,
      "learning_rate": 1.3857651458040878e-06,
      "loss": 0.3771,
      "step": 26572
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6039992570877075,
      "learning_rate": 1.3852479526574468e-06,
      "loss": 0.5126,
      "step": 26573
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4908117055892944,
      "learning_rate": 1.38473084885956e-06,
      "loss": 0.3934,
      "step": 26574
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4999693632125854,
      "learning_rate": 1.384213834415794e-06,
      "loss": 0.4224,
      "step": 26575
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4697318077087402,
      "learning_rate": 1.3836969093315068e-06,
      "loss": 0.363,
      "step": 26576
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5275413990020752,
      "learning_rate": 1.3831800736120592e-06,
      "loss": 0.4176,
      "step": 26577
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5262033939361572,
      "learning_rate": 1.382663327262812e-06,
      "loss": 0.3649,
      "step": 26578
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5962356328964233,
      "learning_rate": 1.3821466702891272e-06,
      "loss": 0.4413,
      "step": 26579
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4964954853057861,
      "learning_rate": 1.3816301026963642e-06,
      "loss": 0.3933,
      "step": 26580
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5523796081542969,
      "learning_rate": 1.3811136244898782e-06,
      "loss": 0.3833,
      "step": 26581
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5320098400115967,
      "learning_rate": 1.380597235675024e-06,
      "loss": 0.3782,
      "step": 26582
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5269571542739868,
      "learning_rate": 1.3800809362571599e-06,
      "loss": 0.4132,
      "step": 26583
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5348190069198608,
      "learning_rate": 1.3795647262416422e-06,
      "loss": 0.3555,
      "step": 26584
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6526771783828735,
      "learning_rate": 1.3790486056338204e-06,
      "loss": 0.4885,
      "step": 26585
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4851926565170288,
      "learning_rate": 1.3785325744390532e-06,
      "loss": 0.3629,
      "step": 26586
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5002895593643188,
      "learning_rate": 1.3780166326626876e-06,
      "loss": 0.367,
      "step": 26587
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.584982991218567,
      "learning_rate": 1.3775007803100782e-06,
      "loss": 0.376,
      "step": 26588
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.529417872428894,
      "learning_rate": 1.3769850173865717e-06,
      "loss": 0.4456,
      "step": 26589
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4815298318862915,
      "learning_rate": 1.3764693438975196e-06,
      "loss": 0.357,
      "step": 26590
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6726528406143188,
      "learning_rate": 1.3759537598482731e-06,
      "loss": 0.4125,
      "step": 26591
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.510434627532959,
      "learning_rate": 1.3754382652441757e-06,
      "loss": 0.3664,
      "step": 26592
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5495061874389648,
      "learning_rate": 1.3749228600905729e-06,
      "loss": 0.4084,
      "step": 26593
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5280849933624268,
      "learning_rate": 1.374407544392813e-06,
      "loss": 0.3668,
      "step": 26594
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7535661458969116,
      "learning_rate": 1.373892318156238e-06,
      "loss": 0.5153,
      "step": 26595
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5060383081436157,
      "learning_rate": 1.3733771813861996e-06,
      "loss": 0.3688,
      "step": 26596
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5544079542160034,
      "learning_rate": 1.37286213408803e-06,
      "loss": 0.4107,
      "step": 26597
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4498409032821655,
      "learning_rate": 1.3723471762670748e-06,
      "loss": 0.3565,
      "step": 26598
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6195244789123535,
      "learning_rate": 1.3718323079286754e-06,
      "loss": 0.4384,
      "step": 26599
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6417624950408936,
      "learning_rate": 1.3713175290781755e-06,
      "loss": 0.4129,
      "step": 26600
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6198471784591675,
      "learning_rate": 1.3708028397209105e-06,
      "loss": 0.4339,
      "step": 26601
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5766704082489014,
      "learning_rate": 1.3702882398622165e-06,
      "loss": 0.4018,
      "step": 26602
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5725399255752563,
      "learning_rate": 1.3697737295074332e-06,
      "loss": 0.4117,
      "step": 26603
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4999330043792725,
      "learning_rate": 1.3692593086618977e-06,
      "loss": 0.3566,
      "step": 26604
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6155356168746948,
      "learning_rate": 1.3687449773309435e-06,
      "loss": 0.4106,
      "step": 26605
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.49656081199646,
      "learning_rate": 1.3682307355199076e-06,
      "loss": 0.3901,
      "step": 26606
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.519312858581543,
      "learning_rate": 1.3677165832341199e-06,
      "loss": 0.4047,
      "step": 26607
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5295733213424683,
      "learning_rate": 1.3672025204789173e-06,
      "loss": 0.367,
      "step": 26608
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.727741003036499,
      "learning_rate": 1.3666885472596269e-06,
      "loss": 0.4417,
      "step": 26609
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5101947784423828,
      "learning_rate": 1.366174663581582e-06,
      "loss": 0.3575,
      "step": 26610
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6160938739776611,
      "learning_rate": 1.3656608694501139e-06,
      "loss": 0.4287,
      "step": 26611
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4368263483047485,
      "learning_rate": 1.3651471648705494e-06,
      "loss": 0.3614,
      "step": 26612
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5286104679107666,
      "learning_rate": 1.3646335498482144e-06,
      "loss": 0.4234,
      "step": 26613
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5062165260314941,
      "learning_rate": 1.3641200243884379e-06,
      "loss": 0.3669,
      "step": 26614
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5831271409988403,
      "learning_rate": 1.3636065884965465e-06,
      "loss": 0.452,
      "step": 26615
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5957263708114624,
      "learning_rate": 1.3630932421778676e-06,
      "loss": 0.406,
      "step": 26616
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6795265674591064,
      "learning_rate": 1.3625799854377231e-06,
      "loss": 0.4181,
      "step": 26617
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5216130018234253,
      "learning_rate": 1.3620668182814334e-06,
      "loss": 0.3578,
      "step": 26618
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5426156520843506,
      "learning_rate": 1.3615537407143231e-06,
      "loss": 0.3854,
      "step": 26619
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4282091856002808,
      "learning_rate": 1.3610407527417169e-06,
      "loss": 0.3406,
      "step": 26620
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5677305459976196,
      "learning_rate": 1.3605278543689305e-06,
      "loss": 0.4722,
      "step": 26621
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5253976583480835,
      "learning_rate": 1.3600150456012862e-06,
      "loss": 0.3705,
      "step": 26622
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9226741790771484,
      "learning_rate": 1.3595023264441009e-06,
      "loss": 1.0399,
      "step": 26623
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4483445882797241,
      "learning_rate": 1.358989696902696e-06,
      "loss": 0.9693,
      "step": 26624
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6586319208145142,
      "learning_rate": 1.358477156982383e-06,
      "loss": 0.479,
      "step": 26625
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5049387216567993,
      "learning_rate": 1.3579647066884794e-06,
      "loss": 0.3726,
      "step": 26626
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.65482497215271,
      "learning_rate": 1.3574523460263045e-06,
      "loss": 0.4656,
      "step": 26627
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5117090940475464,
      "learning_rate": 1.3569400750011686e-06,
      "loss": 0.3662,
      "step": 26628
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7825781106948853,
      "learning_rate": 1.3564278936183829e-06,
      "loss": 0.4339,
      "step": 26629
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.625575304031372,
      "learning_rate": 1.3559158018832619e-06,
      "loss": 0.3877,
      "step": 26630
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5965405702590942,
      "learning_rate": 1.355403799801116e-06,
      "loss": 0.4184,
      "step": 26631
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4708715677261353,
      "learning_rate": 1.3548918873772588e-06,
      "loss": 0.3632,
      "step": 26632
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5476452112197876,
      "learning_rate": 1.3543800646169968e-06,
      "loss": 0.4517,
      "step": 26633
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4947384595870972,
      "learning_rate": 1.3538683315256363e-06,
      "loss": 0.371,
      "step": 26634
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5241130590438843,
      "learning_rate": 1.353356688108487e-06,
      "loss": 0.4284,
      "step": 26635
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.441796064376831,
      "learning_rate": 1.3528451343708582e-06,
      "loss": 0.3685,
      "step": 26636
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.60292387008667,
      "learning_rate": 1.3523336703180523e-06,
      "loss": 0.4184,
      "step": 26637
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.455407977104187,
      "learning_rate": 1.351822295955373e-06,
      "loss": 0.3708,
      "step": 26638
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5131810903549194,
      "learning_rate": 1.3513110112881245e-06,
      "loss": 0.3953,
      "step": 26639
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5929721593856812,
      "learning_rate": 1.350799816321614e-06,
      "loss": 0.3731,
      "step": 26640
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.820543646812439,
      "learning_rate": 1.3502887110611374e-06,
      "loss": 0.4625,
      "step": 26641
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6238270998001099,
      "learning_rate": 1.3497776955120012e-06,
      "loss": 0.4152,
      "step": 26642
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5475735664367676,
      "learning_rate": 1.3492667696795004e-06,
      "loss": 0.4349,
      "step": 26643
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4461991786956787,
      "learning_rate": 1.348755933568938e-06,
      "loss": 0.3876,
      "step": 26644
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.692916750907898,
      "learning_rate": 1.3482451871856095e-06,
      "loss": 0.4471,
      "step": 26645
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.528195858001709,
      "learning_rate": 1.347734530534812e-06,
      "loss": 0.3838,
      "step": 26646
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4845603704452515,
      "learning_rate": 1.3472239636218465e-06,
      "loss": 0.3806,
      "step": 26647
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4780265092849731,
      "learning_rate": 1.3467134864520036e-06,
      "loss": 0.3584,
      "step": 26648
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.787616491317749,
      "learning_rate": 1.3462030990305786e-06,
      "loss": 0.4575,
      "step": 26649
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6499651670455933,
      "learning_rate": 1.3456928013628646e-06,
      "loss": 0.3796,
      "step": 26650
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5564662218093872,
      "learning_rate": 1.345182593454155e-06,
      "loss": 0.4034,
      "step": 26651
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5059211254119873,
      "learning_rate": 1.344672475309744e-06,
      "loss": 0.3886,
      "step": 26652
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5300796031951904,
      "learning_rate": 1.3441624469349212e-06,
      "loss": 0.4042,
      "step": 26653
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.562657117843628,
      "learning_rate": 1.3436525083349717e-06,
      "loss": 0.3861,
      "step": 26654
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9453327655792236,
      "learning_rate": 1.3431426595151874e-06,
      "loss": 0.4012,
      "step": 26655
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5271245241165161,
      "learning_rate": 1.3426329004808602e-06,
      "loss": 0.3439,
      "step": 26656
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.769905924797058,
      "learning_rate": 1.3421232312372712e-06,
      "loss": 0.4578,
      "step": 26657
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5391273498535156,
      "learning_rate": 1.3416136517897105e-06,
      "loss": 0.3713,
      "step": 26658
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.627921223640442,
      "learning_rate": 1.3411041621434607e-06,
      "loss": 0.4107,
      "step": 26659
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4821974039077759,
      "learning_rate": 1.3405947623038085e-06,
      "loss": 0.3747,
      "step": 26660
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.572340488433838,
      "learning_rate": 1.340085452276033e-06,
      "loss": 0.411,
      "step": 26661
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5128614902496338,
      "learning_rate": 1.3395762320654193e-06,
      "loss": 0.3734,
      "step": 26662
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5744348764419556,
      "learning_rate": 1.3390671016772516e-06,
      "loss": 0.4204,
      "step": 26663
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.601168155670166,
      "learning_rate": 1.3385580611168082e-06,
      "loss": 0.4175,
      "step": 26664
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.983256459236145,
      "learning_rate": 1.3380491103893645e-06,
      "loss": 0.4135,
      "step": 26665
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.519468903541565,
      "learning_rate": 1.3375402495002031e-06,
      "loss": 0.3807,
      "step": 26666
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5867540836334229,
      "learning_rate": 1.3370314784546012e-06,
      "loss": 0.498,
      "step": 26667
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.551342487335205,
      "learning_rate": 1.3365227972578388e-06,
      "loss": 0.3866,
      "step": 26668
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5192662477493286,
      "learning_rate": 1.336014205915187e-06,
      "loss": 0.4036,
      "step": 26669
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5548681020736694,
      "learning_rate": 1.3355057044319208e-06,
      "loss": 0.3866,
      "step": 26670
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5749377012252808,
      "learning_rate": 1.334997292813316e-06,
      "loss": 0.3907,
      "step": 26671
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6153398752212524,
      "learning_rate": 1.334488971064648e-06,
      "loss": 0.3701,
      "step": 26672
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5142210721969604,
      "learning_rate": 1.333980739191183e-06,
      "loss": 0.4556,
      "step": 26673
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5775587558746338,
      "learning_rate": 1.3334725971981977e-06,
      "loss": 0.3793,
      "step": 26674
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4930819272994995,
      "learning_rate": 1.3329645450909589e-06,
      "loss": 0.4081,
      "step": 26675
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.678526759147644,
      "learning_rate": 1.3324565828747381e-06,
      "loss": 0.4063,
      "step": 26676
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7890149354934692,
      "learning_rate": 1.3319487105548013e-06,
      "loss": 0.9632,
      "step": 26677
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5167324542999268,
      "learning_rate": 1.3314409281364181e-06,
      "loss": 1.0468,
      "step": 26678
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5314724445343018,
      "learning_rate": 1.3309332356248539e-06,
      "loss": 0.4242,
      "step": 26679
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5939851999282837,
      "learning_rate": 1.3304256330253751e-06,
      "loss": 0.389,
      "step": 26680
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.518783688545227,
      "learning_rate": 1.3299181203432443e-06,
      "loss": 0.4348,
      "step": 26681
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.519971489906311,
      "learning_rate": 1.3294106975837273e-06,
      "loss": 0.3888,
      "step": 26682
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5521893501281738,
      "learning_rate": 1.3289033647520878e-06,
      "loss": 0.4489,
      "step": 26683
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.522373914718628,
      "learning_rate": 1.3283961218535867e-06,
      "loss": 0.3745,
      "step": 26684
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.598284125328064,
      "learning_rate": 1.3278889688934815e-06,
      "loss": 0.4123,
      "step": 26685
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4664016962051392,
      "learning_rate": 1.3273819058770354e-06,
      "loss": 0.3731,
      "step": 26686
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6496105194091797,
      "learning_rate": 1.3268749328095076e-06,
      "loss": 0.4019,
      "step": 26687
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.530014991760254,
      "learning_rate": 1.3263680496961573e-06,
      "loss": 0.397,
      "step": 26688
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5510814189910889,
      "learning_rate": 1.3258612565422412e-06,
      "loss": 0.4327,
      "step": 26689
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5096291303634644,
      "learning_rate": 1.325354553353011e-06,
      "loss": 0.4013,
      "step": 26690
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6052730083465576,
      "learning_rate": 1.3248479401337267e-06,
      "loss": 0.4154,
      "step": 26691
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.619364619255066,
      "learning_rate": 1.324341416889643e-06,
      "loss": 0.3729,
      "step": 26692
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5568655729293823,
      "learning_rate": 1.32383498362601e-06,
      "loss": 0.4493,
      "step": 26693
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.549172043800354,
      "learning_rate": 1.3233286403480838e-06,
      "loss": 0.3776,
      "step": 26694
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5300471782684326,
      "learning_rate": 1.3228223870611135e-06,
      "loss": 0.4108,
      "step": 26695
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.63894784450531,
      "learning_rate": 1.3223162237703514e-06,
      "loss": 0.3845,
      "step": 26696
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6475021839141846,
      "learning_rate": 1.3218101504810444e-06,
      "loss": 1.0125,
      "step": 26697
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.404487133026123,
      "learning_rate": 1.3213041671984439e-06,
      "loss": 0.9632,
      "step": 26698
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.537980079650879,
      "learning_rate": 1.320798273927798e-06,
      "loss": 0.4148,
      "step": 26699
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6607370376586914,
      "learning_rate": 1.3202924706743537e-06,
      "loss": 0.4406,
      "step": 26700
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6449769735336304,
      "learning_rate": 1.3197867574433542e-06,
      "loss": 0.5064,
      "step": 26701
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4575517177581787,
      "learning_rate": 1.3192811342400457e-06,
      "loss": 0.3508,
      "step": 26702
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.25890851020813,
      "learning_rate": 1.3187756010696738e-06,
      "loss": 0.4442,
      "step": 26703
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5623546838760376,
      "learning_rate": 1.3182701579374823e-06,
      "loss": 0.3663,
      "step": 26704
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5183615684509277,
      "learning_rate": 1.3177648048487123e-06,
      "loss": 0.4032,
      "step": 26705
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.452606439590454,
      "learning_rate": 1.317259541808602e-06,
      "loss": 0.3732,
      "step": 26706
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6804231405258179,
      "learning_rate": 1.316754368822396e-06,
      "loss": 0.4525,
      "step": 26707
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.512937068939209,
      "learning_rate": 1.3162492858953335e-06,
      "loss": 0.3836,
      "step": 26708
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5597180128097534,
      "learning_rate": 1.3157442930326504e-06,
      "loss": 0.4366,
      "step": 26709
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5068525075912476,
      "learning_rate": 1.3152393902395876e-06,
      "loss": 0.3725,
      "step": 26710
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.655937910079956,
      "learning_rate": 1.3147345775213783e-06,
      "loss": 0.486,
      "step": 26711
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.57525634765625,
      "learning_rate": 1.3142298548832622e-06,
      "loss": 0.384,
      "step": 26712
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.537559151649475,
      "learning_rate": 1.3137252223304687e-06,
      "loss": 0.4408,
      "step": 26713
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5137972831726074,
      "learning_rate": 1.3132206798682369e-06,
      "loss": 0.3858,
      "step": 26714
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5677186250686646,
      "learning_rate": 1.3127162275017957e-06,
      "loss": 0.4059,
      "step": 26715
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5394527912139893,
      "learning_rate": 1.3122118652363813e-06,
      "loss": 0.3686,
      "step": 26716
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.495001196861267,
      "learning_rate": 1.3117075930772193e-06,
      "loss": 0.4599,
      "step": 26717
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5107684135437012,
      "learning_rate": 1.3112034110295435e-06,
      "loss": 0.3691,
      "step": 26718
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6453608274459839,
      "learning_rate": 1.310699319098585e-06,
      "loss": 0.43,
      "step": 26719
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5260133743286133,
      "learning_rate": 1.3101953172895688e-06,
      "loss": 0.3601,
      "step": 26720
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5173131227493286,
      "learning_rate": 1.3096914056077216e-06,
      "loss": 0.3927,
      "step": 26721
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4272624254226685,
      "learning_rate": 1.3091875840582701e-06,
      "loss": 0.3551,
      "step": 26722
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.704581618309021,
      "learning_rate": 1.3086838526464407e-06,
      "loss": 0.4506,
      "step": 26723
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7858972549438477,
      "learning_rate": 1.30818021137746e-06,
      "loss": 0.3826,
      "step": 26724
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5973095893859863,
      "learning_rate": 1.3076766602565494e-06,
      "loss": 0.4167,
      "step": 26725
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.500195026397705,
      "learning_rate": 1.3071731992889302e-06,
      "loss": 0.3794,
      "step": 26726
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7318990230560303,
      "learning_rate": 1.306669828479825e-06,
      "loss": 0.4182,
      "step": 26727
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5184025764465332,
      "learning_rate": 1.3061665478344566e-06,
      "loss": 0.3673,
      "step": 26728
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.551127314567566,
      "learning_rate": 1.3056633573580423e-06,
      "loss": 0.4384,
      "step": 26729
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5052496194839478,
      "learning_rate": 1.3051602570558031e-06,
      "loss": 0.4008,
      "step": 26730
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.666330337524414,
      "learning_rate": 1.3046572469329533e-06,
      "loss": 0.4251,
      "step": 26731
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5664126873016357,
      "learning_rate": 1.3041543269947155e-06,
      "loss": 0.3888,
      "step": 26732
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5179940462112427,
      "learning_rate": 1.3036514972462999e-06,
      "loss": 0.4188,
      "step": 26733
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4586782455444336,
      "learning_rate": 1.3031487576929246e-06,
      "loss": 0.3818,
      "step": 26734
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.489216685295105,
      "learning_rate": 1.3026461083398067e-06,
      "loss": 0.4232,
      "step": 26735
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.645472526550293,
      "learning_rate": 1.3021435491921552e-06,
      "loss": 0.3912,
      "step": 26736
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5938905477523804,
      "learning_rate": 1.3016410802551828e-06,
      "loss": 0.4307,
      "step": 26737
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5411638021469116,
      "learning_rate": 1.3011387015341005e-06,
      "loss": 0.3725,
      "step": 26738
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.524310827255249,
      "learning_rate": 1.300636413034121e-06,
      "loss": 0.3704,
      "step": 26739
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.519254207611084,
      "learning_rate": 1.3001342147604533e-06,
      "loss": 0.3938,
      "step": 26740
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6394026279449463,
      "learning_rate": 1.299632106718307e-06,
      "loss": 0.4651,
      "step": 26741
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5277982950210571,
      "learning_rate": 1.2991300889128867e-06,
      "loss": 0.3688,
      "step": 26742
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4609150886535645,
      "learning_rate": 1.2986281613494002e-06,
      "loss": 0.4014,
      "step": 26743
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.516964316368103,
      "learning_rate": 1.2981263240330532e-06,
      "loss": 0.3619,
      "step": 26744
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.669154167175293,
      "learning_rate": 1.2976245769690543e-06,
      "loss": 0.4412,
      "step": 26745
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4178718328475952,
      "learning_rate": 1.2971229201626034e-06,
      "loss": 0.3409,
      "step": 26746
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.971830129623413,
      "learning_rate": 1.2966213536189031e-06,
      "loss": 0.4579,
      "step": 26747
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.550673246383667,
      "learning_rate": 1.2961198773431572e-06,
      "loss": 0.3935,
      "step": 26748
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4874345064163208,
      "learning_rate": 1.2956184913405678e-06,
      "loss": 0.418,
      "step": 26749
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4833048582077026,
      "learning_rate": 1.2951171956163311e-06,
      "loss": 0.3813,
      "step": 26750
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5957192182540894,
      "learning_rate": 1.2946159901756516e-06,
      "loss": 0.4795,
      "step": 26751
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.541024923324585,
      "learning_rate": 1.2941148750237232e-06,
      "loss": 0.3799,
      "step": 26752
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.542873501777649,
      "learning_rate": 1.2936138501657458e-06,
      "loss": 0.4001,
      "step": 26753
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.48365318775177,
      "learning_rate": 1.2931129156069145e-06,
      "loss": 0.3651,
      "step": 26754
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6263527870178223,
      "learning_rate": 1.292612071352426e-06,
      "loss": 0.4603,
      "step": 26755
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6677651405334473,
      "learning_rate": 1.2921113174074728e-06,
      "loss": 0.4276,
      "step": 26756
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5616419315338135,
      "learning_rate": 1.291610653777251e-06,
      "loss": 0.4124,
      "step": 26757
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5372909307479858,
      "learning_rate": 1.2911100804669508e-06,
      "loss": 0.3743,
      "step": 26758
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6196750402450562,
      "learning_rate": 1.2906095974817656e-06,
      "loss": 0.4515,
      "step": 26759
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.826080083847046,
      "learning_rate": 1.2901092048268883e-06,
      "loss": 0.4149,
      "step": 26760
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.59862220287323,
      "learning_rate": 1.2896089025075053e-06,
      "loss": 0.4556,
      "step": 26761
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5154445171356201,
      "learning_rate": 1.2891086905288053e-06,
      "loss": 0.382,
      "step": 26762
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5299278497695923,
      "learning_rate": 1.288608568895977e-06,
      "loss": 0.4216,
      "step": 26763
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.58296799659729,
      "learning_rate": 1.2881085376142077e-06,
      "loss": 0.3928,
      "step": 26764
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.600583791732788,
      "learning_rate": 1.2876085966886865e-06,
      "loss": 0.4553,
      "step": 26765
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4969768524169922,
      "learning_rate": 1.2871087461245967e-06,
      "loss": 0.3734,
      "step": 26766
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.649713397026062,
      "learning_rate": 1.286608985927119e-06,
      "loss": 0.5195,
      "step": 26767
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.489872932434082,
      "learning_rate": 1.2861093161014394e-06,
      "loss": 0.3691,
      "step": 26768
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6504812240600586,
      "learning_rate": 1.285609736652742e-06,
      "loss": 0.5102,
      "step": 26769
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5091601610183716,
      "learning_rate": 1.2851102475862055e-06,
      "loss": 0.3554,
      "step": 26770
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6422446966171265,
      "learning_rate": 1.2846108489070142e-06,
      "loss": 0.4773,
      "step": 26771
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4849605560302734,
      "learning_rate": 1.284111540620342e-06,
      "loss": 0.3545,
      "step": 26772
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.526957392692566,
      "learning_rate": 1.2836123227313724e-06,
      "loss": 0.416,
      "step": 26773
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5520511865615845,
      "learning_rate": 1.283113195245279e-06,
      "loss": 0.3986,
      "step": 26774
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5746970176696777,
      "learning_rate": 1.2826141581672413e-06,
      "loss": 0.411,
      "step": 26775
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6431493759155273,
      "learning_rate": 1.282115211502437e-06,
      "loss": 0.4017,
      "step": 26776
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5668128728866577,
      "learning_rate": 1.2816163552560378e-06,
      "loss": 0.4083,
      "step": 26777
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5454152822494507,
      "learning_rate": 1.2811175894332172e-06,
      "loss": 0.423,
      "step": 26778
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4756697416305542,
      "learning_rate": 1.2806189140391489e-06,
      "loss": 0.386,
      "step": 26779
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.469385027885437,
      "learning_rate": 1.2801203290790055e-06,
      "loss": 0.3581,
      "step": 26780
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5434107780456543,
      "learning_rate": 1.2796218345579592e-06,
      "loss": 0.4359,
      "step": 26781
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6548049449920654,
      "learning_rate": 1.2791234304811806e-06,
      "loss": 0.3815,
      "step": 26782
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6897166967391968,
      "learning_rate": 1.2786251168538344e-06,
      "loss": 0.4116,
      "step": 26783
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5333449840545654,
      "learning_rate": 1.278126893681092e-06,
      "loss": 0.3765,
      "step": 26784
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.570823073387146,
      "learning_rate": 1.2776287609681237e-06,
      "loss": 0.4598,
      "step": 26785
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4220821857452393,
      "learning_rate": 1.2771307187200899e-06,
      "loss": 0.3501,
      "step": 26786
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5293920040130615,
      "learning_rate": 1.2766327669421618e-06,
      "loss": 0.4177,
      "step": 26787
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4204879999160767,
      "learning_rate": 1.276134905639499e-06,
      "loss": 0.4034,
      "step": 26788
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5655231475830078,
      "learning_rate": 1.2756371348172692e-06,
      "loss": 0.4694,
      "step": 26789
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4944860935211182,
      "learning_rate": 1.2751394544806307e-06,
      "loss": 0.3814,
      "step": 26790
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6760903596878052,
      "learning_rate": 1.2746418646347481e-06,
      "loss": 0.4514,
      "step": 26791
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4890971183776855,
      "learning_rate": 1.274144365284785e-06,
      "loss": 0.3642,
      "step": 26792
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4763654470443726,
      "learning_rate": 1.2736469564358978e-06,
      "loss": 0.3959,
      "step": 26793
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5566370487213135,
      "learning_rate": 1.2731496380932429e-06,
      "loss": 0.3822,
      "step": 26794
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5689820051193237,
      "learning_rate": 1.2726524102619818e-06,
      "loss": 0.4109,
      "step": 26795
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5004163980484009,
      "learning_rate": 1.272155272947273e-06,
      "loss": 0.3728,
      "step": 26796
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5993643999099731,
      "learning_rate": 1.2716582261542709e-06,
      "loss": 0.4605,
      "step": 26797
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5506680011749268,
      "learning_rate": 1.271161269888127e-06,
      "loss": 0.3672,
      "step": 26798
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5000215768814087,
      "learning_rate": 1.2706644041540006e-06,
      "loss": 0.4088,
      "step": 26799
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.609497308731079,
      "learning_rate": 1.270167628957042e-06,
      "loss": 0.3835,
      "step": 26800
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.50260591506958,
      "learning_rate": 1.2696709443024068e-06,
      "loss": 0.4273,
      "step": 26801
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5932998657226562,
      "learning_rate": 1.2691743501952447e-06,
      "loss": 0.3567,
      "step": 26802
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7982515096664429,
      "learning_rate": 1.2686778466407045e-06,
      "loss": 0.9696,
      "step": 26803
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4423729181289673,
      "learning_rate": 1.2681814336439368e-06,
      "loss": 1.0423,
      "step": 26804
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5258808135986328,
      "learning_rate": 1.267685111210092e-06,
      "loss": 0.3912,
      "step": 26805
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4917175769805908,
      "learning_rate": 1.2671888793443144e-06,
      "loss": 0.3973,
      "step": 26806
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5274442434310913,
      "learning_rate": 1.2666927380517546e-06,
      "loss": 0.455,
      "step": 26807
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.550515055656433,
      "learning_rate": 1.2661966873375553e-06,
      "loss": 0.3859,
      "step": 26808
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5119073390960693,
      "learning_rate": 1.2657007272068632e-06,
      "loss": 0.3922,
      "step": 26809
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4593753814697266,
      "learning_rate": 1.2652048576648202e-06,
      "loss": 0.3711,
      "step": 26810
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5626674890518188,
      "learning_rate": 1.2647090787165694e-06,
      "loss": 0.4233,
      "step": 26811
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5323466062545776,
      "learning_rate": 1.2642133903672572e-06,
      "loss": 0.4016,
      "step": 26812
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6705611944198608,
      "learning_rate": 1.2637177926220212e-06,
      "loss": 0.4079,
      "step": 26813
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6152784824371338,
      "learning_rate": 1.2632222854859988e-06,
      "loss": 0.3961,
      "step": 26814
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5254631042480469,
      "learning_rate": 1.2627268689643325e-06,
      "loss": 0.4388,
      "step": 26815
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.509301781654358,
      "learning_rate": 1.26223154306216e-06,
      "loss": 0.3627,
      "step": 26816
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5352174043655396,
      "learning_rate": 1.2617363077846212e-06,
      "loss": 0.4011,
      "step": 26817
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5295133590698242,
      "learning_rate": 1.2612411631368505e-06,
      "loss": 0.3838,
      "step": 26818
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5434335470199585,
      "learning_rate": 1.2607461091239803e-06,
      "loss": 0.378,
      "step": 26819
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6720185279846191,
      "learning_rate": 1.260251145751148e-06,
      "loss": 0.3932,
      "step": 26820
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5315017700195312,
      "learning_rate": 1.2597562730234892e-06,
      "loss": 0.4121,
      "step": 26821
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5445928573608398,
      "learning_rate": 1.259261490946131e-06,
      "loss": 0.364,
      "step": 26822
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6203932762145996,
      "learning_rate": 1.2587667995242114e-06,
      "loss": 0.4022,
      "step": 26823
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4614871740341187,
      "learning_rate": 1.2582721987628555e-06,
      "loss": 0.3741,
      "step": 26824
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.586495041847229,
      "learning_rate": 1.2577776886671977e-06,
      "loss": 0.4377,
      "step": 26825
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5641729831695557,
      "learning_rate": 1.257283269242363e-06,
      "loss": 0.3781,
      "step": 26826
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8342978954315186,
      "learning_rate": 1.2567889404934809e-06,
      "loss": 0.4343,
      "step": 26827
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5202109813690186,
      "learning_rate": 1.256294702425681e-06,
      "loss": 0.3919,
      "step": 26828
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5119962692260742,
      "learning_rate": 1.2558005550440867e-06,
      "loss": 0.3977,
      "step": 26829
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5779439210891724,
      "learning_rate": 1.255306498353821e-06,
      "loss": 0.4029,
      "step": 26830
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5109609365463257,
      "learning_rate": 1.2548125323600101e-06,
      "loss": 0.4177,
      "step": 26831
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4530906677246094,
      "learning_rate": 1.25431865706778e-06,
      "loss": 0.3561,
      "step": 26832
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6122006177902222,
      "learning_rate": 1.253824872482251e-06,
      "loss": 0.4311,
      "step": 26833
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.487359642982483,
      "learning_rate": 1.2533311786085412e-06,
      "loss": 0.379,
      "step": 26834
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5548217296600342,
      "learning_rate": 1.252837575451774e-06,
      "loss": 0.4402,
      "step": 26835
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.508787989616394,
      "learning_rate": 1.252344063017068e-06,
      "loss": 0.3946,
      "step": 26836
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.541379451751709,
      "learning_rate": 1.2518506413095455e-06,
      "loss": 0.4223,
      "step": 26837
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.569509744644165,
      "learning_rate": 1.2513573103343202e-06,
      "loss": 0.3888,
      "step": 26838
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1056876182556152,
      "learning_rate": 1.250864070096507e-06,
      "loss": 0.418,
      "step": 26839
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5210816860198975,
      "learning_rate": 1.2503709206012239e-06,
      "loss": 0.415,
      "step": 26840
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5537643432617188,
      "learning_rate": 1.249877861853589e-06,
      "loss": 0.4101,
      "step": 26841
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6182057857513428,
      "learning_rate": 1.2493848938587095e-06,
      "loss": 0.4207,
      "step": 26842
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6291617155075073,
      "learning_rate": 1.2488920166217034e-06,
      "loss": 0.4557,
      "step": 26843
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5873620510101318,
      "learning_rate": 1.24839923014768e-06,
      "loss": 0.3904,
      "step": 26844
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.482981562614441,
      "learning_rate": 1.2479065344417528e-06,
      "loss": 0.3945,
      "step": 26845
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.508731722831726,
      "learning_rate": 1.2474139295090271e-06,
      "loss": 0.3737,
      "step": 26846
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5442616939544678,
      "learning_rate": 1.246921415354616e-06,
      "loss": 0.4236,
      "step": 26847
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4817858934402466,
      "learning_rate": 1.2464289919836292e-06,
      "loss": 0.3674,
      "step": 26848
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.521775484085083,
      "learning_rate": 1.2459366594011702e-06,
      "loss": 0.3948,
      "step": 26849
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4909279346466064,
      "learning_rate": 1.2454444176123447e-06,
      "loss": 0.3842,
      "step": 26850
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5751502513885498,
      "learning_rate": 1.2449522666222603e-06,
      "loss": 0.4605,
      "step": 26851
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5400230884552002,
      "learning_rate": 1.2444602064360212e-06,
      "loss": 0.3645,
      "step": 26852
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5659892559051514,
      "learning_rate": 1.2439682370587326e-06,
      "loss": 0.428,
      "step": 26853
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5386956930160522,
      "learning_rate": 1.2434763584954946e-06,
      "loss": 0.3653,
      "step": 26854
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.575538158416748,
      "learning_rate": 1.2429845707514065e-06,
      "loss": 0.4836,
      "step": 26855
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.586722493171692,
      "learning_rate": 1.2424928738315723e-06,
      "loss": 0.3481,
      "step": 26856
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.580844521522522,
      "learning_rate": 1.242001267741093e-06,
      "loss": 0.3838,
      "step": 26857
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4484304189682007,
      "learning_rate": 1.2415097524850628e-06,
      "loss": 0.3624,
      "step": 26858
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4848536252975464,
      "learning_rate": 1.241018328068585e-06,
      "loss": 0.3946,
      "step": 26859
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.509165644645691,
      "learning_rate": 1.2405269944967502e-06,
      "loss": 0.361,
      "step": 26860
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5932093858718872,
      "learning_rate": 1.240035751774661e-06,
      "loss": 0.4845,
      "step": 26861
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.598443865776062,
      "learning_rate": 1.239544599907405e-06,
      "loss": 0.4046,
      "step": 26862
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.495416283607483,
      "learning_rate": 1.2390535389000813e-06,
      "loss": 0.394,
      "step": 26863
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5701823234558105,
      "learning_rate": 1.238562568757784e-06,
      "loss": 0.3802,
      "step": 26864
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6097588539123535,
      "learning_rate": 1.2380716894856037e-06,
      "loss": 0.4246,
      "step": 26865
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4687656164169312,
      "learning_rate": 1.2375809010886275e-06,
      "loss": 0.3536,
      "step": 26866
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5674631595611572,
      "learning_rate": 1.23709020357195e-06,
      "loss": 0.439,
      "step": 26867
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5407594442367554,
      "learning_rate": 1.2365995969406597e-06,
      "loss": 0.3765,
      "step": 26868
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6540788412094116,
      "learning_rate": 1.23610908119985e-06,
      "loss": 0.4288,
      "step": 26869
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5549975633621216,
      "learning_rate": 1.235618656354598e-06,
      "loss": 0.4179,
      "step": 26870
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6112316846847534,
      "learning_rate": 1.2351283224099952e-06,
      "loss": 0.4402,
      "step": 26871
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4792182445526123,
      "learning_rate": 1.2346380793711277e-06,
      "loss": 0.3777,
      "step": 26872
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6652424335479736,
      "learning_rate": 1.2341479272430822e-06,
      "loss": 0.4778,
      "step": 26873
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.522820234298706,
      "learning_rate": 1.2336578660309406e-06,
      "loss": 0.3801,
      "step": 26874
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6108486652374268,
      "learning_rate": 1.2331678957397819e-06,
      "loss": 0.3992,
      "step": 26875
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5186837911605835,
      "learning_rate": 1.232678016374691e-06,
      "loss": 0.3638,
      "step": 26876
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6398600339889526,
      "learning_rate": 1.2321882279407516e-06,
      "loss": 0.3991,
      "step": 26877
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6511636972427368,
      "learning_rate": 1.2316985304430384e-06,
      "loss": 0.3692,
      "step": 26878
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5494897365570068,
      "learning_rate": 1.2312089238866343e-06,
      "loss": 0.4189,
      "step": 26879
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5795551538467407,
      "learning_rate": 1.2307194082766139e-06,
      "loss": 0.3917,
      "step": 26880
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5562719106674194,
      "learning_rate": 1.2302299836180587e-06,
      "loss": 0.3982,
      "step": 26881
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.462655782699585,
      "learning_rate": 1.229740649916039e-06,
      "loss": 0.3723,
      "step": 26882
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5496364831924438,
      "learning_rate": 1.2292514071756334e-06,
      "loss": 0.3994,
      "step": 26883
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5132726430892944,
      "learning_rate": 1.2287622554019186e-06,
      "loss": 0.361,
      "step": 26884
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5303055047988892,
      "learning_rate": 1.2282731945999637e-06,
      "loss": 0.3981,
      "step": 26885
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5744816064834595,
      "learning_rate": 1.2277842247748417e-06,
      "loss": 0.4012,
      "step": 26886
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5537258386611938,
      "learning_rate": 1.2272953459316238e-06,
      "loss": 0.439,
      "step": 26887
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6549123525619507,
      "learning_rate": 1.2268065580753829e-06,
      "loss": 0.3699,
      "step": 26888
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5757663249969482,
      "learning_rate": 1.226317861211188e-06,
      "loss": 0.4497,
      "step": 26889
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6164584159851074,
      "learning_rate": 1.2258292553441064e-06,
      "loss": 0.3761,
      "step": 26890
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5326428413391113,
      "learning_rate": 1.225340740479205e-06,
      "loss": 0.4368,
      "step": 26891
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.652204155921936,
      "learning_rate": 1.2248523166215508e-06,
      "loss": 0.4088,
      "step": 26892
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5954426527023315,
      "learning_rate": 1.2243639837762123e-06,
      "loss": 0.4713,
      "step": 26893
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5138803720474243,
      "learning_rate": 1.2238757419482505e-06,
      "loss": 0.3767,
      "step": 26894
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5754923820495605,
      "learning_rate": 1.2233875911427329e-06,
      "loss": 0.4766,
      "step": 26895
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5378764867782593,
      "learning_rate": 1.2228995313647185e-06,
      "loss": 0.3818,
      "step": 26896
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5773380994796753,
      "learning_rate": 1.2224115626192734e-06,
      "loss": 0.4615,
      "step": 26897
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.526006817817688,
      "learning_rate": 1.2219236849114547e-06,
      "loss": 0.3619,
      "step": 26898
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.581053376197815,
      "learning_rate": 1.2214358982463236e-06,
      "loss": 0.4042,
      "step": 26899
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5018776655197144,
      "learning_rate": 1.220948202628942e-06,
      "loss": 0.3735,
      "step": 26900
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6262472867965698,
      "learning_rate": 1.2204605980643669e-06,
      "loss": 0.4264,
      "step": 26901
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5564886331558228,
      "learning_rate": 1.2199730845576507e-06,
      "loss": 0.4128,
      "step": 26902
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.524877905845642,
      "learning_rate": 1.219485662113855e-06,
      "loss": 0.4019,
      "step": 26903
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5978519916534424,
      "learning_rate": 1.2189983307380338e-06,
      "loss": 0.3589,
      "step": 26904
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5540647506713867,
      "learning_rate": 1.2185110904352448e-06,
      "loss": 0.4346,
      "step": 26905
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.518645167350769,
      "learning_rate": 1.2180239412105344e-06,
      "loss": 0.3969,
      "step": 26906
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7137818336486816,
      "learning_rate": 1.2175368830689593e-06,
      "loss": 0.4486,
      "step": 26907
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.49019193649292,
      "learning_rate": 1.217049916015569e-06,
      "loss": 0.412,
      "step": 26908
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6336907148361206,
      "learning_rate": 1.2165630400554196e-06,
      "loss": 0.4637,
      "step": 26909
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.475722074508667,
      "learning_rate": 1.2160762551935558e-06,
      "loss": 0.3762,
      "step": 26910
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.582987666130066,
      "learning_rate": 1.2155895614350255e-06,
      "loss": 0.4131,
      "step": 26911
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5187841653823853,
      "learning_rate": 1.2151029587848795e-06,
      "loss": 0.3687,
      "step": 26912
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5823485851287842,
      "learning_rate": 1.2146164472481648e-06,
      "loss": 0.3986,
      "step": 26913
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.504201054573059,
      "learning_rate": 1.214130026829924e-06,
      "loss": 0.3674,
      "step": 26914
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.526939034461975,
      "learning_rate": 1.2136436975352062e-06,
      "loss": 0.4194,
      "step": 26915
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5578442811965942,
      "learning_rate": 1.213157459369051e-06,
      "loss": 0.3953,
      "step": 26916
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6619760990142822,
      "learning_rate": 1.2126713123365065e-06,
      "loss": 0.4507,
      "step": 26917
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4977827072143555,
      "learning_rate": 1.2121852564426095e-06,
      "loss": 0.3739,
      "step": 26918
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5845738649368286,
      "learning_rate": 1.2116992916924042e-06,
      "loss": 0.4454,
      "step": 26919
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.458952784538269,
      "learning_rate": 1.2112134180909319e-06,
      "loss": 0.3563,
      "step": 26920
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5075112581253052,
      "learning_rate": 1.2107276356432308e-06,
      "loss": 0.4343,
      "step": 26921
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5432193279266357,
      "learning_rate": 1.2102419443543367e-06,
      "loss": 0.3889,
      "step": 26922
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5933133363723755,
      "learning_rate": 1.2097563442292904e-06,
      "loss": 0.3945,
      "step": 26923
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5126326084136963,
      "learning_rate": 1.2092708352731253e-06,
      "loss": 0.3753,
      "step": 26924
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.502056360244751,
      "learning_rate": 1.208785417490882e-06,
      "loss": 0.4025,
      "step": 26925
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6013985872268677,
      "learning_rate": 1.208300090887592e-06,
      "loss": 0.4268,
      "step": 26926
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5402995347976685,
      "learning_rate": 1.2078148554682857e-06,
      "loss": 0.438,
      "step": 26927
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6195331811904907,
      "learning_rate": 1.2073297112379989e-06,
      "loss": 0.3703,
      "step": 26928
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4818527698516846,
      "learning_rate": 1.2068446582017656e-06,
      "loss": 0.3921,
      "step": 26929
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.504688024520874,
      "learning_rate": 1.2063596963646118e-06,
      "loss": 0.378,
      "step": 26930
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7065150737762451,
      "learning_rate": 1.2058748257315722e-06,
      "loss": 0.5009,
      "step": 26931
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5311421155929565,
      "learning_rate": 1.2053900463076706e-06,
      "loss": 0.358,
      "step": 26932
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4710109233856201,
      "learning_rate": 1.2049053580979397e-06,
      "loss": 0.4082,
      "step": 26933
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5637730360031128,
      "learning_rate": 1.2044207611074022e-06,
      "loss": 0.3889,
      "step": 26934
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4810694456100464,
      "learning_rate": 1.2039362553410872e-06,
      "loss": 0.3869,
      "step": 26935
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4658282995224,
      "learning_rate": 1.2034518408040198e-06,
      "loss": 0.3731,
      "step": 26936
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5866706371307373,
      "learning_rate": 1.2029675175012235e-06,
      "loss": 0.4201,
      "step": 26937
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5364595651626587,
      "learning_rate": 1.2024832854377188e-06,
      "loss": 0.4118,
      "step": 26938
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5368800163269043,
      "learning_rate": 1.201999144618531e-06,
      "loss": 0.4517,
      "step": 26939
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4688265323638916,
      "learning_rate": 1.201515095048681e-06,
      "loss": 0.3578,
      "step": 26940
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6041467189788818,
      "learning_rate": 1.2010311367331895e-06,
      "loss": 0.4448,
      "step": 26941
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4655611515045166,
      "learning_rate": 1.200547269677077e-06,
      "loss": 0.37,
      "step": 26942
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5744940042495728,
      "learning_rate": 1.2000634938853574e-06,
      "loss": 0.4068,
      "step": 26943
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5184686183929443,
      "learning_rate": 1.1995798093630517e-06,
      "loss": 0.3798,
      "step": 26944
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4859397411346436,
      "learning_rate": 1.1990962161151776e-06,
      "loss": 0.3978,
      "step": 26945
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5402929782867432,
      "learning_rate": 1.1986127141467495e-06,
      "loss": 0.3836,
      "step": 26946
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5333548784255981,
      "learning_rate": 1.1981293034627795e-06,
      "loss": 0.4243,
      "step": 26947
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6444402933120728,
      "learning_rate": 1.197645984068283e-06,
      "loss": 0.3741,
      "step": 26948
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6018790006637573,
      "learning_rate": 1.1971627559682764e-06,
      "loss": 0.415,
      "step": 26949
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5528826713562012,
      "learning_rate": 1.1966796191677655e-06,
      "loss": 0.3561,
      "step": 26950
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4413609504699707,
      "learning_rate": 1.1961965736717673e-06,
      "loss": 0.3866,
      "step": 26951
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5002520084381104,
      "learning_rate": 1.195713619485286e-06,
      "loss": 0.356,
      "step": 26952
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5643672943115234,
      "learning_rate": 1.1952307566133348e-06,
      "loss": 0.4087,
      "step": 26953
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5186805725097656,
      "learning_rate": 1.1947479850609178e-06,
      "loss": 0.3658,
      "step": 26954
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6483159065246582,
      "learning_rate": 1.1942653048330455e-06,
      "loss": 0.4267,
      "step": 26955
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5817060470581055,
      "learning_rate": 1.1937827159347248e-06,
      "loss": 0.3896,
      "step": 26956
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8954524993896484,
      "learning_rate": 1.1933002183709596e-06,
      "loss": 1.0047,
      "step": 26957
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.477359414100647,
      "learning_rate": 1.1928178121467515e-06,
      "loss": 0.9792,
      "step": 26958
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5137100219726562,
      "learning_rate": 1.1923354972671052e-06,
      "loss": 0.4169,
      "step": 26959
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4833394289016724,
      "learning_rate": 1.1918532737370247e-06,
      "loss": 0.3877,
      "step": 26960
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.612980842590332,
      "learning_rate": 1.1913711415615125e-06,
      "loss": 0.4393,
      "step": 26961
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5070573091506958,
      "learning_rate": 1.190889100745568e-06,
      "loss": 0.3891,
      "step": 26962
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5032870769500732,
      "learning_rate": 1.1904071512941873e-06,
      "loss": 0.422,
      "step": 26963
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5678200721740723,
      "learning_rate": 1.1899252932123717e-06,
      "loss": 0.3851,
      "step": 26964
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.635046124458313,
      "learning_rate": 1.1894435265051208e-06,
      "loss": 0.4604,
      "step": 26965
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5148009061813354,
      "learning_rate": 1.1889618511774282e-06,
      "loss": 0.3889,
      "step": 26966
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5076704025268555,
      "learning_rate": 1.188480267234292e-06,
      "loss": 0.371,
      "step": 26967
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5111825466156006,
      "learning_rate": 1.187998774680703e-06,
      "loss": 0.3753,
      "step": 26968
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5113569498062134,
      "learning_rate": 1.1875173735216616e-06,
      "loss": 0.3982,
      "step": 26969
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.515184998512268,
      "learning_rate": 1.1870360637621536e-06,
      "loss": 0.371,
      "step": 26970
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.529855251312256,
      "learning_rate": 1.186554845407174e-06,
      "loss": 0.4585,
      "step": 26971
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.472065806388855,
      "learning_rate": 1.1860737184617178e-06,
      "loss": 0.3889,
      "step": 26972
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5487339496612549,
      "learning_rate": 1.1855926829307696e-06,
      "loss": 0.4359,
      "step": 26973
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5184203386306763,
      "learning_rate": 1.185111738819319e-06,
      "loss": 0.3769,
      "step": 26974
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5589089393615723,
      "learning_rate": 1.1846308861323551e-06,
      "loss": 0.4264,
      "step": 26975
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5025629997253418,
      "learning_rate": 1.1841501248748655e-06,
      "loss": 0.3536,
      "step": 26976
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.565114974975586,
      "learning_rate": 1.183669455051838e-06,
      "loss": 0.3901,
      "step": 26977
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5882205963134766,
      "learning_rate": 1.1831888766682576e-06,
      "loss": 0.4343,
      "step": 26978
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.494486927986145,
      "learning_rate": 1.1827083897291037e-06,
      "loss": 0.4416,
      "step": 26979
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5436288118362427,
      "learning_rate": 1.1822279942393643e-06,
      "loss": 0.3805,
      "step": 26980
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.534264326095581,
      "learning_rate": 1.1817476902040225e-06,
      "loss": 0.3992,
      "step": 26981
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.480954885482788,
      "learning_rate": 1.1812674776280585e-06,
      "loss": 0.3787,
      "step": 26982
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.503760576248169,
      "learning_rate": 1.1807873565164507e-06,
      "loss": 0.427,
      "step": 26983
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5216120481491089,
      "learning_rate": 1.1803073268741804e-06,
      "loss": 0.3812,
      "step": 26984
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5640071630477905,
      "learning_rate": 1.1798273887062272e-06,
      "loss": 0.4265,
      "step": 26985
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6037943363189697,
      "learning_rate": 1.1793475420175693e-06,
      "loss": 0.4126,
      "step": 26986
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4852015972137451,
      "learning_rate": 1.1788677868131826e-06,
      "loss": 0.4108,
      "step": 26987
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5088862180709839,
      "learning_rate": 1.178388123098041e-06,
      "loss": 0.3749,
      "step": 26988
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5518608093261719,
      "learning_rate": 1.1779085508771205e-06,
      "loss": 0.4123,
      "step": 26989
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5550485849380493,
      "learning_rate": 1.177429070155398e-06,
      "loss": 0.3909,
      "step": 26990
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4677332639694214,
      "learning_rate": 1.1769496809378411e-06,
      "loss": 0.4008,
      "step": 26991
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5478590726852417,
      "learning_rate": 1.1764703832294277e-06,
      "loss": 0.4073,
      "step": 26992
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5959413051605225,
      "learning_rate": 1.175991177035124e-06,
      "loss": 0.4475,
      "step": 26993
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4799386262893677,
      "learning_rate": 1.1755120623599036e-06,
      "loss": 0.3785,
      "step": 26994
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5995687246322632,
      "learning_rate": 1.1750330392087327e-06,
      "loss": 0.4121,
      "step": 26995
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5424796342849731,
      "learning_rate": 1.1745541075865807e-06,
      "loss": 0.3904,
      "step": 26996
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.676198959350586,
      "learning_rate": 1.174075267498418e-06,
      "loss": 0.463,
      "step": 26997
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5374361276626587,
      "learning_rate": 1.173596518949207e-06,
      "loss": 0.3563,
      "step": 26998
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4976447820663452,
      "learning_rate": 1.1731178619439131e-06,
      "loss": 0.398,
      "step": 26999
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.471139669418335,
      "learning_rate": 1.1726392964875012e-06,
      "loss": 0.3662,
      "step": 27000
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5011924505233765,
      "learning_rate": 1.1721608225849357e-06,
      "loss": 0.4122,
      "step": 27001
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.551224708557129,
      "learning_rate": 1.1716824402411808e-06,
      "loss": 0.3796,
      "step": 27002
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5523066520690918,
      "learning_rate": 1.171204149461196e-06,
      "loss": 0.4183,
      "step": 27003
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6111726760864258,
      "learning_rate": 1.1707259502499401e-06,
      "loss": 0.398,
      "step": 27004
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5091735124588013,
      "learning_rate": 1.1702478426123753e-06,
      "loss": 0.3953,
      "step": 27005
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5247005224227905,
      "learning_rate": 1.1697698265534617e-06,
      "loss": 0.4003,
      "step": 27006
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6047245264053345,
      "learning_rate": 1.1692919020781534e-06,
      "loss": 0.4271,
      "step": 27007
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6511423587799072,
      "learning_rate": 1.168814069191412e-06,
      "loss": 0.3752,
      "step": 27008
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5376261472702026,
      "learning_rate": 1.1683363278981864e-06,
      "loss": 0.4213,
      "step": 27009
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4770517349243164,
      "learning_rate": 1.1678586782034396e-06,
      "loss": 0.3469,
      "step": 27010
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6575102806091309,
      "learning_rate": 1.1673811201121177e-06,
      "loss": 0.4747,
      "step": 27011
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5765489339828491,
      "learning_rate": 1.1669036536291788e-06,
      "loss": 0.3911,
      "step": 27012
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7101346254348755,
      "learning_rate": 1.166426278759576e-06,
      "loss": 0.3964,
      "step": 27013
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5090891122817993,
      "learning_rate": 1.1659489955082582e-06,
      "loss": 0.377,
      "step": 27014
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.626007318496704,
      "learning_rate": 1.1654718038801727e-06,
      "loss": 0.425,
      "step": 27015
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5339016914367676,
      "learning_rate": 1.1649947038802733e-06,
      "loss": 0.3583,
      "step": 27016
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.596962809562683,
      "learning_rate": 1.1645176955135052e-06,
      "loss": 0.4179,
      "step": 27017
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4844969511032104,
      "learning_rate": 1.1640407787848206e-06,
      "loss": 0.3807,
      "step": 27018
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6529004573822021,
      "learning_rate": 1.1635639536991617e-06,
      "loss": 0.4394,
      "step": 27019
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5249214172363281,
      "learning_rate": 1.1630872202614729e-06,
      "loss": 0.365,
      "step": 27020
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6138242483139038,
      "learning_rate": 1.1626105784767005e-06,
      "loss": 0.4034,
      "step": 27021
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5739262104034424,
      "learning_rate": 1.1621340283497906e-06,
      "loss": 0.4124,
      "step": 27022
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.513126015663147,
      "learning_rate": 1.1616575698856824e-06,
      "loss": 0.4276,
      "step": 27023
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.501010537147522,
      "learning_rate": 1.1611812030893165e-06,
      "loss": 0.3674,
      "step": 27024
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5280053615570068,
      "learning_rate": 1.1607049279656368e-06,
      "loss": 0.3986,
      "step": 27025
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4730989933013916,
      "learning_rate": 1.1602287445195825e-06,
      "loss": 0.3791,
      "step": 27026
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5519450902938843,
      "learning_rate": 1.1597526527560899e-06,
      "loss": 0.4322,
      "step": 27027
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.484880805015564,
      "learning_rate": 1.1592766526801015e-06,
      "loss": 0.3654,
      "step": 27028
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.505731463432312,
      "learning_rate": 1.1588007442965477e-06,
      "loss": 0.4254,
      "step": 27029
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5043258666992188,
      "learning_rate": 1.1583249276103714e-06,
      "loss": 0.3815,
      "step": 27030
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5671240091323853,
      "learning_rate": 1.1578492026265009e-06,
      "loss": 0.4295,
      "step": 27031
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6215122938156128,
      "learning_rate": 1.1573735693498744e-06,
      "loss": 0.3708,
      "step": 27032
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5978412628173828,
      "learning_rate": 1.1568980277854257e-06,
      "loss": 0.4545,
      "step": 27033
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4961031675338745,
      "learning_rate": 1.1564225779380855e-06,
      "loss": 0.3773,
      "step": 27034
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5373164415359497,
      "learning_rate": 1.155947219812782e-06,
      "loss": 0.4343,
      "step": 27035
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5158147811889648,
      "learning_rate": 1.1554719534144488e-06,
      "loss": 0.3802,
      "step": 27036
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6177713871002197,
      "learning_rate": 1.1549967787480143e-06,
      "loss": 0.469,
      "step": 27037
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5621418952941895,
      "learning_rate": 1.1545216958184091e-06,
      "loss": 0.3677,
      "step": 27038
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5059518814086914,
      "learning_rate": 1.1540467046305592e-06,
      "loss": 0.4109,
      "step": 27039
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.568041205406189,
      "learning_rate": 1.1535718051893885e-06,
      "loss": 0.3775,
      "step": 27040
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.616911768913269,
      "learning_rate": 1.153096997499823e-06,
      "loss": 0.4339,
      "step": 27041
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4352220296859741,
      "learning_rate": 1.152622281566792e-06,
      "loss": 0.3605,
      "step": 27042
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4803671836853027,
      "learning_rate": 1.1521476573952128e-06,
      "loss": 0.4501,
      "step": 27043
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5537844896316528,
      "learning_rate": 1.1516731249900126e-06,
      "loss": 0.3901,
      "step": 27044
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.721320629119873,
      "learning_rate": 1.1511986843561096e-06,
      "loss": 0.417,
      "step": 27045
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5367951393127441,
      "learning_rate": 1.1507243354984288e-06,
      "loss": 0.3789,
      "step": 27046
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5065151453018188,
      "learning_rate": 1.150250078421884e-06,
      "loss": 0.4079,
      "step": 27047
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.479481816291809,
      "learning_rate": 1.1497759131313979e-06,
      "loss": 0.3187,
      "step": 27048
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.535436749458313,
      "learning_rate": 1.14930183963189e-06,
      "loss": 0.4224,
      "step": 27049
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4776859283447266,
      "learning_rate": 1.1488278579282753e-06,
      "loss": 0.3668,
      "step": 27050
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.628383755683899,
      "learning_rate": 1.1483539680254673e-06,
      "loss": 0.4524,
      "step": 27051
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5566966533660889,
      "learning_rate": 1.1478801699283837e-06,
      "loss": 0.373,
      "step": 27052
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5333116054534912,
      "learning_rate": 1.147406463641937e-06,
      "loss": 0.3831,
      "step": 27053
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4774539470672607,
      "learning_rate": 1.1469328491710429e-06,
      "loss": 0.356,
      "step": 27054
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.732573390007019,
      "learning_rate": 1.1464593265206136e-06,
      "loss": 0.4179,
      "step": 27055
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5277758836746216,
      "learning_rate": 1.145985895695555e-06,
      "loss": 0.3801,
      "step": 27056
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.534466028213501,
      "learning_rate": 1.1455125567007818e-06,
      "loss": 0.3968,
      "step": 27057
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5942275524139404,
      "learning_rate": 1.1450393095412039e-06,
      "loss": 0.4092,
      "step": 27058
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5032235383987427,
      "learning_rate": 1.1445661542217268e-06,
      "loss": 0.4061,
      "step": 27059
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.573992371559143,
      "learning_rate": 1.1440930907472615e-06,
      "loss": 0.3988,
      "step": 27060
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5840963125228882,
      "learning_rate": 1.1436201191227092e-06,
      "loss": 0.4149,
      "step": 27061
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6590564250946045,
      "learning_rate": 1.1431472393529807e-06,
      "loss": 0.3817,
      "step": 27062
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5669176578521729,
      "learning_rate": 1.1426744514429767e-06,
      "loss": 0.4216,
      "step": 27063
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5136710405349731,
      "learning_rate": 1.1422017553976027e-06,
      "loss": 0.3724,
      "step": 27064
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7048033475875854,
      "learning_rate": 1.1417291512217598e-06,
      "loss": 0.9298,
      "step": 27065
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4400252103805542,
      "learning_rate": 1.1412566389203528e-06,
      "loss": 1.0234,
      "step": 27066
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6547505855560303,
      "learning_rate": 1.1407842184982786e-06,
      "loss": 0.4282,
      "step": 27067
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.487792730331421,
      "learning_rate": 1.1403118899604381e-06,
      "loss": 0.364,
      "step": 27068
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.656256079673767,
      "learning_rate": 1.139839653311733e-06,
      "loss": 0.456,
      "step": 27069
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.543898105621338,
      "learning_rate": 1.1393675085570588e-06,
      "loss": 0.3747,
      "step": 27070
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5933187007904053,
      "learning_rate": 1.13889545570131e-06,
      "loss": 0.4416,
      "step": 27071
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.526471495628357,
      "learning_rate": 1.1384234947493856e-06,
      "loss": 0.3723,
      "step": 27072
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4998142719268799,
      "learning_rate": 1.1379516257061795e-06,
      "loss": 0.4378,
      "step": 27073
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5408755540847778,
      "learning_rate": 1.1374798485765881e-06,
      "loss": 0.4182,
      "step": 27074
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5412160158157349,
      "learning_rate": 1.1370081633655027e-06,
      "loss": 0.4154,
      "step": 27075
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5211282968521118,
      "learning_rate": 1.136536570077813e-06,
      "loss": 0.3772,
      "step": 27076
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5149294137954712,
      "learning_rate": 1.1360650687184126e-06,
      "loss": 0.4081,
      "step": 27077
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5663455724716187,
      "learning_rate": 1.1355936592921946e-06,
      "loss": 0.3719,
      "step": 27078
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5407984256744385,
      "learning_rate": 1.1351223418040424e-06,
      "loss": 0.3961,
      "step": 27079
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4882166385650635,
      "learning_rate": 1.134651116258849e-06,
      "loss": 0.3811,
      "step": 27080
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5346970558166504,
      "learning_rate": 1.1341799826614975e-06,
      "loss": 0.3866,
      "step": 27081
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5104066133499146,
      "learning_rate": 1.13370894101688e-06,
      "loss": 0.367,
      "step": 27082
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4866193532943726,
      "learning_rate": 1.1332379913298751e-06,
      "loss": 0.4224,
      "step": 27083
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6124171018600464,
      "learning_rate": 1.1327671336053724e-06,
      "loss": 0.4065,
      "step": 27084
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5210700035095215,
      "learning_rate": 1.1322963678482556e-06,
      "loss": 0.4172,
      "step": 27085
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5487487316131592,
      "learning_rate": 1.1318256940634054e-06,
      "loss": 0.3985,
      "step": 27086
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5382370948791504,
      "learning_rate": 1.131355112255702e-06,
      "loss": 0.4211,
      "step": 27087
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5164095163345337,
      "learning_rate": 1.1308846224300274e-06,
      "loss": 0.3678,
      "step": 27088
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5413522720336914,
      "learning_rate": 1.130414224591262e-06,
      "loss": 0.4188,
      "step": 27089
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6305817365646362,
      "learning_rate": 1.1299439187442851e-06,
      "loss": 0.3872,
      "step": 27090
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5861084461212158,
      "learning_rate": 1.1294737048939751e-06,
      "loss": 0.4209,
      "step": 27091
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4902684688568115,
      "learning_rate": 1.1290035830452052e-06,
      "loss": 0.363,
      "step": 27092
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.631646990776062,
      "learning_rate": 1.1285335532028529e-06,
      "loss": 0.4879,
      "step": 27093
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5091211795806885,
      "learning_rate": 1.1280636153717972e-06,
      "loss": 0.3682,
      "step": 27094
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5508953332901,
      "learning_rate": 1.127593769556905e-06,
      "loss": 0.4419,
      "step": 27095
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4503268003463745,
      "learning_rate": 1.127124015763057e-06,
      "loss": 0.3372,
      "step": 27096
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8068442344665527,
      "learning_rate": 1.126654353995118e-06,
      "loss": 0.9538,
      "step": 27097
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3401644229888916,
      "learning_rate": 1.1261847842579653e-06,
      "loss": 0.9551,
      "step": 27098
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5204020738601685,
      "learning_rate": 1.1257153065564641e-06,
      "loss": 0.4009,
      "step": 27099
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5095831155776978,
      "learning_rate": 1.1252459208954868e-06,
      "loss": 0.3598,
      "step": 27100
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.502811074256897,
      "learning_rate": 1.1247766272799e-06,
      "loss": 0.3834,
      "step": 27101
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5164473056793213,
      "learning_rate": 1.1243074257145736e-06,
      "loss": 0.3794,
      "step": 27102
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4669396877288818,
      "learning_rate": 1.1238383162043698e-06,
      "loss": 0.3997,
      "step": 27103
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.03983998298645,
      "learning_rate": 1.1233692987541568e-06,
      "loss": 0.3624,
      "step": 27104
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5621998310089111,
      "learning_rate": 1.1229003733688003e-06,
      "loss": 0.4763,
      "step": 27105
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5557527542114258,
      "learning_rate": 1.1224315400531626e-06,
      "loss": 0.3897,
      "step": 27106
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4646985530853271,
      "learning_rate": 1.121962798812104e-06,
      "loss": 0.4307,
      "step": 27107
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.530038833618164,
      "learning_rate": 1.1214941496504872e-06,
      "loss": 0.37,
      "step": 27108
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5772594213485718,
      "learning_rate": 1.1210255925731728e-06,
      "loss": 0.4372,
      "step": 27109
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5517492294311523,
      "learning_rate": 1.1205571275850246e-06,
      "loss": 0.3873,
      "step": 27110
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5842684507369995,
      "learning_rate": 1.1200887546908978e-06,
      "loss": 0.3899,
      "step": 27111
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5411313772201538,
      "learning_rate": 1.1196204738956472e-06,
      "loss": 0.4035,
      "step": 27112
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.58979332447052,
      "learning_rate": 1.1191522852041337e-06,
      "loss": 0.4199,
      "step": 27113
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.507253885269165,
      "learning_rate": 1.1186841886212142e-06,
      "loss": 0.3757,
      "step": 27114
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6105293035507202,
      "learning_rate": 1.1182161841517392e-06,
      "loss": 0.3985,
      "step": 27115
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.679871916770935,
      "learning_rate": 1.1177482718005672e-06,
      "loss": 0.3716,
      "step": 27116
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5750672817230225,
      "learning_rate": 1.1172804515725466e-06,
      "loss": 0.4487,
      "step": 27117
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.463474154472351,
      "learning_rate": 1.1168127234725346e-06,
      "loss": 0.3513,
      "step": 27118
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.534752368927002,
      "learning_rate": 1.116345087505377e-06,
      "loss": 0.4295,
      "step": 27119
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5559033155441284,
      "learning_rate": 1.1158775436759274e-06,
      "loss": 0.422,
      "step": 27120
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5094712972640991,
      "learning_rate": 1.1154100919890353e-06,
      "loss": 0.4029,
      "step": 27121
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6483620405197144,
      "learning_rate": 1.1149427324495487e-06,
      "loss": 0.4263,
      "step": 27122
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.543089747428894,
      "learning_rate": 1.1144754650623113e-06,
      "loss": 0.4176,
      "step": 27123
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4698054790496826,
      "learning_rate": 1.1140082898321713e-06,
      "loss": 0.3696,
      "step": 27124
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5816559791564941,
      "learning_rate": 1.113541206763975e-06,
      "loss": 0.4533,
      "step": 27125
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5370858907699585,
      "learning_rate": 1.1130742158625696e-06,
      "loss": 0.3538,
      "step": 27126
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.492521047592163,
      "learning_rate": 1.1126073171327946e-06,
      "loss": 0.3587,
      "step": 27127
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5355398654937744,
      "learning_rate": 1.1121405105794913e-06,
      "loss": 0.364,
      "step": 27128
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5130162239074707,
      "learning_rate": 1.1116737962075041e-06,
      "loss": 0.3929,
      "step": 27129
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.49796724319458,
      "learning_rate": 1.1112071740216745e-06,
      "loss": 0.3502,
      "step": 27130
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5923030376434326,
      "learning_rate": 1.1107406440268377e-06,
      "loss": 0.4071,
      "step": 27131
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.510806679725647,
      "learning_rate": 1.1102742062278381e-06,
      "loss": 0.367,
      "step": 27132
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5364747047424316,
      "learning_rate": 1.1098078606295071e-06,
      "loss": 0.3958,
      "step": 27133
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.525638461112976,
      "learning_rate": 1.109341607236688e-06,
      "loss": 0.3767,
      "step": 27134
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.546770691871643,
      "learning_rate": 1.1088754460542107e-06,
      "loss": 0.4071,
      "step": 27135
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5348241329193115,
      "learning_rate": 1.108409377086913e-06,
      "loss": 0.3635,
      "step": 27136
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.515103816986084,
      "learning_rate": 1.1079434003396294e-06,
      "loss": 0.4124,
      "step": 27137
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.475854516029358,
      "learning_rate": 1.1074775158171935e-06,
      "loss": 0.3599,
      "step": 27138
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5320873260498047,
      "learning_rate": 1.1070117235244326e-06,
      "loss": 0.4183,
      "step": 27139
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6713017225265503,
      "learning_rate": 1.1065460234661808e-06,
      "loss": 0.4091,
      "step": 27140
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6059623956680298,
      "learning_rate": 1.1060804156472705e-06,
      "loss": 0.4266,
      "step": 27141
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.435685396194458,
      "learning_rate": 1.105614900072528e-06,
      "loss": 0.3476,
      "step": 27142
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9024391174316406,
      "learning_rate": 1.1051494767467797e-06,
      "loss": 0.4376,
      "step": 27143
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4544506072998047,
      "learning_rate": 1.1046841456748546e-06,
      "loss": 0.3464,
      "step": 27144
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6000827550888062,
      "learning_rate": 1.1042189068615804e-06,
      "loss": 0.4285,
      "step": 27145
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5222333669662476,
      "learning_rate": 1.1037537603117821e-06,
      "loss": 0.4063,
      "step": 27146
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5801674127578735,
      "learning_rate": 1.1032887060302833e-06,
      "loss": 0.4205,
      "step": 27147
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4793710708618164,
      "learning_rate": 1.1028237440219058e-06,
      "loss": 0.3583,
      "step": 27148
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6193270683288574,
      "learning_rate": 1.1023588742914726e-06,
      "loss": 0.4661,
      "step": 27149
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5253759622573853,
      "learning_rate": 1.1018940968438085e-06,
      "loss": 0.3992,
      "step": 27150
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5626164674758911,
      "learning_rate": 1.1014294116837299e-06,
      "loss": 0.4107,
      "step": 27151
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.631954550743103,
      "learning_rate": 1.1009648188160593e-06,
      "loss": 0.3967,
      "step": 27152
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8292890787124634,
      "learning_rate": 1.100500318245611e-06,
      "loss": 0.9704,
      "step": 27153
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5153486728668213,
      "learning_rate": 1.1000359099772083e-06,
      "loss": 1.0056,
      "step": 27154
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.554444432258606,
      "learning_rate": 1.0995715940156626e-06,
      "loss": 0.42,
      "step": 27155
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5472493171691895,
      "learning_rate": 1.099107370365793e-06,
      "loss": 0.3793,
      "step": 27156
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5818084478378296,
      "learning_rate": 1.0986432390324142e-06,
      "loss": 0.4682,
      "step": 27157
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4663639068603516,
      "learning_rate": 1.0981792000203394e-06,
      "loss": 0.3732,
      "step": 27158
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5619853734970093,
      "learning_rate": 1.0977152533343782e-06,
      "loss": 0.4108,
      "step": 27159
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5104955434799194,
      "learning_rate": 1.0972513989793464e-06,
      "loss": 0.3524,
      "step": 27160
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4838664531707764,
      "learning_rate": 1.0967876369600528e-06,
      "loss": 0.3955,
      "step": 27161
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4197322130203247,
      "learning_rate": 1.0963239672813109e-06,
      "loss": 0.3527,
      "step": 27162
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7009000778198242,
      "learning_rate": 1.095860389947928e-06,
      "loss": 0.893,
      "step": 27163
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.383044719696045,
      "learning_rate": 1.0953969049647084e-06,
      "loss": 0.9726,
      "step": 27164
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5878576040267944,
      "learning_rate": 1.0949335123364613e-06,
      "loss": 0.4201,
      "step": 27165
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4565954208374023,
      "learning_rate": 1.0944702120679973e-06,
      "loss": 0.3548,
      "step": 27166
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9818986654281616,
      "learning_rate": 1.0940070041641148e-06,
      "loss": 0.4616,
      "step": 27167
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.207805871963501,
      "learning_rate": 1.0935438886296235e-06,
      "loss": 0.3752,
      "step": 27168
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7086008787155151,
      "learning_rate": 1.0930808654693214e-06,
      "loss": 0.4574,
      "step": 27169
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.503706455230713,
      "learning_rate": 1.092617934688016e-06,
      "loss": 0.3917,
      "step": 27170
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5455055236816406,
      "learning_rate": 1.0921550962905048e-06,
      "loss": 0.4671,
      "step": 27171
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4954487085342407,
      "learning_rate": 1.091692350281589e-06,
      "loss": 0.3638,
      "step": 27172
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6190212965011597,
      "learning_rate": 1.0912296966660717e-06,
      "loss": 0.4124,
      "step": 27173
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.462541937828064,
      "learning_rate": 1.090767135448747e-06,
      "loss": 0.3416,
      "step": 27174
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4948772192001343,
      "learning_rate": 1.090304666634412e-06,
      "loss": 0.4092,
      "step": 27175
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.58010995388031,
      "learning_rate": 1.089842290227865e-06,
      "loss": 0.4016,
      "step": 27176
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5452556610107422,
      "learning_rate": 1.0893800062339022e-06,
      "loss": 0.4151,
      "step": 27177
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4753024578094482,
      "learning_rate": 1.088917814657321e-06,
      "loss": 0.3624,
      "step": 27178
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4756499528884888,
      "learning_rate": 1.0884557155029074e-06,
      "loss": 0.3817,
      "step": 27179
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4801279306411743,
      "learning_rate": 1.087993708775459e-06,
      "loss": 0.3499,
      "step": 27180
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5492470264434814,
      "learning_rate": 1.087531794479766e-06,
      "loss": 0.4068,
      "step": 27181
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.452337622642517,
      "learning_rate": 1.0870699726206224e-06,
      "loss": 0.3517,
      "step": 27182
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5142103433609009,
      "learning_rate": 1.0866082432028157e-06,
      "loss": 0.3811,
      "step": 27183
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5199313163757324,
      "learning_rate": 1.0861466062311321e-06,
      "loss": 0.3922,
      "step": 27184
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.501051425933838,
      "learning_rate": 1.085685061710363e-06,
      "loss": 0.3995,
      "step": 27185
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4454900026321411,
      "learning_rate": 1.0852236096452962e-06,
      "loss": 0.3554,
      "step": 27186
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5860179662704468,
      "learning_rate": 1.084762250040714e-06,
      "loss": 0.4381,
      "step": 27187
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5122740268707275,
      "learning_rate": 1.0843009829014062e-06,
      "loss": 0.3716,
      "step": 27188
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.487254023551941,
      "learning_rate": 1.083839808232151e-06,
      "loss": 0.3869,
      "step": 27189
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4734046459197998,
      "learning_rate": 1.083378726037737e-06,
      "loss": 0.4026,
      "step": 27190
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6539171934127808,
      "learning_rate": 1.0829177363229425e-06,
      "loss": 0.4433,
      "step": 27191
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.575541615486145,
      "learning_rate": 1.0824568390925505e-06,
      "loss": 0.3541,
      "step": 27192
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.871013879776001,
      "learning_rate": 1.0819960343513425e-06,
      "loss": 0.5301,
      "step": 27193
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5909148454666138,
      "learning_rate": 1.0815353221040959e-06,
      "loss": 0.381,
      "step": 27194
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5272196531295776,
      "learning_rate": 1.0810747023555879e-06,
      "loss": 0.4292,
      "step": 27195
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4506186246871948,
      "learning_rate": 1.0806141751105958e-06,
      "loss": 0.3619,
      "step": 27196
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6073390245437622,
      "learning_rate": 1.0801537403738994e-06,
      "loss": 0.4916,
      "step": 27197
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4834872484207153,
      "learning_rate": 1.0796933981502734e-06,
      "loss": 0.3623,
      "step": 27198
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9410463571548462,
      "learning_rate": 1.0792331484444917e-06,
      "loss": 1.0363,
      "step": 27199
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4357954263687134,
      "learning_rate": 1.0787729912613242e-06,
      "loss": 1.0006,
      "step": 27200
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5719064474105835,
      "learning_rate": 1.078312926605548e-06,
      "loss": 0.4039,
      "step": 27201
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5449862480163574,
      "learning_rate": 1.077852954481934e-06,
      "loss": 0.3803,
      "step": 27202
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.514789342880249,
      "learning_rate": 1.077393074895251e-06,
      "loss": 0.4211,
      "step": 27203
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4781882762908936,
      "learning_rate": 1.0769332878502713e-06,
      "loss": 0.3732,
      "step": 27204
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6618460416793823,
      "learning_rate": 1.07647359335176e-06,
      "loss": 0.4627,
      "step": 27205
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5716228485107422,
      "learning_rate": 1.0760139914044887e-06,
      "loss": 0.389,
      "step": 27206
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.48861563205719,
      "learning_rate": 1.0755544820132213e-06,
      "loss": 0.3894,
      "step": 27207
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4995397329330444,
      "learning_rate": 1.0750950651827241e-06,
      "loss": 0.3791,
      "step": 27208
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6555812358856201,
      "learning_rate": 1.0746357409177654e-06,
      "loss": 0.4033,
      "step": 27209
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5264333486557007,
      "learning_rate": 1.074176509223106e-06,
      "loss": 0.3532,
      "step": 27210
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.617978811264038,
      "learning_rate": 1.0737173701035075e-06,
      "loss": 0.3979,
      "step": 27211
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5262500047683716,
      "learning_rate": 1.073258323563734e-06,
      "loss": 0.3876,
      "step": 27212
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.729297161102295,
      "learning_rate": 1.072799369608546e-06,
      "loss": 0.4602,
      "step": 27213
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5340265035629272,
      "learning_rate": 1.0723405082427074e-06,
      "loss": 0.3748,
      "step": 27214
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5054265260696411,
      "learning_rate": 1.0718817394709702e-06,
      "loss": 0.3969,
      "step": 27215
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.487045168876648,
      "learning_rate": 1.071423063298096e-06,
      "loss": 0.3539,
      "step": 27216
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6396658420562744,
      "learning_rate": 1.0709644797288433e-06,
      "loss": 0.4341,
      "step": 27217
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4755539894104004,
      "learning_rate": 1.070505988767968e-06,
      "loss": 0.3656,
      "step": 27218
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.604399561882019,
      "learning_rate": 1.0700475904202244e-06,
      "loss": 0.4085,
      "step": 27219
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6463367938995361,
      "learning_rate": 1.0695892846903655e-06,
      "loss": 0.4054,
      "step": 27220
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4951916933059692,
      "learning_rate": 1.069131071583146e-06,
      "loss": 0.4101,
      "step": 27221
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5327730178833008,
      "learning_rate": 1.06867295110332e-06,
      "loss": 0.3764,
      "step": 27222
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.503978967666626,
      "learning_rate": 1.068214923255636e-06,
      "loss": 0.3659,
      "step": 27223
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.612699270248413,
      "learning_rate": 1.0677569880448479e-06,
      "loss": 0.362,
      "step": 27224
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.2807648181915283,
      "learning_rate": 1.067299145475702e-06,
      "loss": 0.4367,
      "step": 27225
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5187807083129883,
      "learning_rate": 1.0668413955529467e-06,
      "loss": 0.3898,
      "step": 27226
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5195428133010864,
      "learning_rate": 1.0663837382813335e-06,
      "loss": 0.4359,
      "step": 27227
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4212912321090698,
      "learning_rate": 1.0659261736656046e-06,
      "loss": 0.3576,
      "step": 27228
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6191295385360718,
      "learning_rate": 1.065468701710509e-06,
      "loss": 0.453,
      "step": 27229
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.452087640762329,
      "learning_rate": 1.0650113224207892e-06,
      "loss": 0.3555,
      "step": 27230
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6052812337875366,
      "learning_rate": 1.0645540358011908e-06,
      "loss": 0.4239,
      "step": 27231
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.485446810722351,
      "learning_rate": 1.0640968418564534e-06,
      "loss": 0.356,
      "step": 27232
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5277115106582642,
      "learning_rate": 1.0636397405913213e-06,
      "loss": 0.5171,
      "step": 27233
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5038130283355713,
      "learning_rate": 1.063182732010537e-06,
      "loss": 0.3905,
      "step": 27234
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.588974118232727,
      "learning_rate": 1.0627258161188392e-06,
      "loss": 0.4366,
      "step": 27235
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5306200981140137,
      "learning_rate": 1.0622689929209628e-06,
      "loss": 0.3682,
      "step": 27236
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5265570878982544,
      "learning_rate": 1.0618122624216498e-06,
      "loss": 0.4164,
      "step": 27237
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.545037865638733,
      "learning_rate": 1.0613556246256373e-06,
      "loss": 0.4039,
      "step": 27238
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5867351293563843,
      "learning_rate": 1.0608990795376616e-06,
      "loss": 0.4898,
      "step": 27239
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5194817781448364,
      "learning_rate": 1.0604426271624569e-06,
      "loss": 0.3726,
      "step": 27240
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5327783823013306,
      "learning_rate": 1.0599862675047557e-06,
      "loss": 0.4162,
      "step": 27241
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5216768980026245,
      "learning_rate": 1.059530000569292e-06,
      "loss": 0.3839,
      "step": 27242
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5654473304748535,
      "learning_rate": 1.0590738263608014e-06,
      "loss": 0.424,
      "step": 27243
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5592670440673828,
      "learning_rate": 1.0586177448840096e-06,
      "loss": 0.3795,
      "step": 27244
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5971423387527466,
      "learning_rate": 1.058161756143653e-06,
      "loss": 0.4228,
      "step": 27245
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4653314352035522,
      "learning_rate": 1.0577058601444546e-06,
      "loss": 0.3622,
      "step": 27246
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6518316268920898,
      "learning_rate": 1.057250056891148e-06,
      "loss": 0.4431,
      "step": 27247
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.471954584121704,
      "learning_rate": 1.0567943463884578e-06,
      "loss": 0.3839,
      "step": 27248
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5535882711410522,
      "learning_rate": 1.0563387286411097e-06,
      "loss": 0.4386,
      "step": 27249
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5330314636230469,
      "learning_rate": 1.0558832036538335e-06,
      "loss": 0.3779,
      "step": 27250
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5760704278945923,
      "learning_rate": 1.05542777143135e-06,
      "loss": 0.4285,
      "step": 27251
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4892354011535645,
      "learning_rate": 1.0549724319783816e-06,
      "loss": 0.3686,
      "step": 27252
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4547452926635742,
      "learning_rate": 1.0545171852996538e-06,
      "loss": 0.4044,
      "step": 27253
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5498582124710083,
      "learning_rate": 1.0540620313998872e-06,
      "loss": 0.3868,
      "step": 27254
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5960657596588135,
      "learning_rate": 1.0536069702838058e-06,
      "loss": 0.4387,
      "step": 27255
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4953683614730835,
      "learning_rate": 1.0531520019561225e-06,
      "loss": 0.3652,
      "step": 27256
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.55448317527771,
      "learning_rate": 1.05269712642156e-06,
      "loss": 0.4214,
      "step": 27257
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.490133285522461,
      "learning_rate": 1.0522423436848361e-06,
      "loss": 0.3738,
      "step": 27258
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.516785740852356,
      "learning_rate": 1.0517876537506688e-06,
      "loss": 0.4034,
      "step": 27259
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4650191068649292,
      "learning_rate": 1.0513330566237723e-06,
      "loss": 0.3442,
      "step": 27260
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5599427223205566,
      "learning_rate": 1.0508785523088594e-06,
      "loss": 0.4068,
      "step": 27261
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5796021223068237,
      "learning_rate": 1.0504241408106464e-06,
      "loss": 0.3716,
      "step": 27262
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5652587413787842,
      "learning_rate": 1.0499698221338483e-06,
      "loss": 0.3888,
      "step": 27263
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5919818878173828,
      "learning_rate": 1.0495155962831738e-06,
      "loss": 0.3904,
      "step": 27264
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5401467084884644,
      "learning_rate": 1.0490614632633357e-06,
      "loss": 0.4095,
      "step": 27265
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4295381307601929,
      "learning_rate": 1.048607423079042e-06,
      "loss": 0.3739,
      "step": 27266
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5999120473861694,
      "learning_rate": 1.0481534757350066e-06,
      "loss": 0.4286,
      "step": 27267
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5244414806365967,
      "learning_rate": 1.0476996212359304e-06,
      "loss": 0.402,
      "step": 27268
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5703524351119995,
      "learning_rate": 1.0472458595865253e-06,
      "loss": 0.4168,
      "step": 27269
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.485355257987976,
      "learning_rate": 1.046792190791499e-06,
      "loss": 0.3646,
      "step": 27270
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5275083780288696,
      "learning_rate": 1.0463386148555555e-06,
      "loss": 0.4388,
      "step": 27271
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6069937944412231,
      "learning_rate": 1.0458851317833951e-06,
      "loss": 0.4074,
      "step": 27272
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6314213275909424,
      "learning_rate": 1.0454317415797244e-06,
      "loss": 0.4787,
      "step": 27273
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.534728765487671,
      "learning_rate": 1.044978444249245e-06,
      "loss": 0.396,
      "step": 27274
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6148077249526978,
      "learning_rate": 1.0445252397966621e-06,
      "loss": 0.4174,
      "step": 27275
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.506585717201233,
      "learning_rate": 1.044072128226672e-06,
      "loss": 0.3723,
      "step": 27276
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5719733238220215,
      "learning_rate": 1.0436191095439741e-06,
      "loss": 0.4396,
      "step": 27277
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4822112321853638,
      "learning_rate": 1.0431661837532669e-06,
      "loss": 0.3864,
      "step": 27278
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5710872411727905,
      "learning_rate": 1.0427133508592524e-06,
      "loss": 0.4389,
      "step": 27279
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5115700960159302,
      "learning_rate": 1.042260610866621e-06,
      "loss": 0.3778,
      "step": 27280
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4885213375091553,
      "learning_rate": 1.0418079637800738e-06,
      "loss": 0.4171,
      "step": 27281
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5079519748687744,
      "learning_rate": 1.0413554096043e-06,
      "loss": 0.3899,
      "step": 27282
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6760538816452026,
      "learning_rate": 1.040902948343999e-06,
      "loss": 0.4746,
      "step": 27283
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.621984601020813,
      "learning_rate": 1.0404505800038577e-06,
      "loss": 0.4311,
      "step": 27284
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.610120415687561,
      "learning_rate": 1.039998304588572e-06,
      "loss": 0.393,
      "step": 27285
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6293002367019653,
      "learning_rate": 1.0395461221028336e-06,
      "loss": 0.385,
      "step": 27286
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5455169677734375,
      "learning_rate": 1.0390940325513298e-06,
      "loss": 0.4376,
      "step": 27287
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5463608503341675,
      "learning_rate": 1.0386420359387484e-06,
      "loss": 0.3718,
      "step": 27288
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5668319463729858,
      "learning_rate": 1.0381901322697796e-06,
      "loss": 0.4268,
      "step": 27289
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4703141450881958,
      "learning_rate": 1.0377383215491088e-06,
      "loss": 0.3753,
      "step": 27290
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6724778413772583,
      "learning_rate": 1.0372866037814277e-06,
      "loss": 0.4212,
      "step": 27291
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6014331579208374,
      "learning_rate": 1.0368349789714128e-06,
      "loss": 0.3662,
      "step": 27292
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5477087497711182,
      "learning_rate": 1.036383447123751e-06,
      "loss": 0.4092,
      "step": 27293
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4887124300003052,
      "learning_rate": 1.0359320082431268e-06,
      "loss": 0.3795,
      "step": 27294
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5248242616653442,
      "learning_rate": 1.035480662334225e-06,
      "loss": 0.4127,
      "step": 27295
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4998201131820679,
      "learning_rate": 1.0350294094017222e-06,
      "loss": 0.3822,
      "step": 27296
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5914479494094849,
      "learning_rate": 1.0345782494502988e-06,
      "loss": 0.4624,
      "step": 27297
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5213040113449097,
      "learning_rate": 1.0341271824846355e-06,
      "loss": 0.3986,
      "step": 27298
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.490113615989685,
      "learning_rate": 1.033676208509412e-06,
      "loss": 0.4266,
      "step": 27299
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4588948488235474,
      "learning_rate": 1.0332253275293024e-06,
      "loss": 0.3667,
      "step": 27300
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6397614479064941,
      "learning_rate": 1.0327745395489874e-06,
      "loss": 0.4272,
      "step": 27301
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6147277355194092,
      "learning_rate": 1.0323238445731365e-06,
      "loss": 0.4041,
      "step": 27302
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6960612535476685,
      "learning_rate": 1.0318732426064304e-06,
      "loss": 0.4602,
      "step": 27303
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4639536142349243,
      "learning_rate": 1.0314227336535376e-06,
      "loss": 0.349,
      "step": 27304
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5449576377868652,
      "learning_rate": 1.0309723177191323e-06,
      "loss": 0.4386,
      "step": 27305
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5275262594223022,
      "learning_rate": 1.030521994807887e-06,
      "loss": 0.3886,
      "step": 27306
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5358266830444336,
      "learning_rate": 1.0300717649244729e-06,
      "loss": 0.3937,
      "step": 27307
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4426857233047485,
      "learning_rate": 1.029621628073556e-06,
      "loss": 0.3427,
      "step": 27308
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5533897876739502,
      "learning_rate": 1.0291715842598071e-06,
      "loss": 0.4415,
      "step": 27309
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5422805547714233,
      "learning_rate": 1.0287216334878935e-06,
      "loss": 0.4082,
      "step": 27310
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5777928829193115,
      "learning_rate": 1.0282717757624848e-06,
      "loss": 0.4264,
      "step": 27311
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6041786670684814,
      "learning_rate": 1.0278220110882452e-06,
      "loss": 0.3926,
      "step": 27312
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5527443885803223,
      "learning_rate": 1.0273723394698354e-06,
      "loss": 0.3925,
      "step": 27313
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4614654779434204,
      "learning_rate": 1.0269227609119225e-06,
      "loss": 0.3548,
      "step": 27314
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5263993740081787,
      "learning_rate": 1.026473275419172e-06,
      "loss": 0.4109,
      "step": 27315
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5509730577468872,
      "learning_rate": 1.02602388299624e-06,
      "loss": 0.3776,
      "step": 27316
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7579796314239502,
      "learning_rate": 1.0255745836477926e-06,
      "loss": 0.4666,
      "step": 27317
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4318314790725708,
      "learning_rate": 1.0251253773784865e-06,
      "loss": 0.3638,
      "step": 27318
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5393654108047485,
      "learning_rate": 1.024676264192983e-06,
      "loss": 0.3871,
      "step": 27319
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.524567723274231,
      "learning_rate": 1.0242272440959377e-06,
      "loss": 0.3841,
      "step": 27320
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.55299973487854,
      "learning_rate": 1.023778317092008e-06,
      "loss": 0.4547,
      "step": 27321
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.574550986289978,
      "learning_rate": 1.0233294831858531e-06,
      "loss": 0.4169,
      "step": 27322
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.484093189239502,
      "learning_rate": 1.0228807423821262e-06,
      "loss": 0.4035,
      "step": 27323
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.529360055923462,
      "learning_rate": 1.0224320946854783e-06,
      "loss": 0.359,
      "step": 27324
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.542528748512268,
      "learning_rate": 1.0219835401005662e-06,
      "loss": 0.4351,
      "step": 27325
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4856860637664795,
      "learning_rate": 1.021535078632041e-06,
      "loss": 0.3636,
      "step": 27326
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7100861072540283,
      "learning_rate": 1.0210867102845568e-06,
      "loss": 0.4163,
      "step": 27327
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5287370681762695,
      "learning_rate": 1.0206384350627607e-06,
      "loss": 0.3837,
      "step": 27328
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5966455936431885,
      "learning_rate": 1.0201902529713016e-06,
      "loss": 0.4578,
      "step": 27329
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4998832941055298,
      "learning_rate": 1.0197421640148286e-06,
      "loss": 0.3834,
      "step": 27330
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.497040033340454,
      "learning_rate": 1.0192941681979918e-06,
      "loss": 0.3817,
      "step": 27331
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5018281936645508,
      "learning_rate": 1.018846265525435e-06,
      "loss": 0.4005,
      "step": 27332
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6214962005615234,
      "learning_rate": 1.0183984560018022e-06,
      "loss": 0.4043,
      "step": 27333
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5563994646072388,
      "learning_rate": 1.017950739631739e-06,
      "loss": 0.3741,
      "step": 27334
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.558437466621399,
      "learning_rate": 1.0175031164198923e-06,
      "loss": 0.4074,
      "step": 27335
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.538191318511963,
      "learning_rate": 1.017055586370901e-06,
      "loss": 0.3599,
      "step": 27336
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6025209426879883,
      "learning_rate": 1.0166081494894075e-06,
      "loss": 0.4357,
      "step": 27337
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5152039527893066,
      "learning_rate": 1.016160805780052e-06,
      "loss": 0.3698,
      "step": 27338
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.677425742149353,
      "learning_rate": 1.0157135552474772e-06,
      "loss": 0.9671,
      "step": 27339
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.440813422203064,
      "learning_rate": 1.0152663978963174e-06,
      "loss": 1.0464,
      "step": 27340
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6575549840927124,
      "learning_rate": 1.0148193337312117e-06,
      "loss": 0.4581,
      "step": 27341
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5256191492080688,
      "learning_rate": 1.0143723627567993e-06,
      "loss": 0.362,
      "step": 27342
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4792513847351074,
      "learning_rate": 1.0139254849777147e-06,
      "loss": 0.3899,
      "step": 27343
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5611755847930908,
      "learning_rate": 1.0134787003985901e-06,
      "loss": 0.415,
      "step": 27344
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4783415794372559,
      "learning_rate": 1.013032009024062e-06,
      "loss": 0.3924,
      "step": 27345
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5429558753967285,
      "learning_rate": 1.012585410858763e-06,
      "loss": 0.376,
      "step": 27346
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.565319538116455,
      "learning_rate": 1.012138905907326e-06,
      "loss": 0.3981,
      "step": 27347
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4998835325241089,
      "learning_rate": 1.011692494174381e-06,
      "loss": 0.3644,
      "step": 27348
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4799691438674927,
      "learning_rate": 1.0112461756645565e-06,
      "loss": 0.4185,
      "step": 27349
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6279627084732056,
      "learning_rate": 1.0107999503824828e-06,
      "loss": 0.3995,
      "step": 27350
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.547634243965149,
      "learning_rate": 1.0103538183327898e-06,
      "loss": 0.4694,
      "step": 27351
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4730242490768433,
      "learning_rate": 1.0099077795201007e-06,
      "loss": 0.3784,
      "step": 27352
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5119552612304688,
      "learning_rate": 1.009461833949047e-06,
      "loss": 0.4136,
      "step": 27353
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5212386846542358,
      "learning_rate": 1.0090159816242472e-06,
      "loss": 0.3719,
      "step": 27354
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5884082317352295,
      "learning_rate": 1.0085702225503312e-06,
      "loss": 0.4264,
      "step": 27355
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.537432074546814,
      "learning_rate": 1.0081245567319187e-06,
      "loss": 0.3821,
      "step": 27356
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5419987440109253,
      "learning_rate": 1.0076789841736334e-06,
      "loss": 0.4353,
      "step": 27357
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5882890224456787,
      "learning_rate": 1.0072335048800974e-06,
      "loss": 0.3951,
      "step": 27358
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5109751224517822,
      "learning_rate": 1.0067881188559314e-06,
      "loss": 0.4034,
      "step": 27359
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4628522396087646,
      "learning_rate": 1.0063428261057507e-06,
      "loss": 0.3568,
      "step": 27360
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5213022232055664,
      "learning_rate": 1.0058976266341768e-06,
      "loss": 0.4028,
      "step": 27361
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5293816328048706,
      "learning_rate": 1.0054525204458266e-06,
      "loss": 0.3727,
      "step": 27362
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5854953527450562,
      "learning_rate": 1.005007507545318e-06,
      "loss": 0.4253,
      "step": 27363
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.543763518333435,
      "learning_rate": 1.0045625879372667e-06,
      "loss": 0.3591,
      "step": 27364
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5281798839569092,
      "learning_rate": 1.0041177616262832e-06,
      "loss": 0.4231,
      "step": 27365
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4934862852096558,
      "learning_rate": 1.0036730286169839e-06,
      "loss": 0.3727,
      "step": 27366
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.573879361152649,
      "learning_rate": 1.0032283889139827e-06,
      "loss": 0.4299,
      "step": 27367
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5400114059448242,
      "learning_rate": 1.0027838425218894e-06,
      "loss": 0.4065,
      "step": 27368
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5616035461425781,
      "learning_rate": 1.0023393894453138e-06,
      "loss": 0.4309,
      "step": 27369
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4985637664794922,
      "learning_rate": 1.0018950296888662e-06,
      "loss": 0.3862,
      "step": 27370
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5310500860214233,
      "learning_rate": 1.001450763257158e-06,
      "loss": 0.3595,
      "step": 27371
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5411404371261597,
      "learning_rate": 1.0010065901547927e-06,
      "loss": 0.3827,
      "step": 27372
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.559762716293335,
      "learning_rate": 1.0005625103863814e-06,
      "loss": 0.4526,
      "step": 27373
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5044372081756592,
      "learning_rate": 1.0001185239565248e-06,
      "loss": 0.3668,
      "step": 27374
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6095601320266724,
      "learning_rate": 9.996746308698336e-07,
      "loss": 0.9522,
      "step": 27375
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.381258249282837,
      "learning_rate": 9.992308311309062e-07,
      "loss": 0.975,
      "step": 27376
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5875248908996582,
      "learning_rate": 9.987871247443492e-07,
      "loss": 0.4199,
      "step": 27377
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4742964506149292,
      "learning_rate": 9.983435117147643e-07,
      "loss": 0.3592,
      "step": 27378
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.549295425415039,
      "learning_rate": 9.978999920467513e-07,
      "loss": 0.4647,
      "step": 27379
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6425727605819702,
      "learning_rate": 9.974565657449086e-07,
      "loss": 0.3895,
      "step": 27380
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5217005014419556,
      "learning_rate": 9.970132328138382e-07,
      "loss": 0.4165,
      "step": 27381
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4704426527023315,
      "learning_rate": 9.965699932581363e-07,
      "loss": 0.3725,
      "step": 27382
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5707579851150513,
      "learning_rate": 9.961268470824025e-07,
      "loss": 0.4393,
      "step": 27383
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5698961019515991,
      "learning_rate": 9.956837942912312e-07,
      "loss": 0.4276,
      "step": 27384
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6810048818588257,
      "learning_rate": 9.952408348892162e-07,
      "loss": 0.415,
      "step": 27385
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5970159769058228,
      "learning_rate": 9.947979688809528e-07,
      "loss": 0.378,
      "step": 27386
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4921157360076904,
      "learning_rate": 9.943551962710362e-07,
      "loss": 0.3829,
      "step": 27387
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4840915203094482,
      "learning_rate": 9.939125170640561e-07,
      "loss": 0.3939,
      "step": 27388
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.480115532875061,
      "learning_rate": 9.934699312646057e-07,
      "loss": 0.4107,
      "step": 27389
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5402525663375854,
      "learning_rate": 9.93027438877272e-07,
      "loss": 0.3848,
      "step": 27390
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.552830696105957,
      "learning_rate": 9.925850399066495e-07,
      "loss": 0.4109,
      "step": 27391
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4317744970321655,
      "learning_rate": 9.921427343573209e-07,
      "loss": 0.3756,
      "step": 27392
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5185989141464233,
      "learning_rate": 9.917005222338772e-07,
      "loss": 0.4267,
      "step": 27393
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5354441404342651,
      "learning_rate": 9.912584035409056e-07,
      "loss": 0.3997,
      "step": 27394
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5747015476226807,
      "learning_rate": 9.908163782829905e-07,
      "loss": 0.4241,
      "step": 27395
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.519834041595459,
      "learning_rate": 9.903744464647135e-07,
      "loss": 0.3757,
      "step": 27396
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.761649489402771,
      "learning_rate": 9.899326080906602e-07,
      "loss": 0.398,
      "step": 27397
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5435185432434082,
      "learning_rate": 9.894908631654143e-07,
      "loss": 0.4013,
      "step": 27398
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5192129611968994,
      "learning_rate": 9.890492116935591e-07,
      "loss": 0.4218,
      "step": 27399
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4933602809906006,
      "learning_rate": 9.886076536796718e-07,
      "loss": 0.3895,
      "step": 27400
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.521764874458313,
      "learning_rate": 9.881661891283312e-07,
      "loss": 0.3986,
      "step": 27401
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5027331113815308,
      "learning_rate": 9.87724818044119e-07,
      "loss": 0.3522,
      "step": 27402
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4603794813156128,
      "learning_rate": 9.872835404316127e-07,
      "loss": 0.3841,
      "step": 27403
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5224318504333496,
      "learning_rate": 9.868423562953866e-07,
      "loss": 0.3457,
      "step": 27404
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4973886013031006,
      "learning_rate": 9.86401265640019e-07,
      "loss": 0.3976,
      "step": 27405
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6382726430892944,
      "learning_rate": 9.85960268470083e-07,
      "loss": 0.4322,
      "step": 27406
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4957561492919922,
      "learning_rate": 9.855193647901552e-07,
      "loss": 0.4396,
      "step": 27407
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4722352027893066,
      "learning_rate": 9.850785546048036e-07,
      "loss": 0.3868,
      "step": 27408
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6150089502334595,
      "learning_rate": 9.846378379186062e-07,
      "loss": 0.4789,
      "step": 27409
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8063647747039795,
      "learning_rate": 9.841972147361278e-07,
      "loss": 0.373,
      "step": 27410
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4709275960922241,
      "learning_rate": 9.837566850619428e-07,
      "loss": 0.4164,
      "step": 27411
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5089935064315796,
      "learning_rate": 9.833162489006176e-07,
      "loss": 0.3534,
      "step": 27412
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.574663519859314,
      "learning_rate": 9.828759062567206e-07,
      "loss": 0.4207,
      "step": 27413
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4820287227630615,
      "learning_rate": 9.824356571348226e-07,
      "loss": 0.3459,
      "step": 27414
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6285476684570312,
      "learning_rate": 9.819955015394856e-07,
      "loss": 0.4492,
      "step": 27415
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6054481267929077,
      "learning_rate": 9.815554394752747e-07,
      "loss": 0.4056,
      "step": 27416
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5536621809005737,
      "learning_rate": 9.811154709467542e-07,
      "loss": 0.4668,
      "step": 27417
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4894633293151855,
      "learning_rate": 9.80675595958489e-07,
      "loss": 0.3659,
      "step": 27418
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5286601781845093,
      "learning_rate": 9.802358145150427e-07,
      "loss": 0.4204,
      "step": 27419
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5542330741882324,
      "learning_rate": 9.797961266209732e-07,
      "loss": 0.3741,
      "step": 27420
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5746046304702759,
      "learning_rate": 9.793565322808407e-07,
      "loss": 0.4462,
      "step": 27421
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5750459432601929,
      "learning_rate": 9.789170314992058e-07,
      "loss": 0.3797,
      "step": 27422
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.655024766921997,
      "learning_rate": 9.784776242806281e-07,
      "loss": 0.4569,
      "step": 27423
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6745377779006958,
      "learning_rate": 9.780383106296621e-07,
      "loss": 0.3736,
      "step": 27424
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5652375221252441,
      "learning_rate": 9.77599090550867e-07,
      "loss": 0.4335,
      "step": 27425
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4834884405136108,
      "learning_rate": 9.77159964048795e-07,
      "loss": 0.3797,
      "step": 27426
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.511875033378601,
      "learning_rate": 9.767209311280046e-07,
      "loss": 0.4019,
      "step": 27427
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5139120817184448,
      "learning_rate": 9.762819917930455e-07,
      "loss": 0.3544,
      "step": 27428
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5872374773025513,
      "learning_rate": 9.75843146048472e-07,
      "loss": 0.4043,
      "step": 27429
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5314502716064453,
      "learning_rate": 9.754043938988366e-07,
      "loss": 0.3752,
      "step": 27430
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.585533618927002,
      "learning_rate": 9.749657353486897e-07,
      "loss": 0.4394,
      "step": 27431
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.603855013847351,
      "learning_rate": 9.745271704025772e-07,
      "loss": 0.3716,
      "step": 27432
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5717096328735352,
      "learning_rate": 9.7408869906505e-07,
      "loss": 0.422,
      "step": 27433
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.512613296508789,
      "learning_rate": 9.736503213406567e-07,
      "loss": 0.3771,
      "step": 27434
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4951151609420776,
      "learning_rate": 9.73212037233946e-07,
      "loss": 0.3953,
      "step": 27435
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4265657663345337,
      "learning_rate": 9.727738467494596e-07,
      "loss": 0.345,
      "step": 27436
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5396027565002441,
      "learning_rate": 9.72335749891743e-07,
      "loss": 0.3886,
      "step": 27437
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5094445943832397,
      "learning_rate": 9.718977466653402e-07,
      "loss": 0.385,
      "step": 27438
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5411640405654907,
      "learning_rate": 9.714598370747962e-07,
      "loss": 0.4303,
      "step": 27439
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5219712257385254,
      "learning_rate": 9.71022021124649e-07,
      "loss": 0.3598,
      "step": 27440
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5350961685180664,
      "learning_rate": 9.705842988194437e-07,
      "loss": 0.4061,
      "step": 27441
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5085138082504272,
      "learning_rate": 9.701466701637164e-07,
      "loss": 0.3674,
      "step": 27442
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4816118478775024,
      "learning_rate": 9.697091351620092e-07,
      "loss": 0.4045,
      "step": 27443
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4761086702346802,
      "learning_rate": 9.692716938188563e-07,
      "loss": 0.3623,
      "step": 27444
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.640861988067627,
      "learning_rate": 9.688343461387972e-07,
      "loss": 0.4079,
      "step": 27445
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4945290088653564,
      "learning_rate": 9.683970921263686e-07,
      "loss": 0.3555,
      "step": 27446
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.50698721408844,
      "learning_rate": 9.679599317861056e-07,
      "loss": 0.4043,
      "step": 27447
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5187945365905762,
      "learning_rate": 9.675228651225389e-07,
      "loss": 0.3781,
      "step": 27448
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8816853761672974,
      "learning_rate": 9.670858921402027e-07,
      "loss": 0.4353,
      "step": 27449
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6261922121047974,
      "learning_rate": 9.666490128436335e-07,
      "loss": 0.3693,
      "step": 27450
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6536527872085571,
      "learning_rate": 9.662122272373574e-07,
      "loss": 0.4987,
      "step": 27451
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5523909330368042,
      "learning_rate": 9.657755353259057e-07,
      "loss": 0.375,
      "step": 27452
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.627353310585022,
      "learning_rate": 9.653389371138078e-07,
      "loss": 0.4531,
      "step": 27453
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5835646390914917,
      "learning_rate": 9.649024326055923e-07,
      "loss": 0.3812,
      "step": 27454
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6004512310028076,
      "learning_rate": 9.64466021805789e-07,
      "loss": 0.4557,
      "step": 27455
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7634305953979492,
      "learning_rate": 9.640297047189207e-07,
      "loss": 0.3806,
      "step": 27456
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8272182941436768,
      "learning_rate": 9.635934813495118e-07,
      "loss": 0.447,
      "step": 27457
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5417563915252686,
      "learning_rate": 9.631573517020886e-07,
      "loss": 0.3801,
      "step": 27458
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6364550590515137,
      "learning_rate": 9.627213157811755e-07,
      "loss": 0.4616,
      "step": 27459
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4696990251541138,
      "learning_rate": 9.62285373591293e-07,
      "loss": 0.3795,
      "step": 27460
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.566123604774475,
      "learning_rate": 9.618495251369642e-07,
      "loss": 0.4044,
      "step": 27461
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4862216711044312,
      "learning_rate": 9.614137704227078e-07,
      "loss": 0.3641,
      "step": 27462
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5818637609481812,
      "learning_rate": 9.609781094530446e-07,
      "loss": 0.435,
      "step": 27463
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6364859342575073,
      "learning_rate": 9.60542542232491e-07,
      "loss": 0.3771,
      "step": 27464
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6022834777832031,
      "learning_rate": 9.601070687655667e-07,
      "loss": 0.4417,
      "step": 27465
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.442240834236145,
      "learning_rate": 9.596716890567891e-07,
      "loss": 0.3542,
      "step": 27466
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6784566640853882,
      "learning_rate": 9.592364031106704e-07,
      "loss": 0.4489,
      "step": 27467
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4682995080947876,
      "learning_rate": 9.58801210931729e-07,
      "loss": 0.3704,
      "step": 27468
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5927246809005737,
      "learning_rate": 9.583661125244747e-07,
      "loss": 0.4379,
      "step": 27469
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.466790795326233,
      "learning_rate": 9.579311078934216e-07,
      "loss": 0.3631,
      "step": 27470
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5399811267852783,
      "learning_rate": 9.574961970430852e-07,
      "loss": 0.4079,
      "step": 27471
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6538645029067993,
      "learning_rate": 9.570613799779716e-07,
      "loss": 0.4056,
      "step": 27472
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5734995603561401,
      "learning_rate": 9.566266567025905e-07,
      "loss": 0.4195,
      "step": 27473
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5166785717010498,
      "learning_rate": 9.56192027221452e-07,
      "loss": 0.3853,
      "step": 27474
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.552555799484253,
      "learning_rate": 9.557574915390643e-07,
      "loss": 0.431,
      "step": 27475
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.565982699394226,
      "learning_rate": 9.553230496599353e-07,
      "loss": 0.3648,
      "step": 27476
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6676568984985352,
      "learning_rate": 9.548887015885699e-07,
      "loss": 0.8658,
      "step": 27477
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4849690198898315,
      "learning_rate": 9.544544473294704e-07,
      "loss": 1.0682,
      "step": 27478
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6325082778930664,
      "learning_rate": 9.54020286887143e-07,
      "loss": 0.3817,
      "step": 27479
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.527617335319519,
      "learning_rate": 9.535862202660918e-07,
      "loss": 0.3775,
      "step": 27480
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5674678087234497,
      "learning_rate": 9.531522474708166e-07,
      "loss": 0.4093,
      "step": 27481
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4538720846176147,
      "learning_rate": 9.527183685058194e-07,
      "loss": 0.372,
      "step": 27482
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6432406902313232,
      "learning_rate": 9.522845833756001e-07,
      "loss": 0.4568,
      "step": 27483
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5265735387802124,
      "learning_rate": 9.518508920846581e-07,
      "loss": 0.3664,
      "step": 27484
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6660399436950684,
      "learning_rate": 9.5141729463749e-07,
      "loss": 0.4449,
      "step": 27485
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.520212173461914,
      "learning_rate": 9.509837910385955e-07,
      "loss": 0.3402,
      "step": 27486
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4904545545578003,
      "learning_rate": 9.505503812924677e-07,
      "loss": 0.4451,
      "step": 27487
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.4696431159973145,
      "learning_rate": 9.501170654036051e-07,
      "loss": 0.3626,
      "step": 27488
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5975028276443481,
      "learning_rate": 9.496838433764977e-07,
      "loss": 0.4191,
      "step": 27489
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5280877351760864,
      "learning_rate": 9.492507152156416e-07,
      "loss": 0.3896,
      "step": 27490
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5429829359054565,
      "learning_rate": 9.4881768092553e-07,
      "loss": 0.4029,
      "step": 27491
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4638457298278809,
      "learning_rate": 9.483847405106528e-07,
      "loss": 0.3788,
      "step": 27492
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5979348421096802,
      "learning_rate": 9.479518939754984e-07,
      "loss": 0.4479,
      "step": 27493
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5043127536773682,
      "learning_rate": 9.475191413245566e-07,
      "loss": 0.3859,
      "step": 27494
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5546252727508545,
      "learning_rate": 9.470864825623183e-07,
      "loss": 0.398,
      "step": 27495
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4929988384246826,
      "learning_rate": 9.466539176932709e-07,
      "loss": 0.3756,
      "step": 27496
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7755696773529053,
      "learning_rate": 9.462214467218999e-07,
      "loss": 0.913,
      "step": 27497
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.396183967590332,
      "learning_rate": 9.457890696526872e-07,
      "loss": 0.973,
      "step": 27498
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6965724229812622,
      "learning_rate": 9.453567864901215e-07,
      "loss": 0.4207,
      "step": 27499
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4806981086730957,
      "learning_rate": 9.449245972386867e-07,
      "loss": 0.3661,
      "step": 27500
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5655875205993652,
      "learning_rate": 9.444925019028606e-07,
      "loss": 0.4207,
      "step": 27501
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.472102403640747,
      "learning_rate": 9.440605004871306e-07,
      "loss": 0.3663,
      "step": 27502
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4874968528747559,
      "learning_rate": 9.436285929959732e-07,
      "loss": 0.3879,
      "step": 27503
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4832895994186401,
      "learning_rate": 9.431967794338692e-07,
      "loss": 0.3382,
      "step": 27504
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5111300945281982,
      "learning_rate": 9.427650598052973e-07,
      "loss": 0.4338,
      "step": 27505
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5782301425933838,
      "learning_rate": 9.423334341147339e-07,
      "loss": 0.381,
      "step": 27506
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5023425817489624,
      "learning_rate": 9.419019023666587e-07,
      "loss": 0.4212,
      "step": 27507
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8243240118026733,
      "learning_rate": 9.414704645655449e-07,
      "loss": 0.3669,
      "step": 27508
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5839197635650635,
      "learning_rate": 9.410391207158664e-07,
      "loss": 0.432,
      "step": 27509
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.567359447479248,
      "learning_rate": 9.406078708220978e-07,
      "loss": 0.3873,
      "step": 27510
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5415414571762085,
      "learning_rate": 9.401767148887131e-07,
      "loss": 0.4263,
      "step": 27511
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5173476934432983,
      "learning_rate": 9.397456529201843e-07,
      "loss": 0.4342,
      "step": 27512
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5186100006103516,
      "learning_rate": 9.393146849209822e-07,
      "loss": 0.4066,
      "step": 27513
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4443626403808594,
      "learning_rate": 9.388838108955722e-07,
      "loss": 0.3569,
      "step": 27514
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.464795470237732,
      "learning_rate": 9.384530308484275e-07,
      "loss": 0.3667,
      "step": 27515
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5030802488327026,
      "learning_rate": 9.380223447840175e-07,
      "loss": 0.391,
      "step": 27516
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5322797298431396,
      "learning_rate": 9.375917527068034e-07,
      "loss": 0.3979,
      "step": 27517
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4551535844802856,
      "learning_rate": 9.371612546212572e-07,
      "loss": 0.3572,
      "step": 27518
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.682921290397644,
      "learning_rate": 9.367308505318384e-07,
      "loss": 0.4531,
      "step": 27519
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4972164630889893,
      "learning_rate": 9.363005404430159e-07,
      "loss": 0.3563,
      "step": 27520
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5052005052566528,
      "learning_rate": 9.358703243592493e-07,
      "loss": 0.4331,
      "step": 27521
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4656023979187012,
      "learning_rate": 9.354402022850007e-07,
      "loss": 0.3603,
      "step": 27522
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6103317737579346,
      "learning_rate": 9.350101742247341e-07,
      "loss": 0.4609,
      "step": 27523
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4940358400344849,
      "learning_rate": 9.345802401829074e-07,
      "loss": 0.3788,
      "step": 27524
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5900768041610718,
      "learning_rate": 9.34150400163979e-07,
      "loss": 0.4329,
      "step": 27525
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4736106395721436,
      "learning_rate": 9.337206541724065e-07,
      "loss": 0.3765,
      "step": 27526
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5893311500549316,
      "learning_rate": 9.332910022126517e-07,
      "loss": 0.441,
      "step": 27527
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4622594118118286,
      "learning_rate": 9.32861444289167e-07,
      "loss": 0.3561,
      "step": 27528
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8102822303771973,
      "learning_rate": 9.324319804064063e-07,
      "loss": 0.4591,
      "step": 27529
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5730528831481934,
      "learning_rate": 9.320026105688262e-07,
      "loss": 0.3747,
      "step": 27530
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.598883032798767,
      "learning_rate": 9.315733347808797e-07,
      "loss": 0.4018,
      "step": 27531
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4893251657485962,
      "learning_rate": 9.3114415304702e-07,
      "loss": 0.3705,
      "step": 27532
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.504179835319519,
      "learning_rate": 9.307150653716979e-07,
      "loss": 0.4043,
      "step": 27533
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5552549362182617,
      "learning_rate": 9.302860717593609e-07,
      "loss": 0.4046,
      "step": 27534
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5465744733810425,
      "learning_rate": 9.2985717221446e-07,
      "loss": 0.4184,
      "step": 27535
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5153465270996094,
      "learning_rate": 9.294283667414473e-07,
      "loss": 0.3709,
      "step": 27536
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6429630517959595,
      "learning_rate": 9.289996553447645e-07,
      "loss": 0.4335,
      "step": 27537
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5249035358428955,
      "learning_rate": 9.285710380288616e-07,
      "loss": 0.3669,
      "step": 27538
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.489086389541626,
      "learning_rate": 9.281425147981827e-07,
      "loss": 0.3792,
      "step": 27539
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6341854333877563,
      "learning_rate": 9.277140856571731e-07,
      "loss": 0.3765,
      "step": 27540
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.624979853630066,
      "learning_rate": 9.27285750610274e-07,
      "loss": 0.429,
      "step": 27541
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4710463285446167,
      "learning_rate": 9.268575096619303e-07,
      "loss": 0.3826,
      "step": 27542
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5306077003479004,
      "learning_rate": 9.264293628165844e-07,
      "loss": 0.4424,
      "step": 27543
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5583133697509766,
      "learning_rate": 9.260013100786758e-07,
      "loss": 0.3882,
      "step": 27544
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.547426462173462,
      "learning_rate": 9.25573351452641e-07,
      "loss": 0.4087,
      "step": 27545
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5195949077606201,
      "learning_rate": 9.251454869429221e-07,
      "loss": 0.3714,
      "step": 27546
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.527134895324707,
      "learning_rate": 9.247177165539556e-07,
      "loss": 0.4099,
      "step": 27547
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5312482118606567,
      "learning_rate": 9.2429004029018e-07,
      "loss": 0.3559,
      "step": 27548
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5095232725143433,
      "learning_rate": 9.238624581560296e-07,
      "loss": 0.3997,
      "step": 27549
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4812885522842407,
      "learning_rate": 9.234349701559375e-07,
      "loss": 0.3677,
      "step": 27550
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6288424730300903,
      "learning_rate": 9.230075762943391e-07,
      "loss": 0.4384,
      "step": 27551
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5571917295455933,
      "learning_rate": 9.225802765756687e-07,
      "loss": 0.377,
      "step": 27552
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5632805824279785,
      "learning_rate": 9.221530710043536e-07,
      "loss": 0.4178,
      "step": 27553
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4906774759292603,
      "learning_rate": 9.217259595848304e-07,
      "loss": 0.3608,
      "step": 27554
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.628864049911499,
      "learning_rate": 9.212989423215235e-07,
      "loss": 0.404,
      "step": 27555
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6112728118896484,
      "learning_rate": 9.208720192188658e-07,
      "loss": 0.3843,
      "step": 27556
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5396451950073242,
      "learning_rate": 9.204451902812827e-07,
      "loss": 0.4212,
      "step": 27557
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5159227848052979,
      "learning_rate": 9.200184555132008e-07,
      "loss": 0.3898,
      "step": 27558
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6250324249267578,
      "learning_rate": 9.195918149190497e-07,
      "loss": 0.4296,
      "step": 27559
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5190403461456299,
      "learning_rate": 9.191652685032515e-07,
      "loss": 0.3547,
      "step": 27560
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5157649517059326,
      "learning_rate": 9.187388162702283e-07,
      "loss": 0.401,
      "step": 27561
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4677975177764893,
      "learning_rate": 9.183124582244052e-07,
      "loss": 0.3861,
      "step": 27562
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6088436841964722,
      "learning_rate": 9.178861943702045e-07,
      "loss": 0.4374,
      "step": 27563
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5049794912338257,
      "learning_rate": 9.174600247120513e-07,
      "loss": 0.3882,
      "step": 27564
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4669795036315918,
      "learning_rate": 9.170339492543568e-07,
      "loss": 0.3811,
      "step": 27565
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6359940767288208,
      "learning_rate": 9.166079680015449e-07,
      "loss": 0.3875,
      "step": 27566
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7500520944595337,
      "learning_rate": 9.161820809580336e-07,
      "loss": 0.4292,
      "step": 27567
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5118764638900757,
      "learning_rate": 9.157562881282411e-07,
      "loss": 0.3753,
      "step": 27568
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.609591245651245,
      "learning_rate": 9.153305895165831e-07,
      "loss": 0.4953,
      "step": 27569
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5349485874176025,
      "learning_rate": 9.149049851274727e-07,
      "loss": 0.384,
      "step": 27570
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7440999746322632,
      "learning_rate": 9.14479474965324e-07,
      "loss": 0.4502,
      "step": 27571
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5274615287780762,
      "learning_rate": 9.140540590345548e-07,
      "loss": 0.3592,
      "step": 27572
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5192350149154663,
      "learning_rate": 9.136287373395714e-07,
      "loss": 0.3901,
      "step": 27573
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4537113904953003,
      "learning_rate": 9.132035098847902e-07,
      "loss": 0.3683,
      "step": 27574
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5116419792175293,
      "learning_rate": 9.127783766746167e-07,
      "loss": 0.4079,
      "step": 27575
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5025440454483032,
      "learning_rate": 9.123533377134652e-07,
      "loss": 0.359,
      "step": 27576
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5111216306686401,
      "learning_rate": 9.119283930057387e-07,
      "loss": 0.4153,
      "step": 27577
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.730027437210083,
      "learning_rate": 9.115035425558482e-07,
      "loss": 0.3768,
      "step": 27578
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6173933744430542,
      "learning_rate": 9.110787863682002e-07,
      "loss": 0.4415,
      "step": 27579
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5028588771820068,
      "learning_rate": 9.106541244472e-07,
      "loss": 0.3749,
      "step": 27580
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5391149520874023,
      "learning_rate": 9.102295567972485e-07,
      "loss": 0.4208,
      "step": 27581
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5485022068023682,
      "learning_rate": 9.098050834227523e-07,
      "loss": 0.3807,
      "step": 27582
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6587599515914917,
      "learning_rate": 9.093807043281133e-07,
      "loss": 0.4229,
      "step": 27583
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4824882745742798,
      "learning_rate": 9.089564195177347e-07,
      "loss": 0.3687,
      "step": 27584
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6251415014266968,
      "learning_rate": 9.085322289960152e-07,
      "loss": 0.4075,
      "step": 27585
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5523134469985962,
      "learning_rate": 9.081081327673535e-07,
      "loss": 0.3714,
      "step": 27586
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4830191135406494,
      "learning_rate": 9.076841308361484e-07,
      "loss": 0.3922,
      "step": 27587
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5365943908691406,
      "learning_rate": 9.072602232067995e-07,
      "loss": 0.3943,
      "step": 27588
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5567195415496826,
      "learning_rate": 9.068364098837013e-07,
      "loss": 0.4251,
      "step": 27589
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4597134590148926,
      "learning_rate": 9.064126908712523e-07,
      "loss": 0.3615,
      "step": 27590
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5833531618118286,
      "learning_rate": 9.059890661738424e-07,
      "loss": 0.3961,
      "step": 27591
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5605486631393433,
      "learning_rate": 9.055655357958703e-07,
      "loss": 0.3858,
      "step": 27592
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.517878770828247,
      "learning_rate": 9.051420997417237e-07,
      "loss": 0.4043,
      "step": 27593
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5579249858856201,
      "learning_rate": 9.047187580157979e-07,
      "loss": 0.3588,
      "step": 27594
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4900054931640625,
      "learning_rate": 9.042955106224838e-07,
      "loss": 0.3717,
      "step": 27595
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5047667026519775,
      "learning_rate": 9.038723575661701e-07,
      "loss": 0.3848,
      "step": 27596
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5725033283233643,
      "learning_rate": 9.034492988512444e-07,
      "loss": 0.4422,
      "step": 27597
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5268657207489014,
      "learning_rate": 9.030263344820944e-07,
      "loss": 0.3895,
      "step": 27598
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5128320455551147,
      "learning_rate": 9.026034644631087e-07,
      "loss": 0.4267,
      "step": 27599
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5767844915390015,
      "learning_rate": 9.021806887986761e-07,
      "loss": 0.4039,
      "step": 27600
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5499790906906128,
      "learning_rate": 9.017580074931753e-07,
      "loss": 0.4285,
      "step": 27601
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.546312928199768,
      "learning_rate": 9.013354205509917e-07,
      "loss": 0.3535,
      "step": 27602
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5768169164657593,
      "learning_rate": 9.009129279765095e-07,
      "loss": 0.4008,
      "step": 27603
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.521959662437439,
      "learning_rate": 9.004905297741118e-07,
      "loss": 0.3754,
      "step": 27604
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5424506664276123,
      "learning_rate": 9.000682259481785e-07,
      "loss": 0.4363,
      "step": 27605
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5247457027435303,
      "learning_rate": 8.996460165030884e-07,
      "loss": 0.3831,
      "step": 27606
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5356022119522095,
      "learning_rate": 8.992239014432203e-07,
      "loss": 0.3803,
      "step": 27607
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.600612998008728,
      "learning_rate": 8.988018807729549e-07,
      "loss": 0.3833,
      "step": 27608
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6345018148422241,
      "learning_rate": 8.983799544966665e-07,
      "loss": 0.3926,
      "step": 27609
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5511730909347534,
      "learning_rate": 8.979581226187328e-07,
      "loss": 0.3648,
      "step": 27610
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9264851808547974,
      "learning_rate": 8.97536385143527e-07,
      "loss": 0.3831,
      "step": 27611
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4751718044281006,
      "learning_rate": 8.971147420754267e-07,
      "loss": 0.3484,
      "step": 27612
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6259199380874634,
      "learning_rate": 8.966931934188005e-07,
      "loss": 0.4156,
      "step": 27613
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4900381565093994,
      "learning_rate": 8.962717391780229e-07,
      "loss": 0.3945,
      "step": 27614
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5720551013946533,
      "learning_rate": 8.958503793574669e-07,
      "loss": 0.4515,
      "step": 27615
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4654219150543213,
      "learning_rate": 8.954291139615013e-07,
      "loss": 0.3588,
      "step": 27616
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6474816799163818,
      "learning_rate": 8.950079429944924e-07,
      "loss": 0.3955,
      "step": 27617
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.489353060722351,
      "learning_rate": 8.945868664608104e-07,
      "loss": 0.3749,
      "step": 27618
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5823980569839478,
      "learning_rate": 8.941658843648237e-07,
      "loss": 0.4139,
      "step": 27619
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4626789093017578,
      "learning_rate": 8.937449967108991e-07,
      "loss": 0.3797,
      "step": 27620
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.568307876586914,
      "learning_rate": 8.933242035033995e-07,
      "loss": 0.4556,
      "step": 27621
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.482534646987915,
      "learning_rate": 8.929035047466894e-07,
      "loss": 0.3563,
      "step": 27622
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.634239673614502,
      "learning_rate": 8.92482900445133e-07,
      "loss": 0.4743,
      "step": 27623
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4784839153289795,
      "learning_rate": 8.920623906030945e-07,
      "loss": 0.3488,
      "step": 27624
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.94595468044281,
      "learning_rate": 8.916419752249317e-07,
      "loss": 0.4364,
      "step": 27625
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.496850609779358,
      "learning_rate": 8.912216543150077e-07,
      "loss": 0.3579,
      "step": 27626
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7979786396026611,
      "learning_rate": 8.9080142787768e-07,
      "loss": 1.0563,
      "step": 27627
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4681705236434937,
      "learning_rate": 8.903812959173086e-07,
      "loss": 1.056,
      "step": 27628
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5513790845870972,
      "learning_rate": 8.899612584382488e-07,
      "loss": 0.4244,
      "step": 27629
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.471100926399231,
      "learning_rate": 8.895413154448585e-07,
      "loss": 0.3732,
      "step": 27630
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.821574330329895,
      "learning_rate": 8.891214669414949e-07,
      "loss": 0.453,
      "step": 27631
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5705814361572266,
      "learning_rate": 8.887017129325104e-07,
      "loss": 0.359,
      "step": 27632
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5935084819793701,
      "learning_rate": 8.88282053422258e-07,
      "loss": 0.4563,
      "step": 27633
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4722667932510376,
      "learning_rate": 8.8786248841509e-07,
      "loss": 0.3367,
      "step": 27634
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6445187330245972,
      "learning_rate": 8.874430179153592e-07,
      "loss": 0.4819,
      "step": 27635
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.3947088718414307,
      "learning_rate": 8.870236419274181e-07,
      "loss": 0.3951,
      "step": 27636
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5395536422729492,
      "learning_rate": 8.86604360455614e-07,
      "loss": 0.3929,
      "step": 27637
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5156277418136597,
      "learning_rate": 8.861851735042948e-07,
      "loss": 0.3638,
      "step": 27638
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6395974159240723,
      "learning_rate": 8.85766081077808e-07,
      "loss": 0.4208,
      "step": 27639
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5103682279586792,
      "learning_rate": 8.853470831805033e-07,
      "loss": 0.3579,
      "step": 27640
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5578563213348389,
      "learning_rate": 8.849281798167253e-07,
      "loss": 0.4501,
      "step": 27641
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5489636659622192,
      "learning_rate": 8.845093709908148e-07,
      "loss": 0.3797,
      "step": 27642
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6797021627426147,
      "learning_rate": 8.840906567071194e-07,
      "loss": 0.4925,
      "step": 27643
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4911288022994995,
      "learning_rate": 8.836720369699825e-07,
      "loss": 0.3721,
      "step": 27644
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.50499427318573,
      "learning_rate": 8.832535117837426e-07,
      "loss": 0.4006,
      "step": 27645
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5449029207229614,
      "learning_rate": 8.828350811527442e-07,
      "loss": 0.3774,
      "step": 27646
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.533855676651001,
      "learning_rate": 8.824167450813226e-07,
      "loss": 0.4086,
      "step": 27647
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.504948616027832,
      "learning_rate": 8.81998503573821e-07,
      "loss": 0.3575,
      "step": 27648
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4594392776489258,
      "learning_rate": 8.815803566345737e-07,
      "loss": 0.404,
      "step": 27649
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.847917914390564,
      "learning_rate": 8.811623042679196e-07,
      "loss": 0.3752,
      "step": 27650
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.525513768196106,
      "learning_rate": 8.807443464781962e-07,
      "loss": 0.3777,
      "step": 27651
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.514305830001831,
      "learning_rate": 8.803264832697356e-07,
      "loss": 0.3625,
      "step": 27652
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5316627025604248,
      "learning_rate": 8.79908714646871e-07,
      "loss": 0.4169,
      "step": 27653
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6003003120422363,
      "learning_rate": 8.794910406139368e-07,
      "loss": 0.4288,
      "step": 27654
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.523988127708435,
      "learning_rate": 8.790734611752649e-07,
      "loss": 0.4191,
      "step": 27655
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5751233100891113,
      "learning_rate": 8.786559763351876e-07,
      "loss": 0.3709,
      "step": 27656
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4787174463272095,
      "learning_rate": 8.782385860980347e-07,
      "loss": 0.3492,
      "step": 27657
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5086137056350708,
      "learning_rate": 8.778212904681316e-07,
      "loss": 0.3598,
      "step": 27658
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4659074544906616,
      "learning_rate": 8.774040894498081e-07,
      "loss": 0.4173,
      "step": 27659
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5833673477172852,
      "learning_rate": 8.769869830473953e-07,
      "loss": 0.3643,
      "step": 27660
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5415414571762085,
      "learning_rate": 8.765699712652131e-07,
      "loss": 0.4659,
      "step": 27661
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5401599407196045,
      "learning_rate": 8.761530541075913e-07,
      "loss": 0.3808,
      "step": 27662
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5458827018737793,
      "learning_rate": 8.75736231578851e-07,
      "loss": 0.4619,
      "step": 27663
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4990971088409424,
      "learning_rate": 8.753195036833173e-07,
      "loss": 0.3722,
      "step": 27664
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4591522216796875,
      "learning_rate": 8.749028704253104e-07,
      "loss": 0.3876,
      "step": 27665
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5236843824386597,
      "learning_rate": 8.744863318091523e-07,
      "loss": 0.3477,
      "step": 27666
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5163832902908325,
      "learning_rate": 8.740698878391641e-07,
      "loss": 0.3929,
      "step": 27667
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5165672302246094,
      "learning_rate": 8.736535385196654e-07,
      "loss": 0.3787,
      "step": 27668
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5590488910675049,
      "learning_rate": 8.732372838549719e-07,
      "loss": 0.4251,
      "step": 27669
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4919251203536987,
      "learning_rate": 8.728211238494011e-07,
      "loss": 0.3612,
      "step": 27670
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6336115598678589,
      "learning_rate": 8.724050585072707e-07,
      "loss": 0.489,
      "step": 27671
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6345189809799194,
      "learning_rate": 8.719890878328973e-07,
      "loss": 0.3822,
      "step": 27672
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4828382730484009,
      "learning_rate": 8.715732118305942e-07,
      "loss": 0.4499,
      "step": 27673
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.539406418800354,
      "learning_rate": 8.711574305046711e-07,
      "loss": 0.3813,
      "step": 27674
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.545582890510559,
      "learning_rate": 8.707417438594445e-07,
      "loss": 0.4317,
      "step": 27675
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7174062728881836,
      "learning_rate": 8.703261518992257e-07,
      "loss": 0.4073,
      "step": 27676
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.596139669418335,
      "learning_rate": 8.699106546283231e-07,
      "loss": 0.4911,
      "step": 27677
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5877004861831665,
      "learning_rate": 8.694952520510458e-07,
      "loss": 0.3875,
      "step": 27678
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.577316403388977,
      "learning_rate": 8.690799441717024e-07,
      "loss": 0.421,
      "step": 27679
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4473137855529785,
      "learning_rate": 8.686647309946039e-07,
      "loss": 0.3335,
      "step": 27680
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7108254432678223,
      "learning_rate": 8.682496125240514e-07,
      "loss": 0.4483,
      "step": 27681
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4652161598205566,
      "learning_rate": 8.678345887643547e-07,
      "loss": 0.367,
      "step": 27682
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6316781044006348,
      "learning_rate": 8.674196597198137e-07,
      "loss": 0.4364,
      "step": 27683
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4763822555541992,
      "learning_rate": 8.670048253947383e-07,
      "loss": 0.352,
      "step": 27684
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6131527423858643,
      "learning_rate": 8.665900857934239e-07,
      "loss": 0.4553,
      "step": 27685
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5715347528457642,
      "learning_rate": 8.661754409201762e-07,
      "loss": 0.3964,
      "step": 27686
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4951486587524414,
      "learning_rate": 8.657608907792958e-07,
      "loss": 0.4005,
      "step": 27687
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.44785737991333,
      "learning_rate": 8.653464353750817e-07,
      "loss": 0.3695,
      "step": 27688
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5351673364639282,
      "learning_rate": 8.649320747118306e-07,
      "loss": 0.3756,
      "step": 27689
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5378766059875488,
      "learning_rate": 8.645178087938422e-07,
      "loss": 0.3674,
      "step": 27690
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5449390411376953,
      "learning_rate": 8.641036376254119e-07,
      "loss": 0.3887,
      "step": 27691
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5276416540145874,
      "learning_rate": 8.636895612108375e-07,
      "loss": 0.37,
      "step": 27692
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5111937522888184,
      "learning_rate": 8.632755795544123e-07,
      "loss": 0.4118,
      "step": 27693
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4542967081069946,
      "learning_rate": 8.628616926604272e-07,
      "loss": 0.3566,
      "step": 27694
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4854238033294678,
      "learning_rate": 8.624479005331788e-07,
      "loss": 0.4045,
      "step": 27695
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.598185658454895,
      "learning_rate": 8.620342031769579e-07,
      "loss": 0.4268,
      "step": 27696
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.618963360786438,
      "learning_rate": 8.616206005960537e-07,
      "loss": 0.4084,
      "step": 27697
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5102537870407104,
      "learning_rate": 8.61207092794758e-07,
      "loss": 0.3557,
      "step": 27698
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6161576509475708,
      "learning_rate": 8.607936797773575e-07,
      "loss": 0.4148,
      "step": 27699
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.486072301864624,
      "learning_rate": 8.603803615481421e-07,
      "loss": 0.381,
      "step": 27700
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5437713861465454,
      "learning_rate": 8.59967138111395e-07,
      "loss": 0.4143,
      "step": 27701
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5371137857437134,
      "learning_rate": 8.59554009471405e-07,
      "loss": 0.3553,
      "step": 27702
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.538954734802246,
      "learning_rate": 8.591409756324575e-07,
      "loss": 0.4613,
      "step": 27703
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4597173929214478,
      "learning_rate": 8.58728036598836e-07,
      "loss": 0.3424,
      "step": 27704
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5071324110031128,
      "learning_rate": 8.583151923748201e-07,
      "loss": 0.4196,
      "step": 27705
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4455102682113647,
      "learning_rate": 8.579024429646932e-07,
      "loss": 0.362,
      "step": 27706
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5211788415908813,
      "learning_rate": 8.574897883727384e-07,
      "loss": 0.4247,
      "step": 27707
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5939764976501465,
      "learning_rate": 8.570772286032347e-07,
      "loss": 0.3821,
      "step": 27708
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5066492557525635,
      "learning_rate": 8.566647636604608e-07,
      "loss": 0.4019,
      "step": 27709
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5611660480499268,
      "learning_rate": 8.562523935486921e-07,
      "loss": 0.4198,
      "step": 27710
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.477800726890564,
      "learning_rate": 8.558401182722087e-07,
      "loss": 0.4147,
      "step": 27711
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5297002792358398,
      "learning_rate": 8.554279378352848e-07,
      "loss": 0.3749,
      "step": 27712
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5285611152648926,
      "learning_rate": 8.55015852242198e-07,
      "loss": 0.4039,
      "step": 27713
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.462138056755066,
      "learning_rate": 8.546038614972207e-07,
      "loss": 0.3601,
      "step": 27714
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5366047620773315,
      "learning_rate": 8.541919656046227e-07,
      "loss": 0.4221,
      "step": 27715
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5655906200408936,
      "learning_rate": 8.537801645686805e-07,
      "loss": 0.3831,
      "step": 27716
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.662569284439087,
      "learning_rate": 8.533684583936642e-07,
      "loss": 0.4548,
      "step": 27717
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4992914199829102,
      "learning_rate": 8.529568470838434e-07,
      "loss": 0.3633,
      "step": 27718
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.330829381942749,
      "learning_rate": 8.52545330643485e-07,
      "loss": 0.4742,
      "step": 27719
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5166330337524414,
      "learning_rate": 8.521339090768599e-07,
      "loss": 0.3644,
      "step": 27720
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.575661063194275,
      "learning_rate": 8.517225823882347e-07,
      "loss": 0.4479,
      "step": 27721
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4847203493118286,
      "learning_rate": 8.513113505818749e-07,
      "loss": 0.3682,
      "step": 27722
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5467751026153564,
      "learning_rate": 8.50900213662047e-07,
      "loss": 0.3948,
      "step": 27723
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4444119930267334,
      "learning_rate": 8.504891716330121e-07,
      "loss": 0.356,
      "step": 27724
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6156082153320312,
      "learning_rate": 8.500782244990369e-07,
      "loss": 0.4473,
      "step": 27725
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4766708612442017,
      "learning_rate": 8.496673722643801e-07,
      "loss": 0.3502,
      "step": 27726
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.784512996673584,
      "learning_rate": 8.492566149333059e-07,
      "loss": 0.3963,
      "step": 27727
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5200716257095337,
      "learning_rate": 8.488459525100745e-07,
      "loss": 0.3883,
      "step": 27728
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5279678106307983,
      "learning_rate": 8.484353849989435e-07,
      "loss": 0.4139,
      "step": 27729
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5801836252212524,
      "learning_rate": 8.480249124041706e-07,
      "loss": 0.4148,
      "step": 27730
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4743053913116455,
      "learning_rate": 8.476145347300136e-07,
      "loss": 0.4265,
      "step": 27731
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.537218689918518,
      "learning_rate": 8.472042519807288e-07,
      "loss": 0.3705,
      "step": 27732
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.532591462135315,
      "learning_rate": 8.467940641605754e-07,
      "loss": 0.4262,
      "step": 27733
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.529758334159851,
      "learning_rate": 8.463839712738031e-07,
      "loss": 0.3714,
      "step": 27734
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.669890284538269,
      "learning_rate": 8.459739733246653e-07,
      "loss": 0.9585,
      "step": 27735
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4468343257904053,
      "learning_rate": 8.45564070317414e-07,
      "loss": 0.9923,
      "step": 27736
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5541447401046753,
      "learning_rate": 8.451542622563047e-07,
      "loss": 0.4223,
      "step": 27737
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0352632999420166,
      "learning_rate": 8.447445491455842e-07,
      "loss": 0.3625,
      "step": 27738
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5525426864624023,
      "learning_rate": 8.443349309895032e-07,
      "loss": 0.4318,
      "step": 27739
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4995428323745728,
      "learning_rate": 8.439254077923076e-07,
      "loss": 0.3768,
      "step": 27740
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5776855945587158,
      "learning_rate": 8.435159795582492e-07,
      "loss": 0.4099,
      "step": 27741
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5285953283309937,
      "learning_rate": 8.431066462915704e-07,
      "loss": 0.3643,
      "step": 27742
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4902576208114624,
      "learning_rate": 8.426974079965189e-07,
      "loss": 0.3855,
      "step": 27743
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4783570766448975,
      "learning_rate": 8.42288264677339e-07,
      "loss": 0.3753,
      "step": 27744
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7929973602294922,
      "learning_rate": 8.418792163382749e-07,
      "loss": 0.4421,
      "step": 27745
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.502450942993164,
      "learning_rate": 8.414702629835647e-07,
      "loss": 0.3699,
      "step": 27746
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5422381162643433,
      "learning_rate": 8.410614046174548e-07,
      "loss": 0.3935,
      "step": 27747
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4834867715835571,
      "learning_rate": 8.406526412441829e-07,
      "loss": 0.3653,
      "step": 27748
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5539616346359253,
      "learning_rate": 8.402439728679912e-07,
      "loss": 0.4129,
      "step": 27749
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6105536222457886,
      "learning_rate": 8.398353994931175e-07,
      "loss": 0.3718,
      "step": 27750
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4925105571746826,
      "learning_rate": 8.394269211237971e-07,
      "loss": 0.4131,
      "step": 27751
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5065367221832275,
      "learning_rate": 8.390185377642668e-07,
      "loss": 0.3785,
      "step": 27752
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5515224933624268,
      "learning_rate": 8.386102494187664e-07,
      "loss": 0.4042,
      "step": 27753
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.483397364616394,
      "learning_rate": 8.382020560915261e-07,
      "loss": 0.3627,
      "step": 27754
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5527840852737427,
      "learning_rate": 8.377939577867799e-07,
      "loss": 0.4738,
      "step": 27755
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5379927158355713,
      "learning_rate": 8.373859545087614e-07,
      "loss": 0.4092,
      "step": 27756
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5794121026992798,
      "learning_rate": 8.36978046261705e-07,
      "loss": 0.4275,
      "step": 27757
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5160986185073853,
      "learning_rate": 8.36570233049836e-07,
      "loss": 0.3578,
      "step": 27758
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5302103757858276,
      "learning_rate": 8.361625148773888e-07,
      "loss": 0.4004,
      "step": 27759
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.505653977394104,
      "learning_rate": 8.357548917485881e-07,
      "loss": 0.3946,
      "step": 27760
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7036947011947632,
      "learning_rate": 8.353473636676657e-07,
      "loss": 0.4411,
      "step": 27761
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5754202604293823,
      "learning_rate": 8.349399306388439e-07,
      "loss": 0.3757,
      "step": 27762
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5832713842391968,
      "learning_rate": 8.345325926663505e-07,
      "loss": 0.4148,
      "step": 27763
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4996614456176758,
      "learning_rate": 8.341253497544122e-07,
      "loss": 0.3427,
      "step": 27764
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5570042133331299,
      "learning_rate": 8.33718201907252e-07,
      "loss": 0.3975,
      "step": 27765
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4800103902816772,
      "learning_rate": 8.33311149129089e-07,
      "loss": 0.3629,
      "step": 27766
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5541422367095947,
      "learning_rate": 8.329041914241476e-07,
      "loss": 0.4192,
      "step": 27767
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6087994575500488,
      "learning_rate": 8.324973287966487e-07,
      "loss": 0.3412,
      "step": 27768
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5563534498214722,
      "learning_rate": 8.320905612508145e-07,
      "loss": 0.4407,
      "step": 27769
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5743350982666016,
      "learning_rate": 8.316838887908607e-07,
      "loss": 0.3655,
      "step": 27770
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.601194977760315,
      "learning_rate": 8.312773114210049e-07,
      "loss": 0.5055,
      "step": 27771
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5307726860046387,
      "learning_rate": 8.308708291454637e-07,
      "loss": 0.3968,
      "step": 27772
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5189404487609863,
      "learning_rate": 8.304644419684571e-07,
      "loss": 0.4125,
      "step": 27773
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6058000326156616,
      "learning_rate": 8.300581498941951e-07,
      "loss": 0.3752,
      "step": 27774
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.755372166633606,
      "learning_rate": 8.296519529268954e-07,
      "loss": 0.5129,
      "step": 27775
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5852251052856445,
      "learning_rate": 8.292458510707669e-07,
      "loss": 0.3928,
      "step": 27776
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7580965757369995,
      "learning_rate": 8.288398443300249e-07,
      "loss": 0.941,
      "step": 27777
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6005127429962158,
      "learning_rate": 8.284339327088786e-07,
      "loss": 1.0121,
      "step": 27778
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5069453716278076,
      "learning_rate": 8.280281162115378e-07,
      "loss": 0.3922,
      "step": 27779
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4865552186965942,
      "learning_rate": 8.276223948422135e-07,
      "loss": 0.3588,
      "step": 27780
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5673872232437134,
      "learning_rate": 8.272167686051136e-07,
      "loss": 0.4189,
      "step": 27781
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5908154249191284,
      "learning_rate": 8.268112375044412e-07,
      "loss": 0.3711,
      "step": 27782
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.589945912361145,
      "learning_rate": 8.264058015444055e-07,
      "loss": 0.407,
      "step": 27783
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6016817092895508,
      "learning_rate": 8.260004607292104e-07,
      "loss": 0.3932,
      "step": 27784
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5370523929595947,
      "learning_rate": 8.255952150630619e-07,
      "loss": 0.422,
      "step": 27785
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4956543445587158,
      "learning_rate": 8.25190064550162e-07,
      "loss": 0.3818,
      "step": 27786
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5856510400772095,
      "learning_rate": 8.247850091947107e-07,
      "loss": 0.403,
      "step": 27787
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5196045637130737,
      "learning_rate": 8.243800490009112e-07,
      "loss": 0.3679,
      "step": 27788
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4400935173034668,
      "learning_rate": 8.239751839729649e-07,
      "loss": 0.4029,
      "step": 27789
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.502402424812317,
      "learning_rate": 8.235704141150669e-07,
      "loss": 0.3768,
      "step": 27790
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.601742148399353,
      "learning_rate": 8.231657394314196e-07,
      "loss": 0.4072,
      "step": 27791
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.568543791770935,
      "learning_rate": 8.227611599262165e-07,
      "loss": 0.3825,
      "step": 27792
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5257965326309204,
      "learning_rate": 8.22356675603656e-07,
      "loss": 0.4184,
      "step": 27793
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5317857265472412,
      "learning_rate": 8.219522864679319e-07,
      "loss": 0.3723,
      "step": 27794
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7977508306503296,
      "learning_rate": 8.215479925232405e-07,
      "loss": 0.4235,
      "step": 27795
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5215585231781006,
      "learning_rate": 8.211437937737721e-07,
      "loss": 0.3768,
      "step": 27796
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5061992406845093,
      "learning_rate": 8.20739690223722e-07,
      "loss": 0.3906,
      "step": 27797
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5240510702133179,
      "learning_rate": 8.203356818772767e-07,
      "loss": 0.3645,
      "step": 27798
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5919138193130493,
      "learning_rate": 8.19931768738631e-07,
      "loss": 0.4239,
      "step": 27799
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4848690032958984,
      "learning_rate": 8.195279508119735e-07,
      "loss": 0.3665,
      "step": 27800
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5185738801956177,
      "learning_rate": 8.191242281014911e-07,
      "loss": 0.3973,
      "step": 27801
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7717151641845703,
      "learning_rate": 8.187206006113701e-07,
      "loss": 0.3613,
      "step": 27802
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5578954219818115,
      "learning_rate": 8.183170683457986e-07,
      "loss": 0.4253,
      "step": 27803
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5640956163406372,
      "learning_rate": 8.179136313089608e-07,
      "loss": 0.4164,
      "step": 27804
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4974913597106934,
      "learning_rate": 8.175102895050436e-07,
      "loss": 0.4008,
      "step": 27805
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6417745351791382,
      "learning_rate": 8.171070429382277e-07,
      "loss": 0.364,
      "step": 27806
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5410100221633911,
      "learning_rate": 8.167038916126946e-07,
      "loss": 0.3948,
      "step": 27807
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6863175630569458,
      "learning_rate": 8.163008355326274e-07,
      "loss": 0.4249,
      "step": 27808
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5664414167404175,
      "learning_rate": 8.158978747022073e-07,
      "loss": 0.4363,
      "step": 27809
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5564842224121094,
      "learning_rate": 8.154950091256098e-07,
      "loss": 0.3994,
      "step": 27810
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5317617654800415,
      "learning_rate": 8.150922388070182e-07,
      "loss": 0.4428,
      "step": 27811
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8320190906524658,
      "learning_rate": 8.146895637506069e-07,
      "loss": 0.3914,
      "step": 27812
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5930513143539429,
      "learning_rate": 8.142869839605527e-07,
      "loss": 0.4423,
      "step": 27813
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4843196868896484,
      "learning_rate": 8.13884499441031e-07,
      "loss": 0.3688,
      "step": 27814
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7156118154525757,
      "learning_rate": 8.134821101962153e-07,
      "loss": 0.9471,
      "step": 27815
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3606806993484497,
      "learning_rate": 8.130798162302822e-07,
      "loss": 0.9585,
      "step": 27816
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5122778415679932,
      "learning_rate": 8.126776175474028e-07,
      "loss": 0.3704,
      "step": 27817
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5228818655014038,
      "learning_rate": 8.122755141517446e-07,
      "loss": 0.3756,
      "step": 27818
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.521963119506836,
      "learning_rate": 8.118735060474825e-07,
      "loss": 0.406,
      "step": 27819
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5658559799194336,
      "learning_rate": 8.114715932387839e-07,
      "loss": 0.3786,
      "step": 27820
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5393226146697998,
      "learning_rate": 8.110697757298202e-07,
      "loss": 0.399,
      "step": 27821
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5929983854293823,
      "learning_rate": 8.106680535247569e-07,
      "loss": 0.3701,
      "step": 27822
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5515458583831787,
      "learning_rate": 8.102664266277582e-07,
      "loss": 0.4683,
      "step": 27823
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5538781881332397,
      "learning_rate": 8.098648950429921e-07,
      "loss": 0.3753,
      "step": 27824
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5067024230957031,
      "learning_rate": 8.094634587746242e-07,
      "loss": 0.3963,
      "step": 27825
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.494046926498413,
      "learning_rate": 8.090621178268155e-07,
      "loss": 0.3363,
      "step": 27826
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5894352197647095,
      "learning_rate": 8.086608722037315e-07,
      "loss": 0.4296,
      "step": 27827
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5208860635757446,
      "learning_rate": 8.082597219095289e-07,
      "loss": 0.3919,
      "step": 27828
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5389420986175537,
      "learning_rate": 8.078586669483745e-07,
      "loss": 0.4341,
      "step": 27829
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5968024730682373,
      "learning_rate": 8.074577073244227e-07,
      "loss": 0.3979,
      "step": 27830
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7026304006576538,
      "learning_rate": 8.070568430418335e-07,
      "loss": 0.4596,
      "step": 27831
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5204029083251953,
      "learning_rate": 8.066560741047669e-07,
      "loss": 0.3727,
      "step": 27832
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5672588348388672,
      "learning_rate": 8.062554005173784e-07,
      "loss": 0.3949,
      "step": 27833
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5228837728500366,
      "learning_rate": 8.058548222838203e-07,
      "loss": 0.3915,
      "step": 27834
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6045494079589844,
      "learning_rate": 8.054543394082503e-07,
      "loss": 0.4258,
      "step": 27835
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5824885368347168,
      "learning_rate": 8.050539518948242e-07,
      "loss": 0.3717,
      "step": 27836
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5543676614761353,
      "learning_rate": 8.046536597476906e-07,
      "loss": 0.4078,
      "step": 27837
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5185796022415161,
      "learning_rate": 8.04253462971002e-07,
      "loss": 0.3846,
      "step": 27838
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5254234075546265,
      "learning_rate": 8.038533615689092e-07,
      "loss": 0.4145,
      "step": 27839
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.498095989227295,
      "learning_rate": 8.034533555455626e-07,
      "loss": 0.3544,
      "step": 27840
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.516395092010498,
      "learning_rate": 8.03053444905113e-07,
      "loss": 0.4184,
      "step": 27841
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5510005950927734,
      "learning_rate": 8.026536296517051e-07,
      "loss": 0.3535,
      "step": 27842
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7038241624832153,
      "learning_rate": 8.022539097894844e-07,
      "loss": 0.4908,
      "step": 27843
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4931262731552124,
      "learning_rate": 8.018542853225997e-07,
      "loss": 0.3644,
      "step": 27844
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4890196323394775,
      "learning_rate": 8.014547562551955e-07,
      "loss": 0.3945,
      "step": 27845
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5738788843154907,
      "learning_rate": 8.010553225914131e-07,
      "loss": 0.4347,
      "step": 27846
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5802886486053467,
      "learning_rate": 8.006559843353979e-07,
      "loss": 0.4647,
      "step": 27847
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4683241844177246,
      "learning_rate": 8.002567414912899e-07,
      "loss": 0.3658,
      "step": 27848
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7098889350891113,
      "learning_rate": 7.998575940632325e-07,
      "loss": 0.4847,
      "step": 27849
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5602771043777466,
      "learning_rate": 7.994585420553613e-07,
      "loss": 0.3823,
      "step": 27850
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5969669818878174,
      "learning_rate": 7.990595854718175e-07,
      "loss": 0.4038,
      "step": 27851
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.541368007659912,
      "learning_rate": 7.98660724316741e-07,
      "loss": 0.3735,
      "step": 27852
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5582491159439087,
      "learning_rate": 7.982619585942664e-07,
      "loss": 0.4015,
      "step": 27853
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5199494361877441,
      "learning_rate": 7.978632883085269e-07,
      "loss": 0.3737,
      "step": 27854
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.675471544265747,
      "learning_rate": 7.974647134636615e-07,
      "loss": 0.4105,
      "step": 27855
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.570762038230896,
      "learning_rate": 7.970662340638014e-07,
      "loss": 0.3491,
      "step": 27856
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4942013025283813,
      "learning_rate": 7.966678501130831e-07,
      "loss": 0.4017,
      "step": 27857
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5194811820983887,
      "learning_rate": 7.962695616156357e-07,
      "loss": 0.3639,
      "step": 27858
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4683196544647217,
      "learning_rate": 7.958713685755893e-07,
      "loss": 0.4172,
      "step": 27859
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.523665189743042,
      "learning_rate": 7.954732709970747e-07,
      "loss": 0.382,
      "step": 27860
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6572202444076538,
      "learning_rate": 7.950752688842222e-07,
      "loss": 0.4321,
      "step": 27861
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5230703353881836,
      "learning_rate": 7.946773622411574e-07,
      "loss": 0.382,
      "step": 27862
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5724916458129883,
      "learning_rate": 7.942795510720103e-07,
      "loss": 0.9379,
      "step": 27863
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4405077695846558,
      "learning_rate": 7.93881835380903e-07,
      "loss": 0.976,
      "step": 27864
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5513405799865723,
      "learning_rate": 7.934842151719647e-07,
      "loss": 0.4066,
      "step": 27865
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5510669946670532,
      "learning_rate": 7.93086690449314e-07,
      "loss": 0.374,
      "step": 27866
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5259937047958374,
      "learning_rate": 7.926892612170778e-07,
      "loss": 0.4018,
      "step": 27867
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.533626675605774,
      "learning_rate": 7.922919274793794e-07,
      "loss": 0.3758,
      "step": 27868
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5095807313919067,
      "learning_rate": 7.918946892403367e-07,
      "loss": 0.4066,
      "step": 27869
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5006380081176758,
      "learning_rate": 7.914975465040686e-07,
      "loss": 0.3767,
      "step": 27870
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6287001371383667,
      "learning_rate": 7.911004992746974e-07,
      "loss": 0.4394,
      "step": 27871
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4647859334945679,
      "learning_rate": 7.907035475563407e-07,
      "loss": 0.3496,
      "step": 27872
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6327080726623535,
      "learning_rate": 7.903066913531144e-07,
      "loss": 0.4051,
      "step": 27873
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6978358030319214,
      "learning_rate": 7.899099306691338e-07,
      "loss": 0.3819,
      "step": 27874
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5699740648269653,
      "learning_rate": 7.895132655085135e-07,
      "loss": 0.4515,
      "step": 27875
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5256625413894653,
      "learning_rate": 7.891166958753704e-07,
      "loss": 0.3812,
      "step": 27876
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.595456838607788,
      "learning_rate": 7.887202217738177e-07,
      "loss": 0.3967,
      "step": 27877
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.520222783088684,
      "learning_rate": 7.883238432079654e-07,
      "loss": 0.3479,
      "step": 27878
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7337874174118042,
      "learning_rate": 7.879275601819237e-07,
      "loss": 0.3976,
      "step": 27879
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6225301027297974,
      "learning_rate": 7.875313726998035e-07,
      "loss": 0.4465,
      "step": 27880
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.49781334400177,
      "learning_rate": 7.871352807657162e-07,
      "loss": 0.4353,
      "step": 27881
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5241870880126953,
      "learning_rate": 7.867392843837662e-07,
      "loss": 0.3854,
      "step": 27882
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7427880764007568,
      "learning_rate": 7.863433835580647e-07,
      "loss": 0.4625,
      "step": 27883
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4856551885604858,
      "learning_rate": 7.859475782927151e-07,
      "loss": 0.3682,
      "step": 27884
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.68759286403656,
      "learning_rate": 7.855518685918229e-07,
      "loss": 0.4973,
      "step": 27885
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6091909408569336,
      "learning_rate": 7.851562544594915e-07,
      "loss": 0.3888,
      "step": 27886
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5931850671768188,
      "learning_rate": 7.847607358998255e-07,
      "loss": 0.4226,
      "step": 27887
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5958881378173828,
      "learning_rate": 7.843653129169282e-07,
      "loss": 0.388,
      "step": 27888
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.503497838973999,
      "learning_rate": 7.839699855148985e-07,
      "loss": 0.3836,
      "step": 27889
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5243349075317383,
      "learning_rate": 7.835747536978367e-07,
      "loss": 0.3713,
      "step": 27890
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5461666584014893,
      "learning_rate": 7.831796174698414e-07,
      "loss": 0.4299,
      "step": 27891
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4795030355453491,
      "learning_rate": 7.827845768350118e-07,
      "loss": 0.3566,
      "step": 27892
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5500996112823486,
      "learning_rate": 7.823896317974477e-07,
      "loss": 0.5112,
      "step": 27893
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5228796005249023,
      "learning_rate": 7.819947823612428e-07,
      "loss": 0.3831,
      "step": 27894
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4867089986801147,
      "learning_rate": 7.816000285304903e-07,
      "loss": 0.4313,
      "step": 27895
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.54976224899292,
      "learning_rate": 7.812053703092859e-07,
      "loss": 0.3744,
      "step": 27896
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5750994682312012,
      "learning_rate": 7.808108077017263e-07,
      "loss": 0.4105,
      "step": 27897
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5724499225616455,
      "learning_rate": 7.804163407118981e-07,
      "loss": 0.3937,
      "step": 27898
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6040149927139282,
      "learning_rate": 7.800219693438981e-07,
      "loss": 0.3982,
      "step": 27899
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7358653545379639,
      "learning_rate": 7.79627693601811e-07,
      "loss": 0.4031,
      "step": 27900
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.543537974357605,
      "learning_rate": 7.792335134897311e-07,
      "loss": 0.419,
      "step": 27901
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.479573130607605,
      "learning_rate": 7.788394290117429e-07,
      "loss": 0.3719,
      "step": 27902
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7612905502319336,
      "learning_rate": 7.784454401719344e-07,
      "loss": 0.4498,
      "step": 27903
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4828832149505615,
      "learning_rate": 7.780515469743955e-07,
      "loss": 0.384,
      "step": 27904
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6134387254714966,
      "learning_rate": 7.776577494232085e-07,
      "loss": 0.4563,
      "step": 27905
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4815634489059448,
      "learning_rate": 7.772640475224557e-07,
      "loss": 0.3712,
      "step": 27906
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.60403573513031,
      "learning_rate": 7.768704412762229e-07,
      "loss": 0.4291,
      "step": 27907
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4657769203186035,
      "learning_rate": 7.76476930688591e-07,
      "loss": 0.3852,
      "step": 27908
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5503723621368408,
      "learning_rate": 7.76083515763647e-07,
      "loss": 0.4157,
      "step": 27909
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5529147386550903,
      "learning_rate": 7.756901965054642e-07,
      "loss": 0.3895,
      "step": 27910
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7386339902877808,
      "learning_rate": 7.752969729181226e-07,
      "loss": 0.4453,
      "step": 27911
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.528205394744873,
      "learning_rate": 7.749038450057033e-07,
      "loss": 0.3599,
      "step": 27912
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6135046482086182,
      "learning_rate": 7.745108127722845e-07,
      "loss": 0.4639,
      "step": 27913
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5932319164276123,
      "learning_rate": 7.741178762219414e-07,
      "loss": 0.3957,
      "step": 27914
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.572314977645874,
      "learning_rate": 7.737250353587467e-07,
      "loss": 0.4214,
      "step": 27915
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4998462200164795,
      "learning_rate": 7.733322901867768e-07,
      "loss": 0.4028,
      "step": 27916
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6144410371780396,
      "learning_rate": 7.729396407101075e-07,
      "loss": 0.4414,
      "step": 27917
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6803468465805054,
      "learning_rate": 7.725470869328067e-07,
      "loss": 0.3658,
      "step": 27918
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5480692386627197,
      "learning_rate": 7.721546288589499e-07,
      "loss": 0.432,
      "step": 27919
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6140148639678955,
      "learning_rate": 7.717622664926039e-07,
      "loss": 0.3717,
      "step": 27920
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6995302438735962,
      "learning_rate": 7.713699998378421e-07,
      "loss": 0.3955,
      "step": 27921
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.580161690711975,
      "learning_rate": 7.709778288987279e-07,
      "loss": 0.3742,
      "step": 27922
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.585399866104126,
      "learning_rate": 7.705857536793327e-07,
      "loss": 0.3948,
      "step": 27923
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4588040113449097,
      "learning_rate": 7.701937741837228e-07,
      "loss": 0.3625,
      "step": 27924
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.517242431640625,
      "learning_rate": 7.698018904159632e-07,
      "loss": 0.4343,
      "step": 27925
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.45781672000885,
      "learning_rate": 7.694101023801159e-07,
      "loss": 0.3897,
      "step": 27926
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5006248950958252,
      "learning_rate": 7.690184100802456e-07,
      "loss": 0.3764,
      "step": 27927
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5931310653686523,
      "learning_rate": 7.686268135204167e-07,
      "loss": 0.3675,
      "step": 27928
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6056129932403564,
      "learning_rate": 7.682353127046904e-07,
      "loss": 0.4556,
      "step": 27929
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4591326713562012,
      "learning_rate": 7.678439076371269e-07,
      "loss": 0.3426,
      "step": 27930
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5727201700210571,
      "learning_rate": 7.674525983217829e-07,
      "loss": 0.4299,
      "step": 27931
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5505727529525757,
      "learning_rate": 7.670613847627184e-07,
      "loss": 0.3747,
      "step": 27932
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7114005088806152,
      "learning_rate": 7.666702669639936e-07,
      "loss": 0.4096,
      "step": 27933
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.45511794090271,
      "learning_rate": 7.662792449296619e-07,
      "loss": 0.3666,
      "step": 27934
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5422334671020508,
      "learning_rate": 7.658883186637811e-07,
      "loss": 0.3775,
      "step": 27935
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5159947872161865,
      "learning_rate": 7.654974881704036e-07,
      "loss": 0.3843,
      "step": 27936
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5801653861999512,
      "learning_rate": 7.65106753453585e-07,
      "loss": 0.4196,
      "step": 27937
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5988017320632935,
      "learning_rate": 7.647161145173743e-07,
      "loss": 0.3777,
      "step": 27938
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5284191370010376,
      "learning_rate": 7.64325571365826e-07,
      "loss": 0.4073,
      "step": 27939
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5717334747314453,
      "learning_rate": 7.639351240029924e-07,
      "loss": 0.3713,
      "step": 27940
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6301394701004028,
      "learning_rate": 7.635447724329203e-07,
      "loss": 0.4296,
      "step": 27941
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5350189208984375,
      "learning_rate": 7.631545166596566e-07,
      "loss": 0.3728,
      "step": 27942
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6655982732772827,
      "learning_rate": 7.627643566872523e-07,
      "loss": 0.4362,
      "step": 27943
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.448376178741455,
      "learning_rate": 7.623742925197519e-07,
      "loss": 0.363,
      "step": 27944
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5828523635864258,
      "learning_rate": 7.619843241612046e-07,
      "loss": 0.4178,
      "step": 27945
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6804051399230957,
      "learning_rate": 7.615944516156493e-07,
      "loss": 0.3817,
      "step": 27946
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5760548114776611,
      "learning_rate": 7.612046748871327e-07,
      "loss": 0.45,
      "step": 27947
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5254050493240356,
      "learning_rate": 7.608149939796972e-07,
      "loss": 0.3982,
      "step": 27948
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5801767110824585,
      "learning_rate": 7.604254088973839e-07,
      "loss": 0.4201,
      "step": 27949
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5577510595321655,
      "learning_rate": 7.600359196442386e-07,
      "loss": 0.3847,
      "step": 27950
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.495586633682251,
      "learning_rate": 7.596465262242914e-07,
      "loss": 0.446,
      "step": 27951
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4789438247680664,
      "learning_rate": 7.592572286415867e-07,
      "loss": 0.3533,
      "step": 27952
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5066944360733032,
      "learning_rate": 7.588680269001614e-07,
      "loss": 0.4622,
      "step": 27953
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4704924821853638,
      "learning_rate": 7.584789210040533e-07,
      "loss": 0.3744,
      "step": 27954
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5953935384750366,
      "learning_rate": 7.580899109572959e-07,
      "loss": 0.4276,
      "step": 27955
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5665439367294312,
      "learning_rate": 7.577009967639237e-07,
      "loss": 0.4003,
      "step": 27956
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6520951986312866,
      "learning_rate": 7.573121784279713e-07,
      "loss": 0.448,
      "step": 27957
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5555826425552368,
      "learning_rate": 7.569234559534733e-07,
      "loss": 0.4223,
      "step": 27958
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4504272937774658,
      "learning_rate": 7.565348293444586e-07,
      "loss": 0.3964,
      "step": 27959
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7805360555648804,
      "learning_rate": 7.561462986049595e-07,
      "loss": 0.4001,
      "step": 27960
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.546248435974121,
      "learning_rate": 7.557578637390029e-07,
      "loss": 0.4167,
      "step": 27961
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5084666013717651,
      "learning_rate": 7.553695247506221e-07,
      "loss": 0.3846,
      "step": 27962
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.36034893989563,
      "learning_rate": 7.549812816438395e-07,
      "loss": 0.4302,
      "step": 27963
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5360080003738403,
      "learning_rate": 7.545931344226853e-07,
      "loss": 0.3806,
      "step": 27964
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.454896092414856,
      "learning_rate": 7.542050830911862e-07,
      "loss": 0.3657,
      "step": 27965
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5286521911621094,
      "learning_rate": 7.538171276533645e-07,
      "loss": 0.3614,
      "step": 27966
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7856647968292236,
      "learning_rate": 7.534292681132438e-07,
      "loss": 0.4414,
      "step": 27967
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5116413831710815,
      "learning_rate": 7.530415044748473e-07,
      "loss": 0.3775,
      "step": 27968
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5385147333145142,
      "learning_rate": 7.526538367421965e-07,
      "loss": 0.4444,
      "step": 27969
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5276670455932617,
      "learning_rate": 7.522662649193146e-07,
      "loss": 0.4049,
      "step": 27970
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.633712887763977,
      "learning_rate": 7.518787890102198e-07,
      "loss": 0.4683,
      "step": 27971
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6081782579421997,
      "learning_rate": 7.514914090189284e-07,
      "loss": 0.4084,
      "step": 27972
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.469834566116333,
      "learning_rate": 7.511041249494599e-07,
      "loss": 0.3874,
      "step": 27973
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5016316175460815,
      "learning_rate": 7.507169368058342e-07,
      "loss": 0.3773,
      "step": 27974
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.502996563911438,
      "learning_rate": 7.503298445920615e-07,
      "loss": 0.4176,
      "step": 27975
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6225656270980835,
      "learning_rate": 7.499428483121618e-07,
      "loss": 0.3912,
      "step": 27976
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5277235507965088,
      "learning_rate": 7.495559479701442e-07,
      "loss": 0.4388,
      "step": 27977
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.47451651096344,
      "learning_rate": 7.491691435700243e-07,
      "loss": 0.3565,
      "step": 27978
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.537258505821228,
      "learning_rate": 7.487824351158135e-07,
      "loss": 0.424,
      "step": 27979
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5406525135040283,
      "learning_rate": 7.483958226115206e-07,
      "loss": 0.3856,
      "step": 27980
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6192206144332886,
      "learning_rate": 7.480093060611593e-07,
      "loss": 0.4033,
      "step": 27981
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5002598762512207,
      "learning_rate": 7.476228854687373e-07,
      "loss": 0.3981,
      "step": 27982
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5193309783935547,
      "learning_rate": 7.472365608382581e-07,
      "loss": 0.408,
      "step": 27983
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5847111940383911,
      "learning_rate": 7.468503321737319e-07,
      "loss": 0.3641,
      "step": 27984
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5525078773498535,
      "learning_rate": 7.464641994791643e-07,
      "loss": 0.3996,
      "step": 27985
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4905140399932861,
      "learning_rate": 7.460781627585645e-07,
      "loss": 0.3593,
      "step": 27986
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5436125993728638,
      "learning_rate": 7.45692222015928e-07,
      "loss": 0.4337,
      "step": 27987
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.566348671913147,
      "learning_rate": 7.453063772552616e-07,
      "loss": 0.3505,
      "step": 27988
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5329499244689941,
      "learning_rate": 7.449206284805666e-07,
      "loss": 0.443,
      "step": 27989
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5132371187210083,
      "learning_rate": 7.445349756958464e-07,
      "loss": 0.34,
      "step": 27990
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5119593143463135,
      "learning_rate": 7.441494189050979e-07,
      "loss": 0.3973,
      "step": 27991
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.53421151638031,
      "learning_rate": 7.437639581123202e-07,
      "loss": 0.3564,
      "step": 27992
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4178657531738281,
      "learning_rate": 7.43378593321511e-07,
      "loss": 0.3813,
      "step": 27993
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5556652545928955,
      "learning_rate": 7.429933245366694e-07,
      "loss": 0.408,
      "step": 27994
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5428991317749023,
      "learning_rate": 7.42608151761789e-07,
      "loss": 0.3933,
      "step": 27995
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5764042139053345,
      "learning_rate": 7.422230750008663e-07,
      "loss": 0.393,
      "step": 27996
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5602246522903442,
      "learning_rate": 7.418380942578929e-07,
      "loss": 0.4363,
      "step": 27997
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4862557649612427,
      "learning_rate": 7.414532095368654e-07,
      "loss": 0.3785,
      "step": 27998
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5656732320785522,
      "learning_rate": 7.410684208417706e-07,
      "loss": 0.4057,
      "step": 27999
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.483625888824463,
      "learning_rate": 7.406837281766021e-07,
      "loss": 0.3547,
      "step": 28000
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.566109538078308,
      "learning_rate": 7.402991315453512e-07,
      "loss": 0.4491,
      "step": 28001
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5531930923461914,
      "learning_rate": 7.399146309520056e-07,
      "loss": 0.3513,
      "step": 28002
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5376163721084595,
      "learning_rate": 7.395302264005522e-07,
      "loss": 0.4265,
      "step": 28003
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5542629957199097,
      "learning_rate": 7.391459178949778e-07,
      "loss": 0.3721,
      "step": 28004
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5718698501586914,
      "learning_rate": 7.387617054392681e-07,
      "loss": 0.4106,
      "step": 28005
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5180814266204834,
      "learning_rate": 7.383775890374123e-07,
      "loss": 0.3504,
      "step": 28006
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5390928983688354,
      "learning_rate": 7.379935686933903e-07,
      "loss": 0.4113,
      "step": 28007
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5379620790481567,
      "learning_rate": 7.376096444111835e-07,
      "loss": 0.37,
      "step": 28008
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6134527921676636,
      "learning_rate": 7.372258161947766e-07,
      "loss": 0.4168,
      "step": 28009
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5208662748336792,
      "learning_rate": 7.368420840481516e-07,
      "loss": 0.3812,
      "step": 28010
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4858900308609009,
      "learning_rate": 7.364584479752845e-07,
      "loss": 0.397,
      "step": 28011
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5236248970031738,
      "learning_rate": 7.360749079801587e-07,
      "loss": 0.368,
      "step": 28012
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6921045780181885,
      "learning_rate": 7.356914640667478e-07,
      "loss": 0.4464,
      "step": 28013
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5024256706237793,
      "learning_rate": 7.35308116239033e-07,
      "loss": 0.3703,
      "step": 28014
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6382793188095093,
      "learning_rate": 7.349248645009866e-07,
      "loss": 0.4241,
      "step": 28015
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7111092805862427,
      "learning_rate": 7.345417088565843e-07,
      "loss": 0.348,
      "step": 28016
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5714020729064941,
      "learning_rate": 7.34158649309803e-07,
      "loss": 0.4041,
      "step": 28017
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5307153463363647,
      "learning_rate": 7.337756858646128e-07,
      "loss": 0.3908,
      "step": 28018
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.540602684020996,
      "learning_rate": 7.333928185249839e-07,
      "loss": 0.4232,
      "step": 28019
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5278775691986084,
      "learning_rate": 7.330100472948909e-07,
      "loss": 0.3602,
      "step": 28020
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.569598913192749,
      "learning_rate": 7.326273721783006e-07,
      "loss": 0.4449,
      "step": 28021
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.528870940208435,
      "learning_rate": 7.322447931791866e-07,
      "loss": 0.3808,
      "step": 28022
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5227283239364624,
      "learning_rate": 7.318623103015143e-07,
      "loss": 0.4011,
      "step": 28023
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5235135555267334,
      "learning_rate": 7.314799235492487e-07,
      "loss": 0.3612,
      "step": 28024
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.572838306427002,
      "learning_rate": 7.310976329263564e-07,
      "loss": 0.4732,
      "step": 28025
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.492531418800354,
      "learning_rate": 7.307154384368054e-07,
      "loss": 0.3854,
      "step": 28026
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.52521550655365,
      "learning_rate": 7.303333400845581e-07,
      "loss": 0.427,
      "step": 28027
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5334413051605225,
      "learning_rate": 7.299513378735745e-07,
      "loss": 0.3712,
      "step": 28028
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5873819589614868,
      "learning_rate": 7.295694318078206e-07,
      "loss": 0.4111,
      "step": 28029
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4859651327133179,
      "learning_rate": 7.291876218912564e-07,
      "loss": 0.3585,
      "step": 28030
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6591335535049438,
      "learning_rate": 7.288059081278398e-07,
      "loss": 0.3997,
      "step": 28031
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5429829359054565,
      "learning_rate": 7.284242905215333e-07,
      "loss": 0.3629,
      "step": 28032
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5540611743927002,
      "learning_rate": 7.280427690762903e-07,
      "loss": 0.427,
      "step": 28033
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.54288649559021,
      "learning_rate": 7.276613437960733e-07,
      "loss": 0.3854,
      "step": 28034
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5344274044036865,
      "learning_rate": 7.272800146848324e-07,
      "loss": 0.3986,
      "step": 28035
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5037916898727417,
      "learning_rate": 7.268987817465268e-07,
      "loss": 0.3937,
      "step": 28036
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7271087169647217,
      "learning_rate": 7.265176449851108e-07,
      "loss": 0.3985,
      "step": 28037
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.468675971031189,
      "learning_rate": 7.261366044045348e-07,
      "loss": 0.3657,
      "step": 28038
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5654830932617188,
      "learning_rate": 7.257556600087512e-07,
      "loss": 0.4272,
      "step": 28039
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5058047771453857,
      "learning_rate": 7.253748118017113e-07,
      "loss": 0.3731,
      "step": 28040
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4844715595245361,
      "learning_rate": 7.249940597873661e-07,
      "loss": 0.4092,
      "step": 28041
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5734208822250366,
      "learning_rate": 7.24613403969665e-07,
      "loss": 0.3821,
      "step": 28042
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5608479976654053,
      "learning_rate": 7.242328443525548e-07,
      "loss": 0.4171,
      "step": 28043
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5413358211517334,
      "learning_rate": 7.238523809399822e-07,
      "loss": 0.4179,
      "step": 28044
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6256072521209717,
      "learning_rate": 7.234720137358919e-07,
      "loss": 0.4578,
      "step": 28045
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5326350927352905,
      "learning_rate": 7.230917427442341e-07,
      "loss": 0.3836,
      "step": 28046
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.732358694076538,
      "learning_rate": 7.227115679689467e-07,
      "loss": 1.0196,
      "step": 28047
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5905483961105347,
      "learning_rate": 7.223314894139777e-07,
      "loss": 1.0318,
      "step": 28048
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5209733247756958,
      "learning_rate": 7.21951507083265e-07,
      "loss": 0.4054,
      "step": 28049
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5642565488815308,
      "learning_rate": 7.215716209807533e-07,
      "loss": 0.3672,
      "step": 28050
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5496288537979126,
      "learning_rate": 7.211918311103794e-07,
      "loss": 0.4214,
      "step": 28051
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.9898171424865723,
      "learning_rate": 7.208121374760835e-07,
      "loss": 0.3932,
      "step": 28052
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7095545530319214,
      "learning_rate": 7.204325400818057e-07,
      "loss": 0.451,
      "step": 28053
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.473296046257019,
      "learning_rate": 7.200530389314809e-07,
      "loss": 0.3736,
      "step": 28054
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6294275522232056,
      "learning_rate": 7.196736340290445e-07,
      "loss": 0.4166,
      "step": 28055
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5298908948898315,
      "learning_rate": 7.192943253784324e-07,
      "loss": 0.397,
      "step": 28056
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4405713081359863,
      "learning_rate": 7.18915112983578e-07,
      "loss": 0.3764,
      "step": 28057
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5601996183395386,
      "learning_rate": 7.185359968484173e-07,
      "loss": 0.38,
      "step": 28058
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8356910943984985,
      "learning_rate": 7.181569769768792e-07,
      "loss": 0.471,
      "step": 28059
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5645971298217773,
      "learning_rate": 7.177780533728951e-07,
      "loss": 0.3786,
      "step": 28060
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5607800483703613,
      "learning_rate": 7.173992260403951e-07,
      "loss": 0.3962,
      "step": 28061
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5307435989379883,
      "learning_rate": 7.170204949833115e-07,
      "loss": 0.3706,
      "step": 28062
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5721760988235474,
      "learning_rate": 7.166418602055681e-07,
      "loss": 0.4643,
      "step": 28063
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5496174097061157,
      "learning_rate": 7.162633217110926e-07,
      "loss": 0.3776,
      "step": 28064
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5019744634628296,
      "learning_rate": 7.15884879503812e-07,
      "loss": 0.447,
      "step": 28065
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.0504462718963623,
      "learning_rate": 7.155065335876521e-07,
      "loss": 0.4033,
      "step": 28066
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4872232675552368,
      "learning_rate": 7.151282839665353e-07,
      "loss": 0.4007,
      "step": 28067
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4557042121887207,
      "learning_rate": 7.147501306443872e-07,
      "loss": 0.3533,
      "step": 28068
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5235130786895752,
      "learning_rate": 7.143720736251259e-07,
      "loss": 0.3933,
      "step": 28069
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4955055713653564,
      "learning_rate": 7.13994112912676e-07,
      "loss": 0.3681,
      "step": 28070
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4836632013320923,
      "learning_rate": 7.136162485109555e-07,
      "loss": 0.4098,
      "step": 28071
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5288516283035278,
      "learning_rate": 7.132384804238834e-07,
      "loss": 0.3716,
      "step": 28072
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5962213277816772,
      "learning_rate": 7.128608086553801e-07,
      "loss": 0.4204,
      "step": 28073
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6237192153930664,
      "learning_rate": 7.124832332093623e-07,
      "loss": 0.3846,
      "step": 28074
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4852356910705566,
      "learning_rate": 7.121057540897424e-07,
      "loss": 0.4397,
      "step": 28075
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.486458420753479,
      "learning_rate": 7.117283713004375e-07,
      "loss": 0.3538,
      "step": 28076
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5422917604446411,
      "learning_rate": 7.113510848453631e-07,
      "loss": 0.4244,
      "step": 28077
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5561389923095703,
      "learning_rate": 7.109738947284317e-07,
      "loss": 0.3906,
      "step": 28078
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6768052577972412,
      "learning_rate": 7.105968009535558e-07,
      "loss": 0.3929,
      "step": 28079
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5823639631271362,
      "learning_rate": 7.102198035246433e-07,
      "loss": 0.358,
      "step": 28080
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6897664070129395,
      "learning_rate": 7.098429024456077e-07,
      "loss": 0.4024,
      "step": 28081
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5969856977462769,
      "learning_rate": 7.094660977203571e-07,
      "loss": 0.3891,
      "step": 28082
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7062724828720093,
      "learning_rate": 7.090893893527984e-07,
      "loss": 0.4329,
      "step": 28083
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5763911008834839,
      "learning_rate": 7.087127773468416e-07,
      "loss": 0.3823,
      "step": 28084
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5318154096603394,
      "learning_rate": 7.083362617063894e-07,
      "loss": 0.4012,
      "step": 28085
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5428396463394165,
      "learning_rate": 7.079598424353496e-07,
      "loss": 0.3747,
      "step": 28086
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5635993480682373,
      "learning_rate": 7.075835195376223e-07,
      "loss": 0.4337,
      "step": 28087
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.66937255859375,
      "learning_rate": 7.072072930171148e-07,
      "loss": 0.3791,
      "step": 28088
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.51729416847229,
      "learning_rate": 7.068311628777281e-07,
      "loss": 0.4501,
      "step": 28089
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.525922417640686,
      "learning_rate": 7.064551291233635e-07,
      "loss": 0.375,
      "step": 28090
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6806986331939697,
      "learning_rate": 7.06079191757918e-07,
      "loss": 0.4502,
      "step": 28091
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5105180740356445,
      "learning_rate": 7.057033507852939e-07,
      "loss": 0.3755,
      "step": 28092
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5321193933486938,
      "learning_rate": 7.053276062093884e-07,
      "loss": 0.4046,
      "step": 28093
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5344902276992798,
      "learning_rate": 7.049519580340991e-07,
      "loss": 0.3854,
      "step": 28094
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5420539379119873,
      "learning_rate": 7.045764062633221e-07,
      "loss": 0.4073,
      "step": 28095
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5006870031356812,
      "learning_rate": 7.042009509009506e-07,
      "loss": 0.3697,
      "step": 28096
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6275972127914429,
      "learning_rate": 7.038255919508796e-07,
      "loss": 0.4468,
      "step": 28097
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.462090253829956,
      "learning_rate": 7.034503294170037e-07,
      "loss": 0.3675,
      "step": 28098
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5961788892745972,
      "learning_rate": 7.030751633032129e-07,
      "loss": 0.4442,
      "step": 28099
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5609445571899414,
      "learning_rate": 7.02700093613401e-07,
      "loss": 0.3854,
      "step": 28100
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5467489957809448,
      "learning_rate": 7.023251203514547e-07,
      "loss": 0.4291,
      "step": 28101
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6101093292236328,
      "learning_rate": 7.019502435212655e-07,
      "loss": 0.3898,
      "step": 28102
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5808334350585938,
      "learning_rate": 7.015754631267191e-07,
      "loss": 0.4306,
      "step": 28103
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4958369731903076,
      "learning_rate": 7.012007791717057e-07,
      "loss": 0.3834,
      "step": 28104
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5536682605743408,
      "learning_rate": 7.008261916601089e-07,
      "loss": 0.426,
      "step": 28105
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6372733116149902,
      "learning_rate": 7.004517005958156e-07,
      "loss": 0.3633,
      "step": 28106
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5623584985733032,
      "learning_rate": 7.000773059827071e-07,
      "loss": 0.4548,
      "step": 28107
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.455133318901062,
      "learning_rate": 6.997030078246691e-07,
      "loss": 0.3593,
      "step": 28108
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.545058012008667,
      "learning_rate": 6.993288061255843e-07,
      "loss": 0.4229,
      "step": 28109
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5453580617904663,
      "learning_rate": 6.989547008893315e-07,
      "loss": 0.3783,
      "step": 28110
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6792134046554565,
      "learning_rate": 6.985806921197902e-07,
      "loss": 0.491,
      "step": 28111
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5188368558883667,
      "learning_rate": 6.982067798208414e-07,
      "loss": 0.3648,
      "step": 28112
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.543075442314148,
      "learning_rate": 6.978329639963622e-07,
      "loss": 0.4069,
      "step": 28113
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.512475609779358,
      "learning_rate": 6.974592446502315e-07,
      "loss": 0.3583,
      "step": 28114
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5874505043029785,
      "learning_rate": 6.970856217863242e-07,
      "loss": 0.4124,
      "step": 28115
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5893614292144775,
      "learning_rate": 6.967120954085138e-07,
      "loss": 0.377,
      "step": 28116
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.523472785949707,
      "learning_rate": 6.963386655206749e-07,
      "loss": 0.4087,
      "step": 28117
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5874824523925781,
      "learning_rate": 6.959653321266835e-07,
      "loss": 0.4283,
      "step": 28118
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.594205617904663,
      "learning_rate": 6.955920952304085e-07,
      "loss": 0.4138,
      "step": 28119
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5457395315170288,
      "learning_rate": 6.952189548357225e-07,
      "loss": 0.3584,
      "step": 28120
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5386707782745361,
      "learning_rate": 6.948459109464944e-07,
      "loss": 0.404,
      "step": 28121
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.485569953918457,
      "learning_rate": 6.944729635665948e-07,
      "loss": 0.3693,
      "step": 28122
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6288495063781738,
      "learning_rate": 6.941001126998892e-07,
      "loss": 0.4166,
      "step": 28123
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6044267416000366,
      "learning_rate": 6.937273583502468e-07,
      "loss": 0.3946,
      "step": 28124
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5663491487503052,
      "learning_rate": 6.933547005215346e-07,
      "loss": 0.4044,
      "step": 28125
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4680382013320923,
      "learning_rate": 6.929821392176162e-07,
      "loss": 0.3285,
      "step": 28126
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5516818761825562,
      "learning_rate": 6.926096744423538e-07,
      "loss": 0.4183,
      "step": 28127
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4872238636016846,
      "learning_rate": 6.922373061996123e-07,
      "loss": 0.3612,
      "step": 28128
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5931586027145386,
      "learning_rate": 6.918650344932543e-07,
      "loss": 0.4331,
      "step": 28129
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4701457023620605,
      "learning_rate": 6.914928593271419e-07,
      "loss": 0.3604,
      "step": 28130
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5033825635910034,
      "learning_rate": 6.911207807051335e-07,
      "loss": 0.4056,
      "step": 28131
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5187747478485107,
      "learning_rate": 6.907487986310868e-07,
      "loss": 0.3723,
      "step": 28132
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5229737758636475,
      "learning_rate": 6.903769131088611e-07,
      "loss": 0.4164,
      "step": 28133
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5267603397369385,
      "learning_rate": 6.900051241423155e-07,
      "loss": 0.3802,
      "step": 28134
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5358548164367676,
      "learning_rate": 6.896334317353026e-07,
      "loss": 0.3834,
      "step": 28135
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5351163148880005,
      "learning_rate": 6.892618358916803e-07,
      "loss": 0.3657,
      "step": 28136
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5602108240127563,
      "learning_rate": 6.888903366153e-07,
      "loss": 0.4339,
      "step": 28137
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.528752088546753,
      "learning_rate": 6.885189339100174e-07,
      "loss": 0.3673,
      "step": 28138
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5516300201416016,
      "learning_rate": 6.881476277796817e-07,
      "loss": 0.4392,
      "step": 28139
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5477937459945679,
      "learning_rate": 6.877764182281455e-07,
      "loss": 0.3592,
      "step": 28140
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5738822221755981,
      "learning_rate": 6.874053052592611e-07,
      "loss": 0.4359,
      "step": 28141
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4598886966705322,
      "learning_rate": 6.870342888768744e-07,
      "loss": 0.3477,
      "step": 28142
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5674535036087036,
      "learning_rate": 6.866633690848334e-07,
      "loss": 0.4355,
      "step": 28143
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.537168264389038,
      "learning_rate": 6.862925458869863e-07,
      "loss": 0.385,
      "step": 28144
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4659898281097412,
      "learning_rate": 6.859218192871797e-07,
      "loss": 0.354,
      "step": 28145
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5736377239227295,
      "learning_rate": 6.855511892892574e-07,
      "loss": 0.4236,
      "step": 28146
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5727746486663818,
      "learning_rate": 6.851806558970631e-07,
      "loss": 0.4622,
      "step": 28147
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5294835567474365,
      "learning_rate": 6.848102191144401e-07,
      "loss": 0.3597,
      "step": 28148
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5606156587600708,
      "learning_rate": 6.844398789452312e-07,
      "loss": 0.4133,
      "step": 28149
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.548699140548706,
      "learning_rate": 6.840696353932786e-07,
      "loss": 0.4053,
      "step": 28150
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.913451910018921,
      "learning_rate": 6.836994884624215e-07,
      "loss": 0.4429,
      "step": 28151
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.528783917427063,
      "learning_rate": 6.833294381564958e-07,
      "loss": 0.3585,
      "step": 28152
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5428916215896606,
      "learning_rate": 6.829594844793419e-07,
      "loss": 0.456,
      "step": 28153
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.504442572593689,
      "learning_rate": 6.825896274347987e-07,
      "loss": 0.3771,
      "step": 28154
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6015574932098389,
      "learning_rate": 6.822198670266989e-07,
      "loss": 0.4313,
      "step": 28155
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4600862264633179,
      "learning_rate": 6.818502032588802e-07,
      "loss": 0.3635,
      "step": 28156
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6030805110931396,
      "learning_rate": 6.814806361351733e-07,
      "loss": 0.4399,
      "step": 28157
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5896795988082886,
      "learning_rate": 6.811111656594161e-07,
      "loss": 0.3984,
      "step": 28158
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.514718770980835,
      "learning_rate": 6.807417918354354e-07,
      "loss": 0.4298,
      "step": 28159
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5010805130004883,
      "learning_rate": 6.803725146670648e-07,
      "loss": 0.3616,
      "step": 28160
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7746005058288574,
      "learning_rate": 6.800033341581358e-07,
      "loss": 0.4887,
      "step": 28161
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.510714054107666,
      "learning_rate": 6.796342503124753e-07,
      "loss": 0.3852,
      "step": 28162
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8894354104995728,
      "learning_rate": 6.792652631339103e-07,
      "loss": 0.9761,
      "step": 28163
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4546599388122559,
      "learning_rate": 6.788963726262687e-07,
      "loss": 0.9893,
      "step": 28164
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6621490716934204,
      "learning_rate": 6.785275787933776e-07,
      "loss": 0.4448,
      "step": 28165
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4457106590270996,
      "learning_rate": 6.781588816390616e-07,
      "loss": 0.3555,
      "step": 28166
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.599378228187561,
      "learning_rate": 6.777902811671455e-07,
      "loss": 0.4202,
      "step": 28167
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5686067342758179,
      "learning_rate": 6.774217773814495e-07,
      "loss": 0.3766,
      "step": 28168
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.547865867614746,
      "learning_rate": 6.770533702857973e-07,
      "loss": 0.4215,
      "step": 28169
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5344339609146118,
      "learning_rate": 6.766850598840114e-07,
      "loss": 0.3697,
      "step": 28170
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5840445756912231,
      "learning_rate": 6.763168461799085e-07,
      "loss": 0.4325,
      "step": 28171
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4295061826705933,
      "learning_rate": 6.759487291773104e-07,
      "loss": 0.3652,
      "step": 28172
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.565588116645813,
      "learning_rate": 6.755807088800325e-07,
      "loss": 0.4432,
      "step": 28173
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6601002216339111,
      "learning_rate": 6.752127852918943e-07,
      "loss": 0.435,
      "step": 28174
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4961804151535034,
      "learning_rate": 6.748449584167105e-07,
      "loss": 0.4157,
      "step": 28175
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5073668956756592,
      "learning_rate": 6.744772282582945e-07,
      "loss": 0.3703,
      "step": 28176
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4644482135772705,
      "learning_rate": 6.741095948204646e-07,
      "loss": 0.4029,
      "step": 28177
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5722695589065552,
      "learning_rate": 6.737420581070309e-07,
      "loss": 0.383,
      "step": 28178
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.858575463294983,
      "learning_rate": 6.733746181218037e-07,
      "loss": 0.4522,
      "step": 28179
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4822444915771484,
      "learning_rate": 6.730072748685956e-07,
      "loss": 0.3816,
      "step": 28180
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6026920080184937,
      "learning_rate": 6.72640028351218e-07,
      "loss": 0.4529,
      "step": 28181
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4626567363739014,
      "learning_rate": 6.722728785734789e-07,
      "loss": 0.3603,
      "step": 28182
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7459518909454346,
      "learning_rate": 6.719058255391842e-07,
      "loss": 0.4368,
      "step": 28183
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7215410470962524,
      "learning_rate": 6.71538869252143e-07,
      "loss": 0.3895,
      "step": 28184
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6823240518569946,
      "learning_rate": 6.7117200971616e-07,
      "loss": 0.4403,
      "step": 28185
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6112213134765625,
      "learning_rate": 6.708052469350424e-07,
      "loss": 0.3789,
      "step": 28186
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5751183032989502,
      "learning_rate": 6.704385809125924e-07,
      "loss": 0.4249,
      "step": 28187
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4991517066955566,
      "learning_rate": 6.700720116526116e-07,
      "loss": 0.3705,
      "step": 28188
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5983482599258423,
      "learning_rate": 6.697055391589036e-07,
      "loss": 0.4008,
      "step": 28189
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4558289051055908,
      "learning_rate": 6.693391634352686e-07,
      "loss": 0.3486,
      "step": 28190
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5400124788284302,
      "learning_rate": 6.689728844855092e-07,
      "loss": 0.4201,
      "step": 28191
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5084782838821411,
      "learning_rate": 6.686067023134224e-07,
      "loss": 0.3741,
      "step": 28192
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5641272068023682,
      "learning_rate": 6.682406169228029e-07,
      "loss": 0.4929,
      "step": 28193
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4010001420974731,
      "learning_rate": 6.678746283174509e-07,
      "loss": 0.3659,
      "step": 28194
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5735111236572266,
      "learning_rate": 6.675087365011645e-07,
      "loss": 0.4716,
      "step": 28195
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.577673316001892,
      "learning_rate": 6.671429414777331e-07,
      "loss": 0.3612,
      "step": 28196
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6410744190216064,
      "learning_rate": 6.667772432509556e-07,
      "loss": 0.9571,
      "step": 28197
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3736966848373413,
      "learning_rate": 6.664116418246202e-07,
      "loss": 0.9918,
      "step": 28198
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4797229766845703,
      "learning_rate": 6.66046137202524e-07,
      "loss": 0.3613,
      "step": 28199
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5954123735427856,
      "learning_rate": 6.656807293884526e-07,
      "loss": 0.3806,
      "step": 28200
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5741620063781738,
      "learning_rate": 6.653154183861988e-07,
      "loss": 0.4488,
      "step": 28201
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5279651880264282,
      "learning_rate": 6.649502041995526e-07,
      "loss": 0.3929,
      "step": 28202
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.567763090133667,
      "learning_rate": 6.645850868323e-07,
      "loss": 0.4655,
      "step": 28203
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4950190782546997,
      "learning_rate": 6.64220066288227e-07,
      "loss": 0.3783,
      "step": 28204
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5176422595977783,
      "learning_rate": 6.638551425711204e-07,
      "loss": 0.3829,
      "step": 28205
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4873582124710083,
      "learning_rate": 6.634903156847649e-07,
      "loss": 0.3732,
      "step": 28206
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6298329830169678,
      "learning_rate": 6.631255856329466e-07,
      "loss": 0.3976,
      "step": 28207
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4690086841583252,
      "learning_rate": 6.627609524194456e-07,
      "loss": 0.3595,
      "step": 28208
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5524145364761353,
      "learning_rate": 6.623964160480434e-07,
      "loss": 0.4105,
      "step": 28209
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4862614870071411,
      "learning_rate": 6.620319765225225e-07,
      "loss": 0.3635,
      "step": 28210
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5712660551071167,
      "learning_rate": 6.616676338466632e-07,
      "loss": 0.4608,
      "step": 28211
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4361728429794312,
      "learning_rate": 6.613033880242425e-07,
      "loss": 0.363,
      "step": 28212
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5530697107315063,
      "learning_rate": 6.609392390590397e-07,
      "loss": 0.3917,
      "step": 28213
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4878802299499512,
      "learning_rate": 6.605751869548294e-07,
      "loss": 0.3564,
      "step": 28214
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5314555168151855,
      "learning_rate": 6.602112317153908e-07,
      "loss": 0.418,
      "step": 28215
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.517099142074585,
      "learning_rate": 6.598473733444954e-07,
      "loss": 0.3743,
      "step": 28216
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7106273174285889,
      "learning_rate": 6.59483611845918e-07,
      "loss": 0.4449,
      "step": 28217
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.523695945739746,
      "learning_rate": 6.591199472234333e-07,
      "loss": 0.3813,
      "step": 28218
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.6244149208068848,
      "learning_rate": 6.587563794808127e-07,
      "loss": 0.4282,
      "step": 28219
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4902894496917725,
      "learning_rate": 6.583929086218232e-07,
      "loss": 0.3501,
      "step": 28220
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.62728750705719,
      "learning_rate": 6.580295346502374e-07,
      "loss": 0.429,
      "step": 28221
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5062013864517212,
      "learning_rate": 6.576662575698256e-07,
      "loss": 0.3661,
      "step": 28222
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.492214560508728,
      "learning_rate": 6.573030773843536e-07,
      "loss": 0.4064,
      "step": 28223
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5091625452041626,
      "learning_rate": 6.569399940975863e-07,
      "loss": 0.3483,
      "step": 28224
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5588289499282837,
      "learning_rate": 6.565770077132927e-07,
      "loss": 0.436,
      "step": 28225
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.585915446281433,
      "learning_rate": 6.562141182352355e-07,
      "loss": 0.3668,
      "step": 28226
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5576921701431274,
      "learning_rate": 6.558513256671817e-07,
      "loss": 0.4526,
      "step": 28227
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5141572952270508,
      "learning_rate": 6.554886300128916e-07,
      "loss": 0.3821,
      "step": 28228
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5612150430679321,
      "learning_rate": 6.551260312761254e-07,
      "loss": 0.3933,
      "step": 28229
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.585381269454956,
      "learning_rate": 6.547635294606447e-07,
      "loss": 0.3727,
      "step": 28230
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.539409875869751,
      "learning_rate": 6.54401124570212e-07,
      "loss": 0.3968,
      "step": 28231
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.65529203414917,
      "learning_rate": 6.540388166085831e-07,
      "loss": 0.4095,
      "step": 28232
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5983177423477173,
      "learning_rate": 6.536766055795174e-07,
      "loss": 0.4775,
      "step": 28233
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.566510796546936,
      "learning_rate": 6.533144914867695e-07,
      "loss": 0.3882,
      "step": 28234
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4857542514801025,
      "learning_rate": 6.529524743340987e-07,
      "loss": 0.4168,
      "step": 28235
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.563962697982788,
      "learning_rate": 6.525905541252553e-07,
      "loss": 0.3761,
      "step": 28236
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7331451177597046,
      "learning_rate": 6.522287308639952e-07,
      "loss": 0.4604,
      "step": 28237
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.501253366470337,
      "learning_rate": 6.51867004554072e-07,
      "loss": 0.3531,
      "step": 28238
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.574562907218933,
      "learning_rate": 6.515053751992373e-07,
      "loss": 0.4306,
      "step": 28239
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5136843919754028,
      "learning_rate": 6.51143842803238e-07,
      "loss": 0.3584,
      "step": 28240
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5500308275222778,
      "learning_rate": 6.507824073698277e-07,
      "loss": 0.4119,
      "step": 28241
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5328633785247803,
      "learning_rate": 6.504210689027535e-07,
      "loss": 0.3777,
      "step": 28242
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.556773066520691,
      "learning_rate": 6.500598274057657e-07,
      "loss": 0.4023,
      "step": 28243
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6149617433547974,
      "learning_rate": 6.49698682882608e-07,
      "loss": 0.3842,
      "step": 28244
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5787367820739746,
      "learning_rate": 6.493376353370262e-07,
      "loss": 0.4367,
      "step": 28245
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8484723567962646,
      "learning_rate": 6.489766847727652e-07,
      "loss": 0.3674,
      "step": 28246
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5869197845458984,
      "learning_rate": 6.486158311935698e-07,
      "loss": 0.4375,
      "step": 28247
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5524696111679077,
      "learning_rate": 6.482550746031812e-07,
      "loss": 0.3604,
      "step": 28248
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5605872869491577,
      "learning_rate": 6.478944150053434e-07,
      "loss": 0.4112,
      "step": 28249
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6118779182434082,
      "learning_rate": 6.475338524037933e-07,
      "loss": 0.3728,
      "step": 28250
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.52859365940094,
      "learning_rate": 6.471733868022745e-07,
      "loss": 0.3849,
      "step": 28251
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5721927881240845,
      "learning_rate": 6.468130182045207e-07,
      "loss": 0.4089,
      "step": 28252
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5459952354431152,
      "learning_rate": 6.464527466142734e-07,
      "loss": 0.3985,
      "step": 28253
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5188868045806885,
      "learning_rate": 6.460925720352707e-07,
      "loss": 0.3706,
      "step": 28254
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5265588760375977,
      "learning_rate": 6.45732494471244e-07,
      "loss": 0.4552,
      "step": 28255
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5194125175476074,
      "learning_rate": 6.453725139259292e-07,
      "loss": 0.3838,
      "step": 28256
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5753117799758911,
      "learning_rate": 6.450126304030591e-07,
      "loss": 0.4187,
      "step": 28257
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4940011501312256,
      "learning_rate": 6.446528439063704e-07,
      "loss": 0.3581,
      "step": 28258
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5695199966430664,
      "learning_rate": 6.442931544395903e-07,
      "loss": 0.4042,
      "step": 28259
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5103358030319214,
      "learning_rate": 6.439335620064502e-07,
      "loss": 0.4015,
      "step": 28260
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5307261943817139,
      "learning_rate": 6.435740666106793e-07,
      "loss": 0.4084,
      "step": 28261
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5384169816970825,
      "learning_rate": 6.43214668256007e-07,
      "loss": 0.3545,
      "step": 28262
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5017826557159424,
      "learning_rate": 6.428553669461634e-07,
      "loss": 0.4349,
      "step": 28263
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5189956426620483,
      "learning_rate": 6.424961626848714e-07,
      "loss": 0.3846,
      "step": 28264
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5577036142349243,
      "learning_rate": 6.421370554758566e-07,
      "loss": 0.4267,
      "step": 28265
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4842404127120972,
      "learning_rate": 6.41778045322845e-07,
      "loss": 0.3579,
      "step": 28266
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5419288873672485,
      "learning_rate": 6.414191322295604e-07,
      "loss": 0.423,
      "step": 28267
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5348570346832275,
      "learning_rate": 6.410603161997231e-07,
      "loss": 0.3742,
      "step": 28268
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.532717227935791,
      "learning_rate": 6.407015972370579e-07,
      "loss": 0.412,
      "step": 28269
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5215027332305908,
      "learning_rate": 6.403429753452816e-07,
      "loss": 0.3546,
      "step": 28270
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5284616947174072,
      "learning_rate": 6.399844505281172e-07,
      "loss": 0.4039,
      "step": 28271
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4385632276535034,
      "learning_rate": 6.396260227892803e-07,
      "loss": 0.3525,
      "step": 28272
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5092267990112305,
      "learning_rate": 6.392676921324881e-07,
      "loss": 0.429,
      "step": 28273
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4780136346817017,
      "learning_rate": 6.389094585614619e-07,
      "loss": 0.361,
      "step": 28274
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5529165267944336,
      "learning_rate": 6.385513220799122e-07,
      "loss": 0.4037,
      "step": 28275
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.522374153137207,
      "learning_rate": 6.381932826915549e-07,
      "loss": 0.3872,
      "step": 28276
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7343511581420898,
      "learning_rate": 6.378353404001025e-07,
      "loss": 0.4213,
      "step": 28277
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.575700283050537,
      "learning_rate": 6.374774952092688e-07,
      "loss": 0.3739,
      "step": 28278
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6726481914520264,
      "learning_rate": 6.371197471227664e-07,
      "loss": 0.4613,
      "step": 28279
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.602146863937378,
      "learning_rate": 6.367620961443044e-07,
      "loss": 0.3754,
      "step": 28280
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4769378900527954,
      "learning_rate": 6.3640454227759e-07,
      "loss": 0.4083,
      "step": 28281
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.792195439338684,
      "learning_rate": 6.360470855263334e-07,
      "loss": 0.3985,
      "step": 28282
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6127662658691406,
      "learning_rate": 6.356897258942452e-07,
      "loss": 0.4371,
      "step": 28283
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6885731220245361,
      "learning_rate": 6.353324633850267e-07,
      "loss": 0.4603,
      "step": 28284
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5605474710464478,
      "learning_rate": 6.34975298002386e-07,
      "loss": 0.4089,
      "step": 28285
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5348314046859741,
      "learning_rate": 6.346182297500258e-07,
      "loss": 0.3879,
      "step": 28286
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5936367511749268,
      "learning_rate": 6.342612586316532e-07,
      "loss": 0.4187,
      "step": 28287
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5775457620620728,
      "learning_rate": 6.339043846509651e-07,
      "loss": 0.3976,
      "step": 28288
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6876550912857056,
      "learning_rate": 6.335476078116664e-07,
      "loss": 0.9467,
      "step": 28289
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4673210382461548,
      "learning_rate": 6.331909281174575e-07,
      "loss": 1.0448,
      "step": 28290
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.612676978111267,
      "learning_rate": 6.328343455720376e-07,
      "loss": 0.43,
      "step": 28291
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.511366844177246,
      "learning_rate": 6.324778601791026e-07,
      "loss": 0.363,
      "step": 28292
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4643641710281372,
      "learning_rate": 6.321214719423518e-07,
      "loss": 0.3862,
      "step": 28293
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5330907106399536,
      "learning_rate": 6.317651808654813e-07,
      "loss": 0.3738,
      "step": 28294
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5652345418930054,
      "learning_rate": 6.3140898695219e-07,
      "loss": 0.4304,
      "step": 28295
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7372386455535889,
      "learning_rate": 6.310528902061641e-07,
      "loss": 0.3865,
      "step": 28296
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5144596099853516,
      "learning_rate": 6.306968906311029e-07,
      "loss": 0.4159,
      "step": 28297
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4380730390548706,
      "learning_rate": 6.303409882306966e-07,
      "loss": 0.3704,
      "step": 28298
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5753464698791504,
      "learning_rate": 6.299851830086389e-07,
      "loss": 0.4209,
      "step": 28299
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4988741874694824,
      "learning_rate": 6.296294749686182e-07,
      "loss": 0.3686,
      "step": 28300
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5615371465682983,
      "learning_rate": 6.292738641143226e-07,
      "loss": 0.4018,
      "step": 28301
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4375998973846436,
      "learning_rate": 6.2891835044944e-07,
      "loss": 0.3748,
      "step": 28302
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5870968103408813,
      "learning_rate": 6.285629339776622e-07,
      "loss": 0.4221,
      "step": 28303
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5367215871810913,
      "learning_rate": 6.282076147026705e-07,
      "loss": 0.3758,
      "step": 28304
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5780746936798096,
      "learning_rate": 6.278523926281544e-07,
      "loss": 0.4459,
      "step": 28305
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5322405099868774,
      "learning_rate": 6.27497267757794e-07,
      "loss": 0.3854,
      "step": 28306
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5685089826583862,
      "learning_rate": 6.271422400952753e-07,
      "loss": 0.4064,
      "step": 28307
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5287665128707886,
      "learning_rate": 6.267873096442778e-07,
      "loss": 0.3401,
      "step": 28308
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4752129316329956,
      "learning_rate": 6.26432476408485e-07,
      "loss": 0.4059,
      "step": 28309
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4768791198730469,
      "learning_rate": 6.260777403915775e-07,
      "loss": 0.3635,
      "step": 28310
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.612973928451538,
      "learning_rate": 6.257231015972343e-07,
      "loss": 0.4768,
      "step": 28311
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5246037244796753,
      "learning_rate": 6.253685600291304e-07,
      "loss": 0.3683,
      "step": 28312
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5487123727798462,
      "learning_rate": 6.250141156909462e-07,
      "loss": 0.4226,
      "step": 28313
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.597536563873291,
      "learning_rate": 6.246597685863564e-07,
      "loss": 0.3746,
      "step": 28314
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.649715542793274,
      "learning_rate": 6.243055187190383e-07,
      "loss": 0.4359,
      "step": 28315
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5294618606567383,
      "learning_rate": 6.239513660926644e-07,
      "loss": 0.3757,
      "step": 28316
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5266426801681519,
      "learning_rate": 6.235973107109061e-07,
      "loss": 0.3998,
      "step": 28317
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.534661054611206,
      "learning_rate": 6.232433525774384e-07,
      "loss": 0.3597,
      "step": 28318
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5007679462432861,
      "learning_rate": 6.228894916959316e-07,
      "loss": 0.4459,
      "step": 28319
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8314354419708252,
      "learning_rate": 6.22535728070055e-07,
      "loss": 0.3775,
      "step": 28320
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5011240243911743,
      "learning_rate": 6.221820617034791e-07,
      "loss": 0.4196,
      "step": 28321
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4415788650512695,
      "learning_rate": 6.218284925998696e-07,
      "loss": 0.3595,
      "step": 28322
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5406450033187866,
      "learning_rate": 6.214750207628972e-07,
      "loss": 0.4184,
      "step": 28323
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4563184976577759,
      "learning_rate": 6.211216461962244e-07,
      "loss": 0.3589,
      "step": 28324
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.519566297531128,
      "learning_rate": 6.207683689035171e-07,
      "loss": 0.3964,
      "step": 28325
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.499587893486023,
      "learning_rate": 6.204151888884413e-07,
      "loss": 0.3778,
      "step": 28326
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5935163497924805,
      "learning_rate": 6.200621061546597e-07,
      "loss": 0.481,
      "step": 28327
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.608128309249878,
      "learning_rate": 6.197091207058304e-07,
      "loss": 0.3767,
      "step": 28328
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5997697114944458,
      "learning_rate": 6.193562325456182e-07,
      "loss": 0.4282,
      "step": 28329
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.728493332862854,
      "learning_rate": 6.190034416776825e-07,
      "loss": 0.4199,
      "step": 28330
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5310866832733154,
      "learning_rate": 6.186507481056847e-07,
      "loss": 0.4308,
      "step": 28331
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.505198359489441,
      "learning_rate": 6.182981518332776e-07,
      "loss": 0.3837,
      "step": 28332
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5590616464614868,
      "learning_rate": 6.179456528641203e-07,
      "loss": 0.41,
      "step": 28333
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.552578091621399,
      "learning_rate": 6.17593251201869e-07,
      "loss": 0.366,
      "step": 28334
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8328503370285034,
      "learning_rate": 6.172409468501817e-07,
      "loss": 0.4486,
      "step": 28335
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5419793128967285,
      "learning_rate": 6.168887398127088e-07,
      "loss": 0.3674,
      "step": 28336
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.520311951637268,
      "learning_rate": 6.16536630093102e-07,
      "loss": 0.3855,
      "step": 28337
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6209162473678589,
      "learning_rate": 6.16184617695016e-07,
      "loss": 0.401,
      "step": 28338
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7398232221603394,
      "learning_rate": 6.158327026221022e-07,
      "loss": 0.4347,
      "step": 28339
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5361742973327637,
      "learning_rate": 6.154808848780069e-07,
      "loss": 0.3966,
      "step": 28340
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.556365966796875,
      "learning_rate": 6.151291644663848e-07,
      "loss": 0.3983,
      "step": 28341
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6977753639221191,
      "learning_rate": 6.147775413908774e-07,
      "loss": 0.4045,
      "step": 28342
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.065759658813477,
      "learning_rate": 6.144260156551373e-07,
      "loss": 0.4027,
      "step": 28343
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5595706701278687,
      "learning_rate": 6.140745872628051e-07,
      "loss": 0.3692,
      "step": 28344
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7433445453643799,
      "learning_rate": 6.137232562175288e-07,
      "loss": 0.4611,
      "step": 28345
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.505275011062622,
      "learning_rate": 6.133720225229533e-07,
      "loss": 0.3578,
      "step": 28346
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5386650562286377,
      "learning_rate": 6.130208861827203e-07,
      "loss": 0.391,
      "step": 28347
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.494231939315796,
      "learning_rate": 6.126698472004688e-07,
      "loss": 0.3503,
      "step": 28348
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.581040859222412,
      "learning_rate": 6.123189055798429e-07,
      "loss": 0.4108,
      "step": 28349
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4751225709915161,
      "learning_rate": 6.119680613244816e-07,
      "loss": 0.3724,
      "step": 28350
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5311224460601807,
      "learning_rate": 6.116173144380244e-07,
      "loss": 0.4213,
      "step": 28351
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5693124532699585,
      "learning_rate": 6.112666649241094e-07,
      "loss": 0.3807,
      "step": 28352
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.511141061782837,
      "learning_rate": 6.109161127863705e-07,
      "loss": 0.3874,
      "step": 28353
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.287024736404419,
      "learning_rate": 6.105656580284447e-07,
      "loss": 0.3856,
      "step": 28354
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5196245908737183,
      "learning_rate": 6.102153006539701e-07,
      "loss": 0.4223,
      "step": 28355
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5467511415481567,
      "learning_rate": 6.098650406665763e-07,
      "loss": 0.3931,
      "step": 28356
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5871315002441406,
      "learning_rate": 6.095148780698978e-07,
      "loss": 0.404,
      "step": 28357
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.56111741065979,
      "learning_rate": 6.091648128675665e-07,
      "loss": 0.3771,
      "step": 28358
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5033570528030396,
      "learning_rate": 6.088148450632125e-07,
      "loss": 0.4295,
      "step": 28359
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6228888034820557,
      "learning_rate": 6.084649746604653e-07,
      "loss": 0.4072,
      "step": 28360
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6725821495056152,
      "learning_rate": 6.081152016629543e-07,
      "loss": 0.4031,
      "step": 28361
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5439587831497192,
      "learning_rate": 6.077655260743087e-07,
      "loss": 0.3919,
      "step": 28362
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5023536682128906,
      "learning_rate": 6.074159478981534e-07,
      "loss": 0.4258,
      "step": 28363
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.547073483467102,
      "learning_rate": 6.070664671381121e-07,
      "loss": 0.3763,
      "step": 28364
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.481308937072754,
      "learning_rate": 6.06717083797812e-07,
      "loss": 0.4239,
      "step": 28365
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5166475772857666,
      "learning_rate": 6.063677978808769e-07,
      "loss": 0.3678,
      "step": 28366
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5345971584320068,
      "learning_rate": 6.060186093909293e-07,
      "loss": 0.4088,
      "step": 28367
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4721450805664062,
      "learning_rate": 6.056695183315908e-07,
      "loss": 0.3668,
      "step": 28368
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5259277820587158,
      "learning_rate": 6.053205247064808e-07,
      "loss": 0.4346,
      "step": 28369
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5850731134414673,
      "learning_rate": 6.049716285192198e-07,
      "loss": 0.3743,
      "step": 28370
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5932626724243164,
      "learning_rate": 6.046228297734269e-07,
      "loss": 0.4431,
      "step": 28371
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4549601078033447,
      "learning_rate": 6.042741284727204e-07,
      "loss": 0.3608,
      "step": 28372
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5585682392120361,
      "learning_rate": 6.039255246207132e-07,
      "loss": 0.3888,
      "step": 28373
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5903358459472656,
      "learning_rate": 6.035770182210243e-07,
      "loss": 0.3665,
      "step": 28374
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5047653913497925,
      "learning_rate": 6.032286092772687e-07,
      "loss": 0.4394,
      "step": 28375
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5413719415664673,
      "learning_rate": 6.02880297793057e-07,
      "loss": 0.3724,
      "step": 28376
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5323501825332642,
      "learning_rate": 6.025320837720039e-07,
      "loss": 0.4237,
      "step": 28377
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5010297298431396,
      "learning_rate": 6.021839672177199e-07,
      "loss": 0.3684,
      "step": 28378
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4858800172805786,
      "learning_rate": 6.018359481338176e-07,
      "loss": 0.3786,
      "step": 28379
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4672439098358154,
      "learning_rate": 6.014880265239032e-07,
      "loss": 0.3548,
      "step": 28380
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5069223642349243,
      "learning_rate": 6.011402023915869e-07,
      "loss": 0.4048,
      "step": 28381
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4419991970062256,
      "learning_rate": 6.007924757404782e-07,
      "loss": 0.3639,
      "step": 28382
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.531050443649292,
      "learning_rate": 6.004448465741808e-07,
      "loss": 0.3856,
      "step": 28383
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.602186679840088,
      "learning_rate": 6.000973148962996e-07,
      "loss": 0.4132,
      "step": 28384
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5814063549041748,
      "learning_rate": 5.997498807104407e-07,
      "loss": 0.4354,
      "step": 28385
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6061725616455078,
      "learning_rate": 5.994025440202078e-07,
      "loss": 0.3891,
      "step": 28386
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4620468616485596,
      "learning_rate": 5.990553048292036e-07,
      "loss": 0.4133,
      "step": 28387
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5922375917434692,
      "learning_rate": 5.987081631410297e-07,
      "loss": 0.4001,
      "step": 28388
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5161540508270264,
      "learning_rate": 5.983611189592831e-07,
      "loss": 0.3963,
      "step": 28389
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.515729546546936,
      "learning_rate": 5.980141722875665e-07,
      "loss": 0.3675,
      "step": 28390
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4674524068832397,
      "learning_rate": 5.976673231294783e-07,
      "loss": 0.3835,
      "step": 28391
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4488416910171509,
      "learning_rate": 5.973205714886132e-07,
      "loss": 0.3535,
      "step": 28392
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5046415328979492,
      "learning_rate": 5.969739173685718e-07,
      "loss": 0.4122,
      "step": 28393
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6262037754058838,
      "learning_rate": 5.966273607729456e-07,
      "loss": 0.3928,
      "step": 28394
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5477770566940308,
      "learning_rate": 5.962809017053306e-07,
      "loss": 0.4097,
      "step": 28395
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4385290145874023,
      "learning_rate": 5.959345401693195e-07,
      "loss": 0.3502,
      "step": 28396
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5787442922592163,
      "learning_rate": 5.95588276168505e-07,
      "loss": 0.3975,
      "step": 28397
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4849159717559814,
      "learning_rate": 5.952421097064809e-07,
      "loss": 0.3774,
      "step": 28398
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5424998998641968,
      "learning_rate": 5.948960407868343e-07,
      "loss": 0.3997,
      "step": 28399
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4740419387817383,
      "learning_rate": 5.945500694131534e-07,
      "loss": 0.3687,
      "step": 28400
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.571094274520874,
      "learning_rate": 5.942041955890288e-07,
      "loss": 0.4227,
      "step": 28401
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4956748485565186,
      "learning_rate": 5.938584193180463e-07,
      "loss": 0.3721,
      "step": 28402
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6177301406860352,
      "learning_rate": 5.935127406037955e-07,
      "loss": 0.4271,
      "step": 28403
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4916467666625977,
      "learning_rate": 5.9316715944986e-07,
      "loss": 0.3854,
      "step": 28404
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4952812194824219,
      "learning_rate": 5.928216758598204e-07,
      "loss": 0.3868,
      "step": 28405
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5446497201919556,
      "learning_rate": 5.924762898372638e-07,
      "loss": 0.3623,
      "step": 28406
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.569158911705017,
      "learning_rate": 5.921310013857728e-07,
      "loss": 0.4667,
      "step": 28407
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5465549230575562,
      "learning_rate": 5.917858105089259e-07,
      "loss": 0.3933,
      "step": 28408
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.556382417678833,
      "learning_rate": 5.914407172103065e-07,
      "loss": 0.4007,
      "step": 28409
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5079177618026733,
      "learning_rate": 5.910957214934898e-07,
      "loss": 0.3723,
      "step": 28410
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.638957142829895,
      "learning_rate": 5.907508233620574e-07,
      "loss": 0.429,
      "step": 28411
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.834970474243164,
      "learning_rate": 5.90406022819584e-07,
      "loss": 0.3736,
      "step": 28412
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8817709684371948,
      "learning_rate": 5.900613198696481e-07,
      "loss": 0.4338,
      "step": 28413
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.230379819869995,
      "learning_rate": 5.897167145158222e-07,
      "loss": 0.3584,
      "step": 28414
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4641705751419067,
      "learning_rate": 5.893722067616836e-07,
      "loss": 0.379,
      "step": 28415
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4861520528793335,
      "learning_rate": 5.890277966108015e-07,
      "loss": 0.3817,
      "step": 28416
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.714510202407837,
      "learning_rate": 5.886834840667511e-07,
      "loss": 0.5063,
      "step": 28417
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4859594106674194,
      "learning_rate": 5.883392691331024e-07,
      "loss": 0.3683,
      "step": 28418
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5216562747955322,
      "learning_rate": 5.879951518134264e-07,
      "loss": 0.3845,
      "step": 28419
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5302249193191528,
      "learning_rate": 5.876511321112898e-07,
      "loss": 0.3579,
      "step": 28420
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5650092363357544,
      "learning_rate": 5.873072100302613e-07,
      "loss": 0.3971,
      "step": 28421
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5219385623931885,
      "learning_rate": 5.869633855739098e-07,
      "loss": 0.3816,
      "step": 28422
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.790025234222412,
      "learning_rate": 5.866196587458006e-07,
      "loss": 0.46,
      "step": 28423
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5018776655197144,
      "learning_rate": 5.862760295494996e-07,
      "loss": 0.3494,
      "step": 28424
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5167127847671509,
      "learning_rate": 5.859324979885672e-07,
      "loss": 0.409,
      "step": 28425
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5398603677749634,
      "learning_rate": 5.855890640665684e-07,
      "loss": 0.3837,
      "step": 28426
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.493476152420044,
      "learning_rate": 5.852457277870671e-07,
      "loss": 0.3715,
      "step": 28427
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.457137942314148,
      "learning_rate": 5.849024891536214e-07,
      "loss": 0.3679,
      "step": 28428
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6611207723617554,
      "learning_rate": 5.845593481697931e-07,
      "loss": 0.4525,
      "step": 28429
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.462477445602417,
      "learning_rate": 5.842163048391391e-07,
      "loss": 0.3558,
      "step": 28430
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5961885452270508,
      "learning_rate": 5.83873359165219e-07,
      "loss": 0.4471,
      "step": 28431
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6023019552230835,
      "learning_rate": 5.83530511151591e-07,
      "loss": 0.3929,
      "step": 28432
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5505305528640747,
      "learning_rate": 5.831877608018077e-07,
      "loss": 0.3954,
      "step": 28433
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.489831566810608,
      "learning_rate": 5.828451081194265e-07,
      "loss": 0.3505,
      "step": 28434
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5536143779754639,
      "learning_rate": 5.825025531079997e-07,
      "loss": 0.4118,
      "step": 28435
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.540766716003418,
      "learning_rate": 5.821600957710816e-07,
      "loss": 0.3829,
      "step": 28436
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5087664127349854,
      "learning_rate": 5.818177361122224e-07,
      "loss": 0.4234,
      "step": 28437
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5440853834152222,
      "learning_rate": 5.814754741349726e-07,
      "loss": 0.3861,
      "step": 28438
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8923090696334839,
      "learning_rate": 5.811333098428851e-07,
      "loss": 0.966,
      "step": 28439
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4674103260040283,
      "learning_rate": 5.807912432395068e-07,
      "loss": 0.9514,
      "step": 28440
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.498072862625122,
      "learning_rate": 5.804492743283841e-07,
      "loss": 0.395,
      "step": 28441
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5744640827178955,
      "learning_rate": 5.80107403113065e-07,
      "loss": 0.3539,
      "step": 28442
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5746046304702759,
      "learning_rate": 5.797656295970955e-07,
      "loss": 0.4488,
      "step": 28443
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7078118324279785,
      "learning_rate": 5.794239537840218e-07,
      "loss": 0.4424,
      "step": 28444
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5964059829711914,
      "learning_rate": 5.790823756773856e-07,
      "loss": 0.4643,
      "step": 28445
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5104999542236328,
      "learning_rate": 5.787408952807283e-07,
      "loss": 0.3867,
      "step": 28446
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5992960929870605,
      "learning_rate": 5.783995125975927e-07,
      "loss": 0.3763,
      "step": 28447
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4659090042114258,
      "learning_rate": 5.780582276315228e-07,
      "loss": 0.3438,
      "step": 28448
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5397368669509888,
      "learning_rate": 5.777170403860544e-07,
      "loss": 0.4266,
      "step": 28449
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4870326519012451,
      "learning_rate": 5.773759508647258e-07,
      "loss": 0.3619,
      "step": 28450
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5310947895050049,
      "learning_rate": 5.770349590710777e-07,
      "loss": 0.407,
      "step": 28451
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.520932912826538,
      "learning_rate": 5.76694065008645e-07,
      "loss": 0.3899,
      "step": 28452
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7339075803756714,
      "learning_rate": 5.763532686809625e-07,
      "loss": 0.434,
      "step": 28453
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.617817759513855,
      "learning_rate": 5.760125700915675e-07,
      "loss": 0.3864,
      "step": 28454
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5524462461471558,
      "learning_rate": 5.756719692439905e-07,
      "loss": 0.4102,
      "step": 28455
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5062839984893799,
      "learning_rate": 5.753314661417663e-07,
      "loss": 0.3755,
      "step": 28456
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.735711932182312,
      "learning_rate": 5.749910607884246e-07,
      "loss": 0.4399,
      "step": 28457
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4836066961288452,
      "learning_rate": 5.746507531874967e-07,
      "loss": 0.3588,
      "step": 28458
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5234521627426147,
      "learning_rate": 5.743105433425145e-07,
      "loss": 0.3814,
      "step": 28459
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5592045783996582,
      "learning_rate": 5.739704312570038e-07,
      "loss": 0.3725,
      "step": 28460
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7329603433609009,
      "learning_rate": 5.736304169344908e-07,
      "loss": 0.4105,
      "step": 28461
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7018083333969116,
      "learning_rate": 5.73290500378505e-07,
      "loss": 0.3723,
      "step": 28462
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5751149654388428,
      "learning_rate": 5.7295068159257e-07,
      "loss": 0.3916,
      "step": 28463
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5335044860839844,
      "learning_rate": 5.726109605802133e-07,
      "loss": 0.4076,
      "step": 28464
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.594321608543396,
      "learning_rate": 5.722713373449551e-07,
      "loss": 0.3924,
      "step": 28465
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5660449266433716,
      "learning_rate": 5.719318118903172e-07,
      "loss": 0.3561,
      "step": 28466
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5721778869628906,
      "learning_rate": 5.715923842198234e-07,
      "loss": 0.4083,
      "step": 28467
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4695281982421875,
      "learning_rate": 5.712530543369954e-07,
      "loss": 0.3717,
      "step": 28468
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5132369995117188,
      "learning_rate": 5.70913822245348e-07,
      "loss": 0.4275,
      "step": 28469
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5016509294509888,
      "learning_rate": 5.705746879484042e-07,
      "loss": 0.3903,
      "step": 28470
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8112461566925049,
      "learning_rate": 5.702356514496776e-07,
      "loss": 0.9856,
      "step": 28471
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4410338401794434,
      "learning_rate": 5.698967127526877e-07,
      "loss": 0.9455,
      "step": 28472
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5691173076629639,
      "learning_rate": 5.695578718609473e-07,
      "loss": 0.4223,
      "step": 28473
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5161476135253906,
      "learning_rate": 5.692191287779714e-07,
      "loss": 0.3851,
      "step": 28474
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6035220623016357,
      "learning_rate": 5.688804835072748e-07,
      "loss": 0.4954,
      "step": 28475
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4712575674057007,
      "learning_rate": 5.685419360523692e-07,
      "loss": 0.371,
      "step": 28476
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6186109781265259,
      "learning_rate": 5.68203486416764e-07,
      "loss": 0.4105,
      "step": 28477
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5930842161178589,
      "learning_rate": 5.67865134603971e-07,
      "loss": 0.3973,
      "step": 28478
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.620902419090271,
      "learning_rate": 5.675268806174983e-07,
      "loss": 0.4604,
      "step": 28479
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4951157569885254,
      "learning_rate": 5.671887244608576e-07,
      "loss": 0.3607,
      "step": 28480
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.717842698097229,
      "learning_rate": 5.668506661375539e-07,
      "loss": 0.4584,
      "step": 28481
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.506742000579834,
      "learning_rate": 5.665127056510911e-07,
      "loss": 0.3575,
      "step": 28482
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4784364700317383,
      "learning_rate": 5.661748430049763e-07,
      "loss": 0.3828,
      "step": 28483
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.520192265510559,
      "learning_rate": 5.658370782027156e-07,
      "loss": 0.379,
      "step": 28484
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5242677927017212,
      "learning_rate": 5.654994112478085e-07,
      "loss": 0.4005,
      "step": 28485
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5822193622589111,
      "learning_rate": 5.65161842143761e-07,
      "loss": 0.4167,
      "step": 28486
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5233362913131714,
      "learning_rate": 5.648243708940715e-07,
      "loss": 0.4111,
      "step": 28487
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.50174880027771,
      "learning_rate": 5.644869975022405e-07,
      "loss": 0.3705,
      "step": 28488
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7487890720367432,
      "learning_rate": 5.641497219717673e-07,
      "loss": 0.4106,
      "step": 28489
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5618412494659424,
      "learning_rate": 5.638125443061515e-07,
      "loss": 0.3668,
      "step": 28490
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5378700494766235,
      "learning_rate": 5.634754645088869e-07,
      "loss": 0.4238,
      "step": 28491
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5576527118682861,
      "learning_rate": 5.63138482583474e-07,
      "loss": 0.3764,
      "step": 28492
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5370008945465088,
      "learning_rate": 5.628015985334034e-07,
      "loss": 0.3815,
      "step": 28493
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4630900621414185,
      "learning_rate": 5.624648123621701e-07,
      "loss": 0.3608,
      "step": 28494
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6534886360168457,
      "learning_rate": 5.621281240732713e-07,
      "loss": 0.4306,
      "step": 28495
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6091912984848022,
      "learning_rate": 5.61791533670194e-07,
      "loss": 0.3657,
      "step": 28496
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5564610958099365,
      "learning_rate": 5.614550411564302e-07,
      "loss": 0.4155,
      "step": 28497
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4759492874145508,
      "learning_rate": 5.611186465354712e-07,
      "loss": 0.3671,
      "step": 28498
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7469133138656616,
      "learning_rate": 5.607823498108045e-07,
      "loss": 0.54,
      "step": 28499
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4775384664535522,
      "learning_rate": 5.604461509859205e-07,
      "loss": 0.3626,
      "step": 28500
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6441768407821655,
      "learning_rate": 5.601100500643053e-07,
      "loss": 0.4764,
      "step": 28501
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5562430620193481,
      "learning_rate": 5.597740470494417e-07,
      "loss": 0.3755,
      "step": 28502
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5630964040756226,
      "learning_rate": 5.59438141944818e-07,
      "loss": 0.4095,
      "step": 28503
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5877149105072021,
      "learning_rate": 5.591023347539171e-07,
      "loss": 0.4057,
      "step": 28504
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5717778205871582,
      "learning_rate": 5.587666254802215e-07,
      "loss": 0.4447,
      "step": 28505
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5052034854888916,
      "learning_rate": 5.584310141272143e-07,
      "loss": 0.3831,
      "step": 28506
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5195978879928589,
      "learning_rate": 5.580955006983735e-07,
      "loss": 0.4059,
      "step": 28507
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5162293910980225,
      "learning_rate": 5.577600851971821e-07,
      "loss": 0.3519,
      "step": 28508
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5481865406036377,
      "learning_rate": 5.574247676271161e-07,
      "loss": 0.447,
      "step": 28509
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5295495986938477,
      "learning_rate": 5.57089547991655e-07,
      "loss": 0.3887,
      "step": 28510
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6659252643585205,
      "learning_rate": 5.56754426294277e-07,
      "loss": 0.9062,
      "step": 28511
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3987805843353271,
      "learning_rate": 5.564194025384561e-07,
      "loss": 1.0148,
      "step": 28512
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5052595138549805,
      "learning_rate": 5.560844767276652e-07,
      "loss": 0.4461,
      "step": 28513
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4701467752456665,
      "learning_rate": 5.557496488653791e-07,
      "loss": 0.3611,
      "step": 28514
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5689499378204346,
      "learning_rate": 5.554149189550729e-07,
      "loss": 0.4349,
      "step": 28515
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4955875873565674,
      "learning_rate": 5.55080287000217e-07,
      "loss": 0.3704,
      "step": 28516
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6842824220657349,
      "learning_rate": 5.547457530042821e-07,
      "loss": 0.4665,
      "step": 28517
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.552426815032959,
      "learning_rate": 5.544113169707354e-07,
      "loss": 0.3744,
      "step": 28518
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.615276575088501,
      "learning_rate": 5.540769789030487e-07,
      "loss": 0.4307,
      "step": 28519
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4952337741851807,
      "learning_rate": 5.537427388046889e-07,
      "loss": 0.3611,
      "step": 28520
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5231947898864746,
      "learning_rate": 5.534085966791214e-07,
      "loss": 0.4545,
      "step": 28521
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.580788254737854,
      "learning_rate": 5.530745525298142e-07,
      "loss": 0.3683,
      "step": 28522
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5743651390075684,
      "learning_rate": 5.527406063602292e-07,
      "loss": 0.4311,
      "step": 28523
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6595702171325684,
      "learning_rate": 5.524067581738312e-07,
      "loss": 0.4054,
      "step": 28524
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5004608631134033,
      "learning_rate": 5.520730079740822e-07,
      "loss": 0.4485,
      "step": 28525
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5792163610458374,
      "learning_rate": 5.517393557644435e-07,
      "loss": 0.3891,
      "step": 28526
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6935389041900635,
      "learning_rate": 5.514058015483781e-07,
      "loss": 0.4744,
      "step": 28527
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5383156538009644,
      "learning_rate": 5.510723453293432e-07,
      "loss": 0.3697,
      "step": 28528
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5297642946243286,
      "learning_rate": 5.507389871107971e-07,
      "loss": 0.4299,
      "step": 28529
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6537436246871948,
      "learning_rate": 5.50405726896197e-07,
      "loss": 0.3987,
      "step": 28530
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5282975435256958,
      "learning_rate": 5.500725646890015e-07,
      "loss": 0.4299,
      "step": 28531
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4669560194015503,
      "learning_rate": 5.497395004926653e-07,
      "loss": 0.3465,
      "step": 28532
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5376384258270264,
      "learning_rate": 5.494065343106403e-07,
      "loss": 0.4097,
      "step": 28533
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5259290933609009,
      "learning_rate": 5.490736661463825e-07,
      "loss": 0.4037,
      "step": 28534
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6451157331466675,
      "learning_rate": 5.487408960033424e-07,
      "loss": 0.4014,
      "step": 28535
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.501002311706543,
      "learning_rate": 5.484082238849752e-07,
      "loss": 0.3618,
      "step": 28536
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.586828589439392,
      "learning_rate": 5.480756497947282e-07,
      "loss": 0.434,
      "step": 28537
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5351893901824951,
      "learning_rate": 5.477431737360495e-07,
      "loss": 0.369,
      "step": 28538
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5157233476638794,
      "learning_rate": 5.474107957123886e-07,
      "loss": 0.4301,
      "step": 28539
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7178572416305542,
      "learning_rate": 5.470785157271952e-07,
      "loss": 0.4089,
      "step": 28540
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5877103805541992,
      "learning_rate": 5.467463337839119e-07,
      "loss": 0.4557,
      "step": 28541
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5034534931182861,
      "learning_rate": 5.46414249885987e-07,
      "loss": 0.3651,
      "step": 28542
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6260952949523926,
      "learning_rate": 5.460822640368623e-07,
      "loss": 0.4622,
      "step": 28543
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4582557678222656,
      "learning_rate": 5.45750376239983e-07,
      "loss": 0.3701,
      "step": 28544
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4912333488464355,
      "learning_rate": 5.454185864987893e-07,
      "loss": 0.4093,
      "step": 28545
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4807045459747314,
      "learning_rate": 5.450868948167243e-07,
      "loss": 0.3621,
      "step": 28546
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6471173763275146,
      "learning_rate": 5.447553011972273e-07,
      "loss": 0.4285,
      "step": 28547
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4728597402572632,
      "learning_rate": 5.44423805643739e-07,
      "loss": 0.3741,
      "step": 28548
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.624056100845337,
      "learning_rate": 5.440924081596932e-07,
      "loss": 0.4058,
      "step": 28549
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5401548147201538,
      "learning_rate": 5.437611087485306e-07,
      "loss": 0.3643,
      "step": 28550
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.563071370124817,
      "learning_rate": 5.434299074136873e-07,
      "loss": 0.4904,
      "step": 28551
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.596160650253296,
      "learning_rate": 5.430988041585983e-07,
      "loss": 0.3645,
      "step": 28552
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5709868669509888,
      "learning_rate": 5.427677989866964e-07,
      "loss": 0.4096,
      "step": 28553
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5281169414520264,
      "learning_rate": 5.424368919014156e-07,
      "loss": 0.3867,
      "step": 28554
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5156112909317017,
      "learning_rate": 5.421060829061863e-07,
      "loss": 0.3876,
      "step": 28555
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5021260976791382,
      "learning_rate": 5.417753720044428e-07,
      "loss": 0.3548,
      "step": 28556
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7561570405960083,
      "learning_rate": 5.414447591996108e-07,
      "loss": 0.996,
      "step": 28557
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4881190061569214,
      "learning_rate": 5.411142444951245e-07,
      "loss": 0.99,
      "step": 28558
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5389997959136963,
      "learning_rate": 5.407838278944066e-07,
      "loss": 0.4182,
      "step": 28559
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5248626470565796,
      "learning_rate": 5.404535094008889e-07,
      "loss": 0.3735,
      "step": 28560
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6313401460647583,
      "learning_rate": 5.40123289017992e-07,
      "loss": 0.4024,
      "step": 28561
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8054921627044678,
      "learning_rate": 5.397931667491451e-07,
      "loss": 0.4183,
      "step": 28562
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7194085121154785,
      "learning_rate": 5.394631425977715e-07,
      "loss": 0.4219,
      "step": 28563
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4682613611221313,
      "learning_rate": 5.391332165672924e-07,
      "loss": 0.3703,
      "step": 28564
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5947140455245972,
      "learning_rate": 5.388033886611311e-07,
      "loss": 0.408,
      "step": 28565
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.537891149520874,
      "learning_rate": 5.384736588827067e-07,
      "loss": 0.3907,
      "step": 28566
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.84699285030365,
      "learning_rate": 5.38144027235441e-07,
      "loss": 0.4215,
      "step": 28567
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6055371761322021,
      "learning_rate": 5.378144937227525e-07,
      "loss": 0.3805,
      "step": 28568
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4831820726394653,
      "learning_rate": 5.374850583480573e-07,
      "loss": 0.3573,
      "step": 28569
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5576590299606323,
      "learning_rate": 5.371557211147737e-07,
      "loss": 0.3807,
      "step": 28570
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.492758870124817,
      "learning_rate": 5.368264820263169e-07,
      "loss": 0.3963,
      "step": 28571
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.556121587753296,
      "learning_rate": 5.36497341086103e-07,
      "loss": 0.3582,
      "step": 28572
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5500521659851074,
      "learning_rate": 5.361682982975447e-07,
      "loss": 0.4394,
      "step": 28573
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5133014917373657,
      "learning_rate": 5.358393536640539e-07,
      "loss": 0.3801,
      "step": 28574
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5939205884933472,
      "learning_rate": 5.355105071890421e-07,
      "loss": 0.4264,
      "step": 28575
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5358284711837769,
      "learning_rate": 5.351817588759223e-07,
      "loss": 0.3829,
      "step": 28576
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5078012943267822,
      "learning_rate": 5.348531087281017e-07,
      "loss": 0.365,
      "step": 28577
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5540565252304077,
      "learning_rate": 5.34524556748991e-07,
      "loss": 0.3747,
      "step": 28578
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5309505462646484,
      "learning_rate": 5.341961029419951e-07,
      "loss": 0.4149,
      "step": 28579
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.524787187576294,
      "learning_rate": 5.338677473105236e-07,
      "loss": 0.3889,
      "step": 28580
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5294936895370483,
      "learning_rate": 5.335394898579793e-07,
      "loss": 0.3979,
      "step": 28581
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5108113288879395,
      "learning_rate": 5.332113305877684e-07,
      "loss": 0.3626,
      "step": 28582
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5414258241653442,
      "learning_rate": 5.32883269503296e-07,
      "loss": 0.3967,
      "step": 28583
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5146628618240356,
      "learning_rate": 5.325553066079614e-07,
      "loss": 0.3739,
      "step": 28584
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4917798042297363,
      "learning_rate": 5.322274419051677e-07,
      "loss": 0.4151,
      "step": 28585
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5505542755126953,
      "learning_rate": 5.318996753983141e-07,
      "loss": 0.3643,
      "step": 28586
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5491337776184082,
      "learning_rate": 5.315720070908015e-07,
      "loss": 0.4451,
      "step": 28587
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.500335454940796,
      "learning_rate": 5.312444369860292e-07,
      "loss": 0.3746,
      "step": 28588
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.534967064857483,
      "learning_rate": 5.309169650873936e-07,
      "loss": 0.4249,
      "step": 28589
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4985995292663574,
      "learning_rate": 5.305895913982895e-07,
      "loss": 0.357,
      "step": 28590
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5678714513778687,
      "learning_rate": 5.302623159221132e-07,
      "loss": 0.4352,
      "step": 28591
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6562211513519287,
      "learning_rate": 5.299351386622608e-07,
      "loss": 0.3613,
      "step": 28592
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5215349197387695,
      "learning_rate": 5.296080596221243e-07,
      "loss": 0.4193,
      "step": 28593
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4910056591033936,
      "learning_rate": 5.292810788050962e-07,
      "loss": 0.3575,
      "step": 28594
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6780986785888672,
      "learning_rate": 5.289541962145672e-07,
      "loss": 0.504,
      "step": 28595
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5525580644607544,
      "learning_rate": 5.286274118539292e-07,
      "loss": 0.3638,
      "step": 28596
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4885244369506836,
      "learning_rate": 5.283007257265693e-07,
      "loss": 0.3859,
      "step": 28597
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8136004209518433,
      "learning_rate": 5.279741378358772e-07,
      "loss": 0.3795,
      "step": 28598
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4717365503311157,
      "learning_rate": 5.276476481852399e-07,
      "loss": 0.3863,
      "step": 28599
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5477657318115234,
      "learning_rate": 5.27321256778045e-07,
      "loss": 0.3743,
      "step": 28600
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5802948474884033,
      "learning_rate": 5.269949636176741e-07,
      "loss": 0.9458,
      "step": 28601
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3679157495498657,
      "learning_rate": 5.266687687075134e-07,
      "loss": 1.0002,
      "step": 28602
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5255568027496338,
      "learning_rate": 5.263426720509469e-07,
      "loss": 0.4441,
      "step": 28603
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.602861762046814,
      "learning_rate": 5.260166736513583e-07,
      "loss": 0.3543,
      "step": 28604
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4976431131362915,
      "learning_rate": 5.256907735121242e-07,
      "loss": 0.3812,
      "step": 28605
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5254141092300415,
      "learning_rate": 5.25364971636626e-07,
      "loss": 0.3509,
      "step": 28606
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7995171546936035,
      "learning_rate": 5.250392680282445e-07,
      "loss": 1.0258,
      "step": 28607
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4402611255645752,
      "learning_rate": 5.247136626903582e-07,
      "loss": 0.95,
      "step": 28608
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5357780456542969,
      "learning_rate": 5.243881556263431e-07,
      "loss": 0.4017,
      "step": 28609
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.816389560699463,
      "learning_rate": 5.240627468395732e-07,
      "loss": 0.3586,
      "step": 28610
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5196561813354492,
      "learning_rate": 5.237374363334258e-07,
      "loss": 0.4298,
      "step": 28611
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5301862955093384,
      "learning_rate": 5.23412224111276e-07,
      "loss": 0.3633,
      "step": 28612
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5546716451644897,
      "learning_rate": 5.230871101764945e-07,
      "loss": 0.3987,
      "step": 28613
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4640470743179321,
      "learning_rate": 5.22762094532454e-07,
      "loss": 0.3603,
      "step": 28614
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.566564917564392,
      "learning_rate": 5.224371771825254e-07,
      "loss": 0.4253,
      "step": 28615
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4794975519180298,
      "learning_rate": 5.221123581300792e-07,
      "loss": 0.3691,
      "step": 28616
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5503175258636475,
      "learning_rate": 5.217876373784825e-07,
      "loss": 0.3809,
      "step": 28617
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5319854021072388,
      "learning_rate": 5.214630149311039e-07,
      "loss": 0.3776,
      "step": 28618
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5956056118011475,
      "learning_rate": 5.21138490791313e-07,
      "loss": 0.3873,
      "step": 28619
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0417940616607666,
      "learning_rate": 5.208140649624737e-07,
      "loss": 0.3648,
      "step": 28620
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5979794263839722,
      "learning_rate": 5.204897374479478e-07,
      "loss": 0.3833,
      "step": 28621
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5138568878173828,
      "learning_rate": 5.201655082511025e-07,
      "loss": 0.3486,
      "step": 28622
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.625933289527893,
      "learning_rate": 5.198413773753009e-07,
      "loss": 0.3757,
      "step": 28623
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.495194673538208,
      "learning_rate": 5.195173448239044e-07,
      "loss": 0.3723,
      "step": 28624
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6186527013778687,
      "learning_rate": 5.191934106002727e-07,
      "loss": 0.4376,
      "step": 28625
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4826067686080933,
      "learning_rate": 5.188695747077643e-07,
      "loss": 0.3727,
      "step": 28626
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.890832543373108,
      "learning_rate": 5.185458371497398e-07,
      "loss": 0.9305,
      "step": 28627
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5324866771697998,
      "learning_rate": 5.182221979295588e-07,
      "loss": 1.0772,
      "step": 28628
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5337696075439453,
      "learning_rate": 5.17898657050574e-07,
      "loss": 0.436,
      "step": 28629
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5210405588150024,
      "learning_rate": 5.175752145161439e-07,
      "loss": 0.3878,
      "step": 28630
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5740383863449097,
      "learning_rate": 5.172518703296203e-07,
      "loss": 0.4248,
      "step": 28631
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5940428972244263,
      "learning_rate": 5.169286244943606e-07,
      "loss": 0.3749,
      "step": 28632
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4984291791915894,
      "learning_rate": 5.16605477013713e-07,
      "loss": 0.3977,
      "step": 28633
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5511677265167236,
      "learning_rate": 5.162824278910317e-07,
      "loss": 0.3611,
      "step": 28634
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.586217999458313,
      "learning_rate": 5.159594771296683e-07,
      "loss": 0.4336,
      "step": 28635
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5490576028823853,
      "learning_rate": 5.156366247329713e-07,
      "loss": 0.4007,
      "step": 28636
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.536013126373291,
      "learning_rate": 5.15313870704287e-07,
      "loss": 0.3824,
      "step": 28637
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5510499477386475,
      "learning_rate": 5.149912150469638e-07,
      "loss": 0.403,
      "step": 28638
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5929843187332153,
      "learning_rate": 5.146686577643501e-07,
      "loss": 0.4829,
      "step": 28639
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.443742275238037,
      "learning_rate": 5.14346198859793e-07,
      "loss": 0.3602,
      "step": 28640
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5652120113372803,
      "learning_rate": 5.140238383366314e-07,
      "loss": 0.4374,
      "step": 28641
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5357121229171753,
      "learning_rate": 5.137015761982112e-07,
      "loss": 0.3947,
      "step": 28642
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.489817500114441,
      "learning_rate": 5.133794124478742e-07,
      "loss": 0.4396,
      "step": 28643
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5695257186889648,
      "learning_rate": 5.130573470889654e-07,
      "loss": 0.3824,
      "step": 28644
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8574390411376953,
      "learning_rate": 5.127353801248214e-07,
      "loss": 0.4073,
      "step": 28645
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5398342609405518,
      "learning_rate": 5.124135115587813e-07,
      "loss": 0.3578,
      "step": 28646
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9373118877410889,
      "learning_rate": 5.12091741394184e-07,
      "loss": 0.3751,
      "step": 28647
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4879536628723145,
      "learning_rate": 5.117700696343697e-07,
      "loss": 0.3519,
      "step": 28648
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.526565432548523,
      "learning_rate": 5.114484962826693e-07,
      "loss": 0.3951,
      "step": 28649
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4626238346099854,
      "learning_rate": 5.111270213424235e-07,
      "loss": 0.3472,
      "step": 28650
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5066285133361816,
      "learning_rate": 5.108056448169629e-07,
      "loss": 0.416,
      "step": 28651
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4394478797912598,
      "learning_rate": 5.104843667096227e-07,
      "loss": 0.3557,
      "step": 28652
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5325134992599487,
      "learning_rate": 5.101631870237323e-07,
      "loss": 0.4555,
      "step": 28653
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5067094564437866,
      "learning_rate": 5.098421057626246e-07,
      "loss": 0.3874,
      "step": 28654
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.556785225868225,
      "learning_rate": 5.095211229296315e-07,
      "loss": 0.4473,
      "step": 28655
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4466062784194946,
      "learning_rate": 5.092002385280803e-07,
      "loss": 0.3483,
      "step": 28656
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.538182020187378,
      "learning_rate": 5.088794525612984e-07,
      "loss": 0.4223,
      "step": 28657
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5154926776885986,
      "learning_rate": 5.085587650326129e-07,
      "loss": 0.3935,
      "step": 28658
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4979907274246216,
      "learning_rate": 5.082381759453514e-07,
      "loss": 0.4388,
      "step": 28659
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5287296772003174,
      "learning_rate": 5.079176853028389e-07,
      "loss": 0.3539,
      "step": 28660
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5421829223632812,
      "learning_rate": 5.075972931083983e-07,
      "loss": 0.4188,
      "step": 28661
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.431548833847046,
      "learning_rate": 5.072769993653526e-07,
      "loss": 0.3273,
      "step": 28662
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.516906976699829,
      "learning_rate": 5.069568040770234e-07,
      "loss": 0.4427,
      "step": 28663
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5237253904342651,
      "learning_rate": 5.066367072467337e-07,
      "loss": 0.3601,
      "step": 28664
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.568611979484558,
      "learning_rate": 5.063167088777998e-07,
      "loss": 0.4215,
      "step": 28665
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4939807653427124,
      "learning_rate": 5.059968089735456e-07,
      "loss": 0.3743,
      "step": 28666
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.573936104774475,
      "learning_rate": 5.056770075372841e-07,
      "loss": 0.4442,
      "step": 28667
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.514643907546997,
      "learning_rate": 5.053573045723359e-07,
      "loss": 0.3905,
      "step": 28668
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4958046674728394,
      "learning_rate": 5.050377000820128e-07,
      "loss": 0.4489,
      "step": 28669
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5965023040771484,
      "learning_rate": 5.047181940696333e-07,
      "loss": 0.4335,
      "step": 28670
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5597167015075684,
      "learning_rate": 5.043987865385102e-07,
      "loss": 0.4314,
      "step": 28671
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5617107152938843,
      "learning_rate": 5.040794774919555e-07,
      "loss": 0.3683,
      "step": 28672
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6266366243362427,
      "learning_rate": 5.03760266933282e-07,
      "loss": 0.395,
      "step": 28673
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4754096269607544,
      "learning_rate": 5.03441154865798e-07,
      "loss": 0.3526,
      "step": 28674
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6129964590072632,
      "learning_rate": 5.031221412928155e-07,
      "loss": 0.4246,
      "step": 28675
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5394169092178345,
      "learning_rate": 5.028032262176453e-07,
      "loss": 0.3783,
      "step": 28676
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9859241247177124,
      "learning_rate": 5.02484409643591e-07,
      "loss": 0.4635,
      "step": 28677
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5034615993499756,
      "learning_rate": 5.021656915739593e-07,
      "loss": 0.3527,
      "step": 28678
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4837087392807007,
      "learning_rate": 5.018470720120572e-07,
      "loss": 0.442,
      "step": 28679
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5374457836151123,
      "learning_rate": 5.0152855096119e-07,
      "loss": 0.3898,
      "step": 28680
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5645942687988281,
      "learning_rate": 5.012101284246629e-07,
      "loss": 0.4331,
      "step": 28681
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5981135368347168,
      "learning_rate": 5.008918044057731e-07,
      "loss": 0.3832,
      "step": 28682
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5029326677322388,
      "learning_rate": 5.005735789078248e-07,
      "loss": 0.4268,
      "step": 28683
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5303072929382324,
      "learning_rate": 5.002554519341197e-07,
      "loss": 0.3703,
      "step": 28684
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.561941146850586,
      "learning_rate": 4.999374234879572e-07,
      "loss": 0.4116,
      "step": 28685
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5166033506393433,
      "learning_rate": 4.996194935726361e-07,
      "loss": 0.3823,
      "step": 28686
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.756986141204834,
      "learning_rate": 4.993016621914503e-07,
      "loss": 0.4658,
      "step": 28687
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5302762985229492,
      "learning_rate": 4.989839293476984e-07,
      "loss": 0.3657,
      "step": 28688
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.50015389919281,
      "learning_rate": 4.986662950446786e-07,
      "loss": 0.3828,
      "step": 28689
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.526191234588623,
      "learning_rate": 4.983487592856806e-07,
      "loss": 0.3908,
      "step": 28690
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8152132034301758,
      "learning_rate": 4.980313220740008e-07,
      "loss": 0.401,
      "step": 28691
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4857960939407349,
      "learning_rate": 4.977139834129296e-07,
      "loss": 0.3597,
      "step": 28692
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8453004360198975,
      "learning_rate": 4.973967433057614e-07,
      "loss": 0.4071,
      "step": 28693
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5188194513320923,
      "learning_rate": 4.970796017557822e-07,
      "loss": 0.3453,
      "step": 28694
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.502461552619934,
      "learning_rate": 4.967625587662839e-07,
      "loss": 0.4519,
      "step": 28695
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5566431283950806,
      "learning_rate": 4.964456143405549e-07,
      "loss": 0.3778,
      "step": 28696
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.655069351196289,
      "learning_rate": 4.961287684818827e-07,
      "loss": 0.4017,
      "step": 28697
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5330753326416016,
      "learning_rate": 4.958120211935513e-07,
      "loss": 0.3842,
      "step": 28698
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5797680616378784,
      "learning_rate": 4.954953724788469e-07,
      "loss": 0.4358,
      "step": 28699
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4580307006835938,
      "learning_rate": 4.951788223410537e-07,
      "loss": 0.351,
      "step": 28700
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.649510145187378,
      "learning_rate": 4.948623707834566e-07,
      "loss": 0.3939,
      "step": 28701
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5140334367752075,
      "learning_rate": 4.945460178093364e-07,
      "loss": 0.3351,
      "step": 28702
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6451311111450195,
      "learning_rate": 4.942297634219728e-07,
      "loss": 0.44,
      "step": 28703
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.578253149986267,
      "learning_rate": 4.939136076246464e-07,
      "loss": 0.3718,
      "step": 28704
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8477497100830078,
      "learning_rate": 4.935975504206391e-07,
      "loss": 0.4834,
      "step": 28705
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.540111780166626,
      "learning_rate": 4.93281591813225e-07,
      "loss": 0.3811,
      "step": 28706
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5535578727722168,
      "learning_rate": 4.929657318056835e-07,
      "loss": 0.4473,
      "step": 28707
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4917843341827393,
      "learning_rate": 4.926499704012888e-07,
      "loss": 0.3567,
      "step": 28708
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5692285299301147,
      "learning_rate": 4.923343076033193e-07,
      "loss": 0.4349,
      "step": 28709
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6150662899017334,
      "learning_rate": 4.920187434150436e-07,
      "loss": 0.379,
      "step": 28710
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.637967586517334,
      "learning_rate": 4.917032778397391e-07,
      "loss": 0.4468,
      "step": 28711
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4923765659332275,
      "learning_rate": 4.913879108806763e-07,
      "loss": 0.3775,
      "step": 28712
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5640078783035278,
      "learning_rate": 4.910726425411272e-07,
      "loss": 0.4264,
      "step": 28713
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6000854969024658,
      "learning_rate": 4.90757472824358e-07,
      "loss": 0.3901,
      "step": 28714
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6027647256851196,
      "learning_rate": 4.904424017336396e-07,
      "loss": 0.4199,
      "step": 28715
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.513944149017334,
      "learning_rate": 4.901274292722403e-07,
      "loss": 0.3845,
      "step": 28716
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6000162363052368,
      "learning_rate": 4.898125554434297e-07,
      "loss": 0.408,
      "step": 28717
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5263314247131348,
      "learning_rate": 4.894977802504675e-07,
      "loss": 0.3914,
      "step": 28718
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5084760189056396,
      "learning_rate": 4.891831036966222e-07,
      "loss": 0.4021,
      "step": 28719
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.581976056098938,
      "learning_rate": 4.888685257851556e-07,
      "loss": 0.3647,
      "step": 28720
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5038292407989502,
      "learning_rate": 4.885540465193328e-07,
      "loss": 0.3773,
      "step": 28721
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4968525171279907,
      "learning_rate": 4.882396659024146e-07,
      "loss": 0.4214,
      "step": 28722
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6098413467407227,
      "learning_rate": 4.879253839376596e-07,
      "loss": 0.4015,
      "step": 28723
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4598884582519531,
      "learning_rate": 4.876112006283295e-07,
      "loss": 0.3405,
      "step": 28724
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5564576387405396,
      "learning_rate": 4.872971159776829e-07,
      "loss": 0.3877,
      "step": 28725
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5520646572113037,
      "learning_rate": 4.86983129988976e-07,
      "loss": 0.3792,
      "step": 28726
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5124354362487793,
      "learning_rate": 4.866692426654674e-07,
      "loss": 0.4055,
      "step": 28727
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5188889503479004,
      "learning_rate": 4.863554540104098e-07,
      "loss": 0.3515,
      "step": 28728
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6437064409255981,
      "learning_rate": 4.860417640270598e-07,
      "loss": 0.4452,
      "step": 28729
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5341038703918457,
      "learning_rate": 4.857281727186703e-07,
      "loss": 0.3716,
      "step": 28730
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6058216094970703,
      "learning_rate": 4.854146800884929e-07,
      "loss": 0.4425,
      "step": 28731
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7973183393478394,
      "learning_rate": 4.85101286139782e-07,
      "loss": 0.3904,
      "step": 28732
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5635257959365845,
      "learning_rate": 4.847879908757847e-07,
      "loss": 0.4,
      "step": 28733
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5203975439071655,
      "learning_rate": 4.844747942997508e-07,
      "loss": 0.3591,
      "step": 28734
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5856854915618896,
      "learning_rate": 4.841616964149298e-07,
      "loss": 0.4195,
      "step": 28735
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4941200017929077,
      "learning_rate": 4.838486972245682e-07,
      "loss": 0.3522,
      "step": 28736
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4868887662887573,
      "learning_rate": 4.835357967319143e-07,
      "loss": 0.3847,
      "step": 28737
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4537237882614136,
      "learning_rate": 4.832229949402112e-07,
      "loss": 0.3457,
      "step": 28738
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6598964929580688,
      "learning_rate": 4.829102918527029e-07,
      "loss": 0.4631,
      "step": 28739
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5218358039855957,
      "learning_rate": 4.825976874726335e-07,
      "loss": 0.3786,
      "step": 28740
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5222673416137695,
      "learning_rate": 4.82285181803246e-07,
      "loss": 0.4091,
      "step": 28741
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5199118852615356,
      "learning_rate": 4.819727748477798e-07,
      "loss": 0.3925,
      "step": 28742
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6913120746612549,
      "learning_rate": 4.816604666094771e-07,
      "loss": 0.4093,
      "step": 28743
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.546385645866394,
      "learning_rate": 4.81348257091575e-07,
      "loss": 0.3788,
      "step": 28744
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9171574115753174,
      "learning_rate": 4.810361462973134e-07,
      "loss": 0.4601,
      "step": 28745
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5028891563415527,
      "learning_rate": 4.807241342299285e-07,
      "loss": 0.3688,
      "step": 28746
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6070055961608887,
      "learning_rate": 4.804122208926554e-07,
      "loss": 0.4237,
      "step": 28747
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4516732692718506,
      "learning_rate": 4.801004062887316e-07,
      "loss": 0.3383,
      "step": 28748
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8567529916763306,
      "learning_rate": 4.797886904213888e-07,
      "loss": 0.9909,
      "step": 28749
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.386355996131897,
      "learning_rate": 4.794770732938603e-07,
      "loss": 0.9991,
      "step": 28750
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.655588984489441,
      "learning_rate": 4.791655549093788e-07,
      "loss": 0.4788,
      "step": 28751
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5315905809402466,
      "learning_rate": 4.788541352711751e-07,
      "loss": 0.3752,
      "step": 28752
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5235838890075684,
      "learning_rate": 4.78542814382481e-07,
      "loss": 0.3873,
      "step": 28753
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5613744258880615,
      "learning_rate": 4.782315922465231e-07,
      "loss": 0.3752,
      "step": 28754
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5735574960708618,
      "learning_rate": 4.779204688665273e-07,
      "loss": 0.4225,
      "step": 28755
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5947128534317017,
      "learning_rate": 4.776094442457236e-07,
      "loss": 0.3724,
      "step": 28756
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.678688883781433,
      "learning_rate": 4.77298518387339e-07,
      "loss": 0.4548,
      "step": 28757
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5136733055114746,
      "learning_rate": 4.769876912945948e-07,
      "loss": 0.3604,
      "step": 28758
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5655053853988647,
      "learning_rate": 4.7667696297071685e-07,
      "loss": 0.401,
      "step": 28759
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.514917254447937,
      "learning_rate": 4.7636633341892615e-07,
      "loss": 0.3682,
      "step": 28760
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.2584803104400635,
      "learning_rate": 4.7605580264244665e-07,
      "loss": 0.4215,
      "step": 28761
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5352414846420288,
      "learning_rate": 4.757453706444981e-07,
      "loss": 0.364,
      "step": 28762
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5394937992095947,
      "learning_rate": 4.754350374283001e-07,
      "loss": 0.4092,
      "step": 28763
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4686397314071655,
      "learning_rate": 4.751248029970701e-07,
      "loss": 0.3628,
      "step": 28764
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5282727479934692,
      "learning_rate": 4.748146673540288e-07,
      "loss": 0.408,
      "step": 28765
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5138798952102661,
      "learning_rate": 4.7450463050238925e-07,
      "loss": 0.3768,
      "step": 28766
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5658313035964966,
      "learning_rate": 4.741946924453689e-07,
      "loss": 0.3899,
      "step": 28767
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4732627868652344,
      "learning_rate": 4.738848531861839e-07,
      "loss": 0.3673,
      "step": 28768
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6027432680130005,
      "learning_rate": 4.7357511272804524e-07,
      "loss": 0.4295,
      "step": 28769
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5311464071273804,
      "learning_rate": 4.7326547107416577e-07,
      "loss": 0.3768,
      "step": 28770
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8319034576416016,
      "learning_rate": 4.729559282277563e-07,
      "loss": 1.0257,
      "step": 28771
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.476817011833191,
      "learning_rate": 4.726464841920297e-07,
      "loss": 1.0476,
      "step": 28772
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6105018854141235,
      "learning_rate": 4.723371389701936e-07,
      "loss": 0.4311,
      "step": 28773
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.470963954925537,
      "learning_rate": 4.7202789256545866e-07,
      "loss": 0.3519,
      "step": 28774
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5635658502578735,
      "learning_rate": 4.717187449810279e-07,
      "loss": 0.3686,
      "step": 28775
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.607313632965088,
      "learning_rate": 4.714096962201109e-07,
      "loss": 0.4216,
      "step": 28776
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7156710624694824,
      "learning_rate": 4.7110074628591294e-07,
      "loss": 0.9433,
      "step": 28777
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3875415325164795,
      "learning_rate": 4.7079189518163704e-07,
      "loss": 0.967,
      "step": 28778
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.547947645187378,
      "learning_rate": 4.7048314291048834e-07,
      "loss": 0.3977,
      "step": 28779
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9059935808181763,
      "learning_rate": 4.7017448947566656e-07,
      "loss": 0.3613,
      "step": 28780
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.53232741355896,
      "learning_rate": 4.698659348803758e-07,
      "loss": 0.4111,
      "step": 28781
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5551855564117432,
      "learning_rate": 4.6955747912781345e-07,
      "loss": 0.3864,
      "step": 28782
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5377568006515503,
      "learning_rate": 4.6924912222118035e-07,
      "loss": 0.3956,
      "step": 28783
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4497346878051758,
      "learning_rate": 4.68940864163675e-07,
      "loss": 0.3889,
      "step": 28784
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5896408557891846,
      "learning_rate": 4.686327049584938e-07,
      "loss": 0.4064,
      "step": 28785
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.430711030960083,
      "learning_rate": 4.683246446088319e-07,
      "loss": 0.3622,
      "step": 28786
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5866835117340088,
      "learning_rate": 4.680166831178845e-07,
      "loss": 0.4241,
      "step": 28787
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5045225620269775,
      "learning_rate": 4.67708820488848e-07,
      "loss": 0.3758,
      "step": 28788
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.72658109664917,
      "learning_rate": 4.6740105672491434e-07,
      "loss": 0.4923,
      "step": 28789
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5306477546691895,
      "learning_rate": 4.6709339182927527e-07,
      "loss": 0.364,
      "step": 28790
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5194576978683472,
      "learning_rate": 4.6678582580511944e-07,
      "loss": 0.3933,
      "step": 28791
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.410593032836914,
      "learning_rate": 4.664783586556398e-07,
      "loss": 0.3694,
      "step": 28792
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5078108310699463,
      "learning_rate": 4.6617099038402615e-07,
      "loss": 0.4194,
      "step": 28793
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4777629375457764,
      "learning_rate": 4.6586372099346246e-07,
      "loss": 0.3617,
      "step": 28794
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4492690563201904,
      "learning_rate": 4.6555655048713953e-07,
      "loss": 0.365,
      "step": 28795
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4530065059661865,
      "learning_rate": 4.6524947886824045e-07,
      "loss": 0.3703,
      "step": 28796
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5126148462295532,
      "learning_rate": 4.649425061399526e-07,
      "loss": 0.4267,
      "step": 28797
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.57197105884552,
      "learning_rate": 4.646356323054568e-07,
      "loss": 0.3717,
      "step": 28798
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5757299661636353,
      "learning_rate": 4.643288573679383e-07,
      "loss": 0.4315,
      "step": 28799
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.351685047149658,
      "learning_rate": 4.6402218133057676e-07,
      "loss": 0.3783,
      "step": 28800
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4974085092544556,
      "learning_rate": 4.637156041965551e-07,
      "loss": 0.3974,
      "step": 28801
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5345497131347656,
      "learning_rate": 4.6340912596905095e-07,
      "loss": 0.3769,
      "step": 28802
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4822417497634888,
      "learning_rate": 4.631027466512439e-07,
      "loss": 0.3829,
      "step": 28803
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6104416847229004,
      "learning_rate": 4.6279646624631356e-07,
      "loss": 0.3715,
      "step": 28804
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5855668783187866,
      "learning_rate": 4.624902847574331e-07,
      "loss": 0.4309,
      "step": 28805
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5487231016159058,
      "learning_rate": 4.621842021877798e-07,
      "loss": 0.3864,
      "step": 28806
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5608587265014648,
      "learning_rate": 4.618782185405268e-07,
      "loss": 0.4361,
      "step": 28807
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4840198755264282,
      "learning_rate": 4.6157233381884937e-07,
      "loss": 0.3699,
      "step": 28808
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4902546405792236,
      "learning_rate": 4.612665480259215e-07,
      "loss": 0.4098,
      "step": 28809
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.914410948753357,
      "learning_rate": 4.609608611649119e-07,
      "loss": 0.4073,
      "step": 28810
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4979349374771118,
      "learning_rate": 4.606552732389913e-07,
      "loss": 0.4129,
      "step": 28811
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.464820146560669,
      "learning_rate": 4.6034978425132937e-07,
      "loss": 0.3669,
      "step": 28812
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8331352472305298,
      "learning_rate": 4.6004439420509583e-07,
      "loss": 0.9117,
      "step": 28813
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4720429182052612,
      "learning_rate": 4.597391031034559e-07,
      "loss": 1.0081,
      "step": 28814
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5355794429779053,
      "learning_rate": 4.5943391094957934e-07,
      "loss": 0.4355,
      "step": 28815
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4962882995605469,
      "learning_rate": 4.591288177466269e-07,
      "loss": 0.3693,
      "step": 28816
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5248589515686035,
      "learning_rate": 4.588238234977671e-07,
      "loss": 0.4224,
      "step": 28817
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5760829448699951,
      "learning_rate": 4.5851892820615975e-07,
      "loss": 0.369,
      "step": 28818
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.55392324924469,
      "learning_rate": 4.5821413187496886e-07,
      "loss": 0.4517,
      "step": 28819
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.511927604675293,
      "learning_rate": 4.579094345073565e-07,
      "loss": 0.377,
      "step": 28820
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.494401454925537,
      "learning_rate": 4.576048361064822e-07,
      "loss": 0.41,
      "step": 28821
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.550382137298584,
      "learning_rate": 4.5730033667550353e-07,
      "loss": 0.3902,
      "step": 28822
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5382314920425415,
      "learning_rate": 4.569959362175791e-07,
      "loss": 0.41,
      "step": 28823
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5124784708023071,
      "learning_rate": 4.5669163473586744e-07,
      "loss": 0.3587,
      "step": 28824
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4961130619049072,
      "learning_rate": 4.56387432233526e-07,
      "loss": 0.4133,
      "step": 28825
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5387005805969238,
      "learning_rate": 4.560833287137067e-07,
      "loss": 0.3595,
      "step": 28826
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5778732299804688,
      "learning_rate": 4.5577932417956495e-07,
      "loss": 0.4338,
      "step": 28827
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5062298774719238,
      "learning_rate": 4.554754186342525e-07,
      "loss": 0.3399,
      "step": 28828
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6401307582855225,
      "learning_rate": 4.551716120809235e-07,
      "loss": 0.433,
      "step": 28829
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.424484133720398,
      "learning_rate": 4.5486790452272776e-07,
      "loss": 0.3517,
      "step": 28830
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5212723016738892,
      "learning_rate": 4.5456429596281603e-07,
      "loss": 0.4266,
      "step": 28831
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5611368417739868,
      "learning_rate": 4.542607864043347e-07,
      "loss": 0.3922,
      "step": 28832
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6431090831756592,
      "learning_rate": 4.5395737585043456e-07,
      "loss": 0.4543,
      "step": 28833
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4903144836425781,
      "learning_rate": 4.5365406430426087e-07,
      "loss": 0.3691,
      "step": 28834
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5103883743286133,
      "learning_rate": 4.533508517689611e-07,
      "loss": 0.4149,
      "step": 28835
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5525364875793457,
      "learning_rate": 4.5304773824767725e-07,
      "loss": 0.4011,
      "step": 28836
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5917450189590454,
      "learning_rate": 4.527447237435556e-07,
      "loss": 0.4105,
      "step": 28837
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5449236631393433,
      "learning_rate": 4.5244180825973703e-07,
      "loss": 0.3913,
      "step": 28838
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4807848930358887,
      "learning_rate": 4.521389917993646e-07,
      "loss": 0.3701,
      "step": 28839
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5159618854522705,
      "learning_rate": 4.518362743655802e-07,
      "loss": 0.3617,
      "step": 28840
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6013308763504028,
      "learning_rate": 4.515336559615213e-07,
      "loss": 0.4143,
      "step": 28841
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.440369963645935,
      "learning_rate": 4.5123113659032547e-07,
      "loss": 0.3579,
      "step": 28842
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5617356300354004,
      "learning_rate": 4.5092871625513345e-07,
      "loss": 0.4331,
      "step": 28843
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.472123146057129,
      "learning_rate": 4.5062639495907947e-07,
      "loss": 0.3713,
      "step": 28844
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.529000997543335,
      "learning_rate": 4.5032417270530095e-07,
      "loss": 0.42,
      "step": 28845
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.519860029220581,
      "learning_rate": 4.5002204949693207e-07,
      "loss": 0.3581,
      "step": 28846
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5145682096481323,
      "learning_rate": 4.4972002533710367e-07,
      "loss": 0.4087,
      "step": 28847
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5651284456253052,
      "learning_rate": 4.4941810022895105e-07,
      "loss": 0.4282,
      "step": 28848
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.604549527168274,
      "learning_rate": 4.491162741756061e-07,
      "loss": 0.3801,
      "step": 28849
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5647130012512207,
      "learning_rate": 4.4881454718019634e-07,
      "loss": 0.3706,
      "step": 28850
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.601243495941162,
      "learning_rate": 4.4851291924585373e-07,
      "loss": 0.4462,
      "step": 28851
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7081525325775146,
      "learning_rate": 4.482113903757057e-07,
      "loss": 0.3603,
      "step": 28852
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4670108556747437,
      "learning_rate": 4.4790996057287983e-07,
      "loss": 0.3814,
      "step": 28853
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4797760248184204,
      "learning_rate": 4.4760862984050134e-07,
      "loss": 0.3694,
      "step": 28854
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5684784650802612,
      "learning_rate": 4.4730739818169665e-07,
      "loss": 0.4084,
      "step": 28855
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5101618766784668,
      "learning_rate": 4.4700626559959105e-07,
      "loss": 0.3858,
      "step": 28856
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5476152896881104,
      "learning_rate": 4.4670523209730643e-07,
      "loss": 0.4792,
      "step": 28857
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.599939227104187,
      "learning_rate": 4.464042976779637e-07,
      "loss": 0.3848,
      "step": 28858
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5040241479873657,
      "learning_rate": 4.461034623446847e-07,
      "loss": 0.4388,
      "step": 28859
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5347824096679688,
      "learning_rate": 4.458027261005904e-07,
      "loss": 0.4083,
      "step": 28860
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5173956155776978,
      "learning_rate": 4.4550208894880154e-07,
      "loss": 0.3831,
      "step": 28861
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4905483722686768,
      "learning_rate": 4.4520155089243455e-07,
      "loss": 0.3723,
      "step": 28862
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5146132707595825,
      "learning_rate": 4.4490111193460586e-07,
      "loss": 0.4166,
      "step": 28863
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5084471702575684,
      "learning_rate": 4.446007720784307e-07,
      "loss": 0.3929,
      "step": 28864
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7800188064575195,
      "learning_rate": 4.4430053132702655e-07,
      "loss": 0.4108,
      "step": 28865
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5544615983963013,
      "learning_rate": 4.4400038968350544e-07,
      "loss": 0.3795,
      "step": 28866
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6701675653457642,
      "learning_rate": 4.4370034715098264e-07,
      "loss": 0.4186,
      "step": 28867
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4631778001785278,
      "learning_rate": 4.4340040373256676e-07,
      "loss": 0.351,
      "step": 28868
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.588491678237915,
      "learning_rate": 4.4310055943137197e-07,
      "loss": 0.4219,
      "step": 28869
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4956914186477661,
      "learning_rate": 4.428008142505058e-07,
      "loss": 0.3606,
      "step": 28870
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5427720546722412,
      "learning_rate": 4.42501168193078e-07,
      "loss": 0.4494,
      "step": 28871
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6974900960922241,
      "learning_rate": 4.422016212621971e-07,
      "loss": 0.3998,
      "step": 28872
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.560261607170105,
      "learning_rate": 4.419021734609685e-07,
      "loss": 0.4396,
      "step": 28873
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5217700004577637,
      "learning_rate": 4.416028247924975e-07,
      "loss": 0.3871,
      "step": 28874
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7361507415771484,
      "learning_rate": 4.413035752598904e-07,
      "loss": 0.4572,
      "step": 28875
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4908745288848877,
      "learning_rate": 4.410044248662515e-07,
      "loss": 0.3483,
      "step": 28876
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8537241220474243,
      "learning_rate": 4.407053736146827e-07,
      "loss": 0.9412,
      "step": 28877
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3574637174606323,
      "learning_rate": 4.404064215082837e-07,
      "loss": 1.0033,
      "step": 28878
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.575600266456604,
      "learning_rate": 4.401075685501566e-07,
      "loss": 0.4144,
      "step": 28879
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.937320590019226,
      "learning_rate": 4.3980881474340096e-07,
      "loss": 0.4013,
      "step": 28880
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5693621635437012,
      "learning_rate": 4.395101600911167e-07,
      "loss": 0.4311,
      "step": 28881
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.513898253440857,
      "learning_rate": 4.392116045964001e-07,
      "loss": 0.3681,
      "step": 28882
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.62295401096344,
      "learning_rate": 4.3891314826234653e-07,
      "loss": 0.4136,
      "step": 28883
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4961483478546143,
      "learning_rate": 4.386147910920524e-07,
      "loss": 0.3531,
      "step": 28884
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7368803024291992,
      "learning_rate": 4.38316533088613e-07,
      "loss": 0.3939,
      "step": 28885
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4817560911178589,
      "learning_rate": 4.3801837425512027e-07,
      "loss": 0.3698,
      "step": 28886
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.618947148323059,
      "learning_rate": 4.377203145946685e-07,
      "loss": 0.4277,
      "step": 28887
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6693552732467651,
      "learning_rate": 4.374223541103473e-07,
      "loss": 0.4073,
      "step": 28888
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6022851467132568,
      "learning_rate": 4.3712449280524873e-07,
      "loss": 0.4877,
      "step": 28889
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5277221202850342,
      "learning_rate": 4.368267306824592e-07,
      "loss": 0.3686,
      "step": 28890
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5993618965148926,
      "learning_rate": 4.365290677450695e-07,
      "loss": 0.4359,
      "step": 28891
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.445203423500061,
      "learning_rate": 4.362315039961662e-07,
      "loss": 0.3491,
      "step": 28892
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5911979675292969,
      "learning_rate": 4.359340394388367e-07,
      "loss": 0.4135,
      "step": 28893
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5105284452438354,
      "learning_rate": 4.3563667407616304e-07,
      "loss": 0.3946,
      "step": 28894
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.576568603515625,
      "learning_rate": 4.3533940791123164e-07,
      "loss": 0.4558,
      "step": 28895
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5243492126464844,
      "learning_rate": 4.3504224094712443e-07,
      "loss": 0.3759,
      "step": 28896
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6095925569534302,
      "learning_rate": 4.347451731869268e-07,
      "loss": 0.4393,
      "step": 28897
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5584384202957153,
      "learning_rate": 4.344482046337162e-07,
      "loss": 0.3599,
      "step": 28898
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.612876296043396,
      "learning_rate": 4.341513352905735e-07,
      "loss": 0.4724,
      "step": 28899
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5514168739318848,
      "learning_rate": 4.3385456516057746e-07,
      "loss": 0.3784,
      "step": 28900
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6190500259399414,
      "learning_rate": 4.3355789424680773e-07,
      "loss": 0.4294,
      "step": 28901
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4839204549789429,
      "learning_rate": 4.332613225523397e-07,
      "loss": 0.3559,
      "step": 28902
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5742435455322266,
      "learning_rate": 4.329648500802497e-07,
      "loss": 0.4415,
      "step": 28903
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5722113847732544,
      "learning_rate": 4.3266847683361203e-07,
      "loss": 0.3749,
      "step": 28904
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5363365411758423,
      "learning_rate": 4.323722028155031e-07,
      "loss": 0.4035,
      "step": 28905
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4385838508605957,
      "learning_rate": 4.3207602802899154e-07,
      "loss": 0.3601,
      "step": 28906
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5352545976638794,
      "learning_rate": 4.3177995247715267e-07,
      "loss": 0.409,
      "step": 28907
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.55100679397583,
      "learning_rate": 4.3148397616305624e-07,
      "loss": 0.3896,
      "step": 28908
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5256677865982056,
      "learning_rate": 4.311880990897721e-07,
      "loss": 0.456,
      "step": 28909
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.482797622680664,
      "learning_rate": 4.3089232126036773e-07,
      "loss": 0.3572,
      "step": 28910
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5971856117248535,
      "learning_rate": 4.305966426779118e-07,
      "loss": 0.435,
      "step": 28911
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.552394986152649,
      "learning_rate": 4.303010633454707e-07,
      "loss": 0.4144,
      "step": 28912
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6010212898254395,
      "learning_rate": 4.3000558326611095e-07,
      "loss": 0.486,
      "step": 28913
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.472739338874817,
      "learning_rate": 4.2971020244289674e-07,
      "loss": 0.3697,
      "step": 28914
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5509719848632812,
      "learning_rate": 4.2941492087889117e-07,
      "loss": 0.4266,
      "step": 28915
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.746914267539978,
      "learning_rate": 4.2911973857715615e-07,
      "loss": 0.4177,
      "step": 28916
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5236071348190308,
      "learning_rate": 4.2882465554075494e-07,
      "loss": 0.4133,
      "step": 28917
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.589770793914795,
      "learning_rate": 4.285296717727483e-07,
      "loss": 0.3769,
      "step": 28918
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5583001375198364,
      "learning_rate": 4.2823478727619274e-07,
      "loss": 0.4303,
      "step": 28919
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5446821451187134,
      "learning_rate": 4.2794000205414797e-07,
      "loss": 0.3655,
      "step": 28920
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5028767585754395,
      "learning_rate": 4.2764531610967276e-07,
      "loss": 0.3869,
      "step": 28921
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5000362396240234,
      "learning_rate": 4.273507294458235e-07,
      "loss": 0.3969,
      "step": 28922
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6360349655151367,
      "learning_rate": 4.270562420656543e-07,
      "loss": 0.4608,
      "step": 28923
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.514723777770996,
      "learning_rate": 4.267618539722185e-07,
      "loss": 0.3974,
      "step": 28924
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6673882007598877,
      "learning_rate": 4.2646756516857124e-07,
      "loss": 0.9232,
      "step": 28925
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4427794218063354,
      "learning_rate": 4.261733756577646e-07,
      "loss": 1.0702,
      "step": 28926
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5267747640609741,
      "learning_rate": 4.258792854428484e-07,
      "loss": 0.3949,
      "step": 28927
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5780822038650513,
      "learning_rate": 4.255852945268757e-07,
      "loss": 0.3707,
      "step": 28928
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.523817777633667,
      "learning_rate": 4.2529140291289185e-07,
      "loss": 0.388,
      "step": 28929
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5065346956253052,
      "learning_rate": 4.2499761060394886e-07,
      "loss": 0.4159,
      "step": 28930
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6346346139907837,
      "learning_rate": 4.2470391760308985e-07,
      "loss": 0.4205,
      "step": 28931
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7078324556350708,
      "learning_rate": 4.2441032391336233e-07,
      "loss": 0.3679,
      "step": 28932
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5476253032684326,
      "learning_rate": 4.24116829537814e-07,
      "loss": 0.4251,
      "step": 28933
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6150050163269043,
      "learning_rate": 4.238234344794867e-07,
      "loss": 0.3778,
      "step": 28934
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5536335706710815,
      "learning_rate": 4.2353013874142256e-07,
      "loss": 0.4032,
      "step": 28935
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.480076789855957,
      "learning_rate": 4.2323694232666466e-07,
      "loss": 0.394,
      "step": 28936
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8564320802688599,
      "learning_rate": 4.229438452382528e-07,
      "loss": 0.942,
      "step": 28937
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4950363636016846,
      "learning_rate": 4.22650847479229e-07,
      "loss": 0.9811,
      "step": 28938
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5084737539291382,
      "learning_rate": 4.2235794905263193e-07,
      "loss": 0.4247,
      "step": 28939
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5938798189163208,
      "learning_rate": 4.2206514996149584e-07,
      "loss": 0.3904,
      "step": 28940
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6158040761947632,
      "learning_rate": 4.217724502088605e-07,
      "loss": 0.4146,
      "step": 28941
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.554702639579773,
      "learning_rate": 4.2147984979776233e-07,
      "loss": 0.3581,
      "step": 28942
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5515779256820679,
      "learning_rate": 4.211873487312346e-07,
      "loss": 0.3942,
      "step": 28943
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.540447473526001,
      "learning_rate": 4.2089494701231136e-07,
      "loss": 0.3764,
      "step": 28944
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5240553617477417,
      "learning_rate": 4.206026446440248e-07,
      "loss": 0.393,
      "step": 28945
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5726186037063599,
      "learning_rate": 4.2031044162940794e-07,
      "loss": 0.3661,
      "step": 28946
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.682295799255371,
      "learning_rate": 4.200183379714895e-07,
      "loss": 0.4713,
      "step": 28947
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5087774991989136,
      "learning_rate": 4.197263336733004e-07,
      "loss": 0.3568,
      "step": 28948
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.566777229309082,
      "learning_rate": 4.194344287378693e-07,
      "loss": 0.4431,
      "step": 28949
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5272314548492432,
      "learning_rate": 4.1914262316822383e-07,
      "loss": 0.3755,
      "step": 28950
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.554280400276184,
      "learning_rate": 4.1885091696738824e-07,
      "loss": 0.4054,
      "step": 28951
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.532841682434082,
      "learning_rate": 4.1855931013839004e-07,
      "loss": 0.3649,
      "step": 28952
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6682692766189575,
      "learning_rate": 4.182678026842546e-07,
      "loss": 0.498,
      "step": 28953
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3967152833938599,
      "learning_rate": 4.17976394608004e-07,
      "loss": 0.3508,
      "step": 28954
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4984886646270752,
      "learning_rate": 4.176850859126591e-07,
      "loss": 0.4131,
      "step": 28955
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5451910495758057,
      "learning_rate": 4.1739387660124197e-07,
      "loss": 0.3604,
      "step": 28956
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.500565528869629,
      "learning_rate": 4.1710276667677464e-07,
      "loss": 0.4076,
      "step": 28957
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5055983066558838,
      "learning_rate": 4.1681175614227687e-07,
      "loss": 0.3856,
      "step": 28958
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5370804071426392,
      "learning_rate": 4.1652084500076516e-07,
      "loss": 0.3612,
      "step": 28959
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4655691385269165,
      "learning_rate": 4.16230033255256e-07,
      "loss": 0.3759,
      "step": 28960
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.642295241355896,
      "learning_rate": 4.15939320908767e-07,
      "loss": 0.4634,
      "step": 28961
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.526692509651184,
      "learning_rate": 4.1564870796431346e-07,
      "loss": 0.3632,
      "step": 28962
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5396608114242554,
      "learning_rate": 4.1535819442490746e-07,
      "loss": 0.3976,
      "step": 28963
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4738739728927612,
      "learning_rate": 4.1506778029356544e-07,
      "loss": 0.3653,
      "step": 28964
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5886313915252686,
      "learning_rate": 4.147774655732961e-07,
      "loss": 0.4021,
      "step": 28965
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4668899774551392,
      "learning_rate": 4.144872502671138e-07,
      "loss": 0.3801,
      "step": 28966
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.521380066871643,
      "learning_rate": 4.141971343780249e-07,
      "loss": 0.398,
      "step": 28967
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4933784008026123,
      "learning_rate": 4.1390711790904037e-07,
      "loss": 0.3536,
      "step": 28968
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.644821286201477,
      "learning_rate": 4.1361720086317e-07,
      "loss": 0.4408,
      "step": 28969
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.468226671218872,
      "learning_rate": 4.1332738324341704e-07,
      "loss": 0.3758,
      "step": 28970
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.527119755744934,
      "learning_rate": 4.13037665052789e-07,
      "loss": 0.3702,
      "step": 28971
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4711326360702515,
      "learning_rate": 4.127480462942901e-07,
      "loss": 0.3591,
      "step": 28972
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6388598680496216,
      "learning_rate": 4.1245852697092467e-07,
      "loss": 0.4452,
      "step": 28973
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5345628261566162,
      "learning_rate": 4.1216910708569814e-07,
      "loss": 0.3661,
      "step": 28974
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7114918231964111,
      "learning_rate": 4.11879786641608e-07,
      "loss": 0.4547,
      "step": 28975
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.451649785041809,
      "learning_rate": 4.1159056564165634e-07,
      "loss": 0.3445,
      "step": 28976
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.614020586013794,
      "learning_rate": 4.11301444088843e-07,
      "loss": 0.4423,
      "step": 28977
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.619477391242981,
      "learning_rate": 4.1101242198616776e-07,
      "loss": 0.3896,
      "step": 28978
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5372949838638306,
      "learning_rate": 4.1072349933662604e-07,
      "loss": 0.3895,
      "step": 28979
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4744582176208496,
      "learning_rate": 4.104346761432165e-07,
      "loss": 0.3813,
      "step": 28980
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.747529149055481,
      "learning_rate": 4.101459524089335e-07,
      "loss": 0.4412,
      "step": 28981
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5195916891098022,
      "learning_rate": 4.0985732813677347e-07,
      "loss": 0.3528,
      "step": 28982
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5510154962539673,
      "learning_rate": 4.0956880332972624e-07,
      "loss": 0.4073,
      "step": 28983
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5366530418395996,
      "learning_rate": 4.0928037799078615e-07,
      "loss": 0.3771,
      "step": 28984
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5637242794036865,
      "learning_rate": 4.0899205212294735e-07,
      "loss": 0.4407,
      "step": 28985
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4925004243850708,
      "learning_rate": 4.0870382572919643e-07,
      "loss": 0.3378,
      "step": 28986
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.565866470336914,
      "learning_rate": 4.084156988125232e-07,
      "loss": 0.4535,
      "step": 28987
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4922958612442017,
      "learning_rate": 4.0812767137591747e-07,
      "loss": 0.3816,
      "step": 28988
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6685808897018433,
      "learning_rate": 4.078397434223658e-07,
      "loss": 0.4361,
      "step": 28989
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5464098453521729,
      "learning_rate": 4.075519149548568e-07,
      "loss": 0.3795,
      "step": 28990
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6126669645309448,
      "learning_rate": 4.0726418597637154e-07,
      "loss": 0.4124,
      "step": 28991
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.570548176765442,
      "learning_rate": 4.0697655648989533e-07,
      "loss": 0.3597,
      "step": 28992
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5600961446762085,
      "learning_rate": 4.0668902649841357e-07,
      "loss": 0.4082,
      "step": 28993
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5005874633789062,
      "learning_rate": 4.0640159600490724e-07,
      "loss": 0.3727,
      "step": 28994
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5486842393875122,
      "learning_rate": 4.061142650123573e-07,
      "loss": 0.421,
      "step": 28995
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5137712955474854,
      "learning_rate": 4.058270335237435e-07,
      "loss": 0.3842,
      "step": 28996
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5846179723739624,
      "learning_rate": 4.055399015420447e-07,
      "loss": 0.4223,
      "step": 28997
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.523222804069519,
      "learning_rate": 4.052528690702406e-07,
      "loss": 0.3571,
      "step": 28998
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.569810390472412,
      "learning_rate": 4.049659361113056e-07,
      "loss": 0.4111,
      "step": 28999
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5329926013946533,
      "learning_rate": 4.0467910266821844e-07,
      "loss": 0.3641,
      "step": 29000
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5245587825775146,
      "learning_rate": 4.0439236874395216e-07,
      "loss": 0.3866,
      "step": 29001
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.861701011657715,
      "learning_rate": 4.0410573434148114e-07,
      "loss": 0.3704,
      "step": 29002
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4824093580245972,
      "learning_rate": 4.0381919946377746e-07,
      "loss": 0.4037,
      "step": 29003
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4632551670074463,
      "learning_rate": 4.0353276411381426e-07,
      "loss": 0.3723,
      "step": 29004
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.517846941947937,
      "learning_rate": 4.032464282945625e-07,
      "loss": 0.4031,
      "step": 29005
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4700472354888916,
      "learning_rate": 4.0296019200899206e-07,
      "loss": 0.3618,
      "step": 29006
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5287516117095947,
      "learning_rate": 4.0267405526006943e-07,
      "loss": 0.4051,
      "step": 29007
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6214213371276855,
      "learning_rate": 4.023880180507633e-07,
      "loss": 0.3827,
      "step": 29008
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5063055753707886,
      "learning_rate": 4.021020803840403e-07,
      "loss": 0.4156,
      "step": 29009
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5470384359359741,
      "learning_rate": 4.0181624226286906e-07,
      "loss": 0.3804,
      "step": 29010
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5788246393203735,
      "learning_rate": 4.015305036902106e-07,
      "loss": 0.4224,
      "step": 29011
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5656355619430542,
      "learning_rate": 4.012448646690292e-07,
      "loss": 0.3929,
      "step": 29012
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6467878818511963,
      "learning_rate": 4.00959325202287e-07,
      "loss": 0.4449,
      "step": 29013
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4985628128051758,
      "learning_rate": 4.0067388529294704e-07,
      "loss": 0.3798,
      "step": 29014
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5844483375549316,
      "learning_rate": 4.0038854494396926e-07,
      "loss": 0.3977,
      "step": 29015
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6278135776519775,
      "learning_rate": 4.0010330415831243e-07,
      "loss": 0.3848,
      "step": 29016
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5233900547027588,
      "learning_rate": 3.9981816293893527e-07,
      "loss": 0.4246,
      "step": 29017
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.497243046760559,
      "learning_rate": 3.995331212887965e-07,
      "loss": 0.3649,
      "step": 29018
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8088754415512085,
      "learning_rate": 3.9924817921084936e-07,
      "loss": 1.0274,
      "step": 29019
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3943263292312622,
      "learning_rate": 3.989633367080514e-07,
      "loss": 0.9239,
      "step": 29020
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5566775798797607,
      "learning_rate": 3.9867859378335703e-07,
      "loss": 0.4367,
      "step": 29021
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5215258598327637,
      "learning_rate": 3.983939504397194e-07,
      "loss": 0.4048,
      "step": 29022
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5078203678131104,
      "learning_rate": 3.9810940668008945e-07,
      "loss": 0.4305,
      "step": 29023
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4707363843917847,
      "learning_rate": 3.978249625074182e-07,
      "loss": 0.3753,
      "step": 29024
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5190147161483765,
      "learning_rate": 3.975406179246577e-07,
      "loss": 0.4785,
      "step": 29025
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5242314338684082,
      "learning_rate": 3.9725637293475784e-07,
      "loss": 0.3825,
      "step": 29026
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5046006441116333,
      "learning_rate": 3.969722275406629e-07,
      "loss": 0.4428,
      "step": 29027
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4219462871551514,
      "learning_rate": 3.9668818174532166e-07,
      "loss": 0.3565,
      "step": 29028
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5955806970596313,
      "learning_rate": 3.964042355516806e-07,
      "loss": 0.4133,
      "step": 29029
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4897397756576538,
      "learning_rate": 3.9612038896268636e-07,
      "loss": 0.327,
      "step": 29030
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5861798524856567,
      "learning_rate": 3.958366419812809e-07,
      "loss": 0.4379,
      "step": 29031
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5167946815490723,
      "learning_rate": 3.955529946104053e-07,
      "loss": 0.3899,
      "step": 29032
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5546034574508667,
      "learning_rate": 3.9526944685300386e-07,
      "loss": 0.4001,
      "step": 29033
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5693151950836182,
      "learning_rate": 3.949859987120186e-07,
      "loss": 0.4027,
      "step": 29034
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.524017095565796,
      "learning_rate": 3.947026501903861e-07,
      "loss": 0.3948,
      "step": 29035
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5339640378952026,
      "learning_rate": 3.944194012910474e-07,
      "loss": 0.4076,
      "step": 29036
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5852817296981812,
      "learning_rate": 3.9413625201693893e-07,
      "loss": 0.4177,
      "step": 29037
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5741679668426514,
      "learning_rate": 3.9385320237099955e-07,
      "loss": 0.4319,
      "step": 29038
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5092833042144775,
      "learning_rate": 3.9357025235616244e-07,
      "loss": 0.4251,
      "step": 29039
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4585940837860107,
      "learning_rate": 3.93287401975363e-07,
      "loss": 0.3557,
      "step": 29040
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5135029554367065,
      "learning_rate": 3.9300465123153664e-07,
      "loss": 0.4313,
      "step": 29041
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5368295907974243,
      "learning_rate": 3.927220001276133e-07,
      "loss": 0.3571,
      "step": 29042
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6286391019821167,
      "learning_rate": 3.9243944866652503e-07,
      "loss": 0.4706,
      "step": 29043
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6001091003417969,
      "learning_rate": 3.921569968512029e-07,
      "loss": 0.3639,
      "step": 29044
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6028846502304077,
      "learning_rate": 3.918746446845767e-07,
      "loss": 0.4502,
      "step": 29045
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5647269487380981,
      "learning_rate": 3.9159239216957634e-07,
      "loss": 0.4142,
      "step": 29046
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.58522629737854,
      "learning_rate": 3.9131023930912615e-07,
      "loss": 0.4643,
      "step": 29047
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.439037561416626,
      "learning_rate": 3.910281861061538e-07,
      "loss": 0.3301,
      "step": 29048
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8487954139709473,
      "learning_rate": 3.9074623256358355e-07,
      "loss": 0.9525,
      "step": 29049
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.370849370956421,
      "learning_rate": 3.904643786843432e-07,
      "loss": 1.0301,
      "step": 29050
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.514175534248352,
      "learning_rate": 3.9018262447135245e-07,
      "loss": 0.4726,
      "step": 29051
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5573487281799316,
      "learning_rate": 3.899009699275347e-07,
      "loss": 0.4002,
      "step": 29052
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.085977792739868,
      "learning_rate": 3.896194150558108e-07,
      "loss": 0.9323,
      "step": 29053
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4789469242095947,
      "learning_rate": 3.8933795985910293e-07,
      "loss": 1.0605,
      "step": 29054
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5932244062423706,
      "learning_rate": 3.8905660434032765e-07,
      "loss": 0.4038,
      "step": 29055
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5323240756988525,
      "learning_rate": 3.8877534850240264e-07,
      "loss": 0.3889,
      "step": 29056
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7445138692855835,
      "learning_rate": 3.8849419234824883e-07,
      "loss": 0.4265,
      "step": 29057
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5268001556396484,
      "learning_rate": 3.882131358807795e-07,
      "loss": 0.3706,
      "step": 29058
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6328983306884766,
      "learning_rate": 3.879321791029089e-07,
      "loss": 0.4854,
      "step": 29059
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.445763111114502,
      "learning_rate": 3.876513220175515e-07,
      "loss": 0.3536,
      "step": 29060
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5011861324310303,
      "learning_rate": 3.8737056462762154e-07,
      "loss": 0.3841,
      "step": 29061
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4832277297973633,
      "learning_rate": 3.870899069360312e-07,
      "loss": 0.3613,
      "step": 29062
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5078104734420776,
      "learning_rate": 3.8680934894569034e-07,
      "loss": 0.4092,
      "step": 29063
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4848027229309082,
      "learning_rate": 3.865288906595066e-07,
      "loss": 0.3837,
      "step": 29064
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5688947439193726,
      "learning_rate": 3.8624853208039214e-07,
      "loss": 0.42,
      "step": 29065
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5227619409561157,
      "learning_rate": 3.859682732112546e-07,
      "loss": 0.3783,
      "step": 29066
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5475668907165527,
      "learning_rate": 3.856881140549995e-07,
      "loss": 0.3842,
      "step": 29067
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5555390119552612,
      "learning_rate": 3.8540805461453224e-07,
      "loss": 0.3802,
      "step": 29068
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6326029300689697,
      "learning_rate": 3.8512809489275714e-07,
      "loss": 0.434,
      "step": 29069
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.598658800125122,
      "learning_rate": 3.848482348925797e-07,
      "loss": 0.3763,
      "step": 29070
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4789901971817017,
      "learning_rate": 3.8456847461690096e-07,
      "loss": 0.3946,
      "step": 29071
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5739200115203857,
      "learning_rate": 3.8428881406862295e-07,
      "loss": 0.3954,
      "step": 29072
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.600783348083496,
      "learning_rate": 3.8400925325064674e-07,
      "loss": 0.4504,
      "step": 29073
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5384557247161865,
      "learning_rate": 3.837297921658711e-07,
      "loss": 0.3748,
      "step": 29074
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5402019023895264,
      "learning_rate": 3.834504308171949e-07,
      "loss": 0.4221,
      "step": 29075
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5412025451660156,
      "learning_rate": 3.831711692075146e-07,
      "loss": 0.3797,
      "step": 29076
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5249762535095215,
      "learning_rate": 3.8289200733972796e-07,
      "loss": 0.3875,
      "step": 29077
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4989241361618042,
      "learning_rate": 3.826129452167304e-07,
      "loss": 0.3638,
      "step": 29078
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4713037014007568,
      "learning_rate": 3.823339828414141e-07,
      "loss": 0.3822,
      "step": 29079
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4858146905899048,
      "learning_rate": 3.8205512021667447e-07,
      "loss": 0.3991,
      "step": 29080
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.602648138999939,
      "learning_rate": 3.817763573454025e-07,
      "loss": 0.4555,
      "step": 29081
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6073771715164185,
      "learning_rate": 3.8149769423049156e-07,
      "loss": 0.3831,
      "step": 29082
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.477547287940979,
      "learning_rate": 3.8121913087483033e-07,
      "loss": 0.3809,
      "step": 29083
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5185798406600952,
      "learning_rate": 3.8094066728130653e-07,
      "loss": 0.4132,
      "step": 29084
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5797480344772339,
      "learning_rate": 3.8066230345280895e-07,
      "loss": 0.4351,
      "step": 29085
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.679324746131897,
      "learning_rate": 3.8038403939222756e-07,
      "loss": 0.3912,
      "step": 29086
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5352309942245483,
      "learning_rate": 3.8010587510244445e-07,
      "loss": 0.4205,
      "step": 29087
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.487388253211975,
      "learning_rate": 3.798278105863484e-07,
      "loss": 0.3512,
      "step": 29088
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6958688497543335,
      "learning_rate": 3.7954984584681944e-07,
      "loss": 0.4264,
      "step": 29089
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4888538122177124,
      "learning_rate": 3.79271980886744e-07,
      "loss": 0.3566,
      "step": 29090
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4893858432769775,
      "learning_rate": 3.7899421570900096e-07,
      "loss": 0.3752,
      "step": 29091
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6235740184783936,
      "learning_rate": 3.787165503164736e-07,
      "loss": 0.4273,
      "step": 29092
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5045362710952759,
      "learning_rate": 3.784389847120407e-07,
      "loss": 0.4201,
      "step": 29093
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5517897605895996,
      "learning_rate": 3.781615188985821e-07,
      "loss": 0.3815,
      "step": 29094
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5469672679901123,
      "learning_rate": 3.778841528789734e-07,
      "loss": 0.3936,
      "step": 29095
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4898756742477417,
      "learning_rate": 3.776068866560922e-07,
      "loss": 0.3653,
      "step": 29096
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5894922018051147,
      "learning_rate": 3.773297202328141e-07,
      "loss": 0.4263,
      "step": 29097
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4539867639541626,
      "learning_rate": 3.770526536120167e-07,
      "loss": 0.3705,
      "step": 29098
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5853161811828613,
      "learning_rate": 3.7677568679656994e-07,
      "loss": 0.43,
      "step": 29099
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4603281021118164,
      "learning_rate": 3.764988197893471e-07,
      "loss": 0.3831,
      "step": 29100
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5629003047943115,
      "learning_rate": 3.762220525932203e-07,
      "loss": 0.4032,
      "step": 29101
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4874497652053833,
      "learning_rate": 3.759453852110606e-07,
      "loss": 0.3633,
      "step": 29102
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.573789119720459,
      "learning_rate": 3.756688176457379e-07,
      "loss": 0.4328,
      "step": 29103
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5329508781433105,
      "learning_rate": 3.7539234990011775e-07,
      "loss": 0.3694,
      "step": 29104
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5330827236175537,
      "learning_rate": 3.7511598197707e-07,
      "loss": 0.4137,
      "step": 29105
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.538565993309021,
      "learning_rate": 3.748397138794613e-07,
      "loss": 0.3725,
      "step": 29106
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6548011302947998,
      "learning_rate": 3.745635456101548e-07,
      "loss": 0.4115,
      "step": 29107
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5039969682693481,
      "learning_rate": 3.742874771720173e-07,
      "loss": 0.3666,
      "step": 29108
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5358422994613647,
      "learning_rate": 3.7401150856790967e-07,
      "loss": 0.426,
      "step": 29109
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5346282720565796,
      "learning_rate": 3.737356398006975e-07,
      "loss": 0.373,
      "step": 29110
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5391265153884888,
      "learning_rate": 3.734598708732384e-07,
      "loss": 0.3923,
      "step": 29111
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6336541175842285,
      "learning_rate": 3.7318420178839355e-07,
      "loss": 0.395,
      "step": 29112
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5551656484603882,
      "learning_rate": 3.7290863254902387e-07,
      "loss": 0.4145,
      "step": 29113
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5130531787872314,
      "learning_rate": 3.7263316315798605e-07,
      "loss": 0.3707,
      "step": 29114
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5590778589248657,
      "learning_rate": 3.723577936181366e-07,
      "loss": 0.4455,
      "step": 29115
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4101935625076294,
      "learning_rate": 3.7208252393233226e-07,
      "loss": 0.3331,
      "step": 29116
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.52541983127594,
      "learning_rate": 3.7180735410342725e-07,
      "loss": 0.4132,
      "step": 29117
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4347319602966309,
      "learning_rate": 3.7153228413427835e-07,
      "loss": 0.3623,
      "step": 29118
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6949102878570557,
      "learning_rate": 3.7125731402773535e-07,
      "loss": 0.5328,
      "step": 29119
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.478302001953125,
      "learning_rate": 3.709824437866505e-07,
      "loss": 0.3727,
      "step": 29120
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5395584106445312,
      "learning_rate": 3.707076734138759e-07,
      "loss": 0.3863,
      "step": 29121
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4956738948822021,
      "learning_rate": 3.704330029122616e-07,
      "loss": 0.3611,
      "step": 29122
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6014955043792725,
      "learning_rate": 3.701584322846541e-07,
      "loss": 0.4568,
      "step": 29123
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5778312683105469,
      "learning_rate": 3.698839615339034e-07,
      "loss": 0.3785,
      "step": 29124
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5566370487213135,
      "learning_rate": 3.696095906628538e-07,
      "loss": 0.428,
      "step": 29125
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5372143983840942,
      "learning_rate": 3.6933531967435544e-07,
      "loss": 0.3804,
      "step": 29126
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5162458419799805,
      "learning_rate": 3.69061148571247e-07,
      "loss": 0.4103,
      "step": 29127
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5379036664962769,
      "learning_rate": 3.687870773563762e-07,
      "loss": 0.3913,
      "step": 29128
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6786203384399414,
      "learning_rate": 3.6851310603258525e-07,
      "loss": 0.4133,
      "step": 29129
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4878718852996826,
      "learning_rate": 3.682392346027153e-07,
      "loss": 0.3635,
      "step": 29130
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6139063835144043,
      "learning_rate": 3.67965463069605e-07,
      "loss": 0.4414,
      "step": 29131
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.496543526649475,
      "learning_rate": 3.6769179143609557e-07,
      "loss": 0.3605,
      "step": 29132
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5036925077438354,
      "learning_rate": 3.674182197050247e-07,
      "loss": 0.3765,
      "step": 29133
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5670251846313477,
      "learning_rate": 3.6714474787923116e-07,
      "loss": 0.3931,
      "step": 29134
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.47809898853302,
      "learning_rate": 3.6687137596155054e-07,
      "loss": 0.4208,
      "step": 29135
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5217827558517456,
      "learning_rate": 3.665981039548172e-07,
      "loss": 0.3751,
      "step": 29136
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6164826154708862,
      "learning_rate": 3.6632493186186556e-07,
      "loss": 0.431,
      "step": 29137
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5321228504180908,
      "learning_rate": 3.6605185968553e-07,
      "loss": 0.3848,
      "step": 29138
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5847781896591187,
      "learning_rate": 3.6577888742864163e-07,
      "loss": 0.4513,
      "step": 29139
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5342777967453003,
      "learning_rate": 3.655060150940337e-07,
      "loss": 0.3781,
      "step": 29140
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5261735916137695,
      "learning_rate": 3.6523324268453286e-07,
      "loss": 0.4101,
      "step": 29141
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5629572868347168,
      "learning_rate": 3.649605702029713e-07,
      "loss": 0.3688,
      "step": 29142
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6317758560180664,
      "learning_rate": 3.6468799765217556e-07,
      "loss": 0.4088,
      "step": 29143
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.596354603767395,
      "learning_rate": 3.6441552503497346e-07,
      "loss": 0.4121,
      "step": 29144
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.559096097946167,
      "learning_rate": 3.641431523541894e-07,
      "loss": 0.4028,
      "step": 29145
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6308879852294922,
      "learning_rate": 3.6387087961265e-07,
      "loss": 0.4202,
      "step": 29146
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4875773191452026,
      "learning_rate": 3.6359870681317743e-07,
      "loss": 0.3869,
      "step": 29147
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5538257360458374,
      "learning_rate": 3.633266339585961e-07,
      "loss": 0.3607,
      "step": 29148
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5773547887802124,
      "learning_rate": 3.630546610517283e-07,
      "loss": 0.4371,
      "step": 29149
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6248117685317993,
      "learning_rate": 3.627827880953938e-07,
      "loss": 0.3637,
      "step": 29150
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.484061598777771,
      "learning_rate": 3.625110150924116e-07,
      "loss": 0.3774,
      "step": 29151
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4938369989395142,
      "learning_rate": 3.6223934204560165e-07,
      "loss": 0.347,
      "step": 29152
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5779778957366943,
      "learning_rate": 3.619677689577805e-07,
      "loss": 0.4405,
      "step": 29153
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4851202964782715,
      "learning_rate": 3.616962958317671e-07,
      "loss": 0.3951,
      "step": 29154
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6567989587783813,
      "learning_rate": 3.614249226703759e-07,
      "loss": 0.4895,
      "step": 29155
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4944288730621338,
      "learning_rate": 3.6115364947642006e-07,
      "loss": 0.3632,
      "step": 29156
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6270267963409424,
      "learning_rate": 3.60882476252713e-07,
      "loss": 0.4704,
      "step": 29157
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7070865631103516,
      "learning_rate": 3.606114030020691e-07,
      "loss": 0.3588,
      "step": 29158
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.560190200805664,
      "learning_rate": 3.6034042972730053e-07,
      "loss": 0.4097,
      "step": 29159
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4792333841323853,
      "learning_rate": 3.600695564312162e-07,
      "loss": 0.339,
      "step": 29160
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6194113492965698,
      "learning_rate": 3.5979878311662385e-07,
      "loss": 0.4081,
      "step": 29161
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5707262754440308,
      "learning_rate": 3.595281097863346e-07,
      "loss": 0.3793,
      "step": 29162
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.555908203125,
      "learning_rate": 3.59257536443155e-07,
      "loss": 0.4137,
      "step": 29163
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9318389892578125,
      "learning_rate": 3.5898706308989063e-07,
      "loss": 0.3739,
      "step": 29164
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5799610614776611,
      "learning_rate": 3.5871668972934813e-07,
      "loss": 0.3741,
      "step": 29165
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4827964305877686,
      "learning_rate": 3.5844641636432973e-07,
      "loss": 0.3712,
      "step": 29166
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.554178237915039,
      "learning_rate": 3.581762429976399e-07,
      "loss": 0.3968,
      "step": 29167
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.506653070449829,
      "learning_rate": 3.579061696320796e-07,
      "loss": 0.376,
      "step": 29168
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5435049533843994,
      "learning_rate": 3.576361962704511e-07,
      "loss": 0.3934,
      "step": 29169
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.532997488975525,
      "learning_rate": 3.573663229155555e-07,
      "loss": 0.3708,
      "step": 29170
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.595701813697815,
      "learning_rate": 3.5709654957019057e-07,
      "loss": 0.4126,
      "step": 29171
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5028897523880005,
      "learning_rate": 3.5682687623715297e-07,
      "loss": 0.3379,
      "step": 29172
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5214840173721313,
      "learning_rate": 3.565573029192404e-07,
      "loss": 0.4151,
      "step": 29173
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5357874631881714,
      "learning_rate": 3.5628782961924957e-07,
      "loss": 0.3662,
      "step": 29174
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5401229858398438,
      "learning_rate": 3.56018456339976e-07,
      "loss": 0.3964,
      "step": 29175
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4554616212844849,
      "learning_rate": 3.55749183084213e-07,
      "loss": 0.3555,
      "step": 29176
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5711044073104858,
      "learning_rate": 3.5548000985475173e-07,
      "loss": 0.4025,
      "step": 29177
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5273010730743408,
      "learning_rate": 3.552109366543843e-07,
      "loss": 0.3634,
      "step": 29178
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5732024908065796,
      "learning_rate": 3.5494196348590416e-07,
      "loss": 0.4154,
      "step": 29179
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4745279550552368,
      "learning_rate": 3.5467309035209786e-07,
      "loss": 0.3744,
      "step": 29180
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.551746129989624,
      "learning_rate": 3.544043172557565e-07,
      "loss": 0.4324,
      "step": 29181
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5543004274368286,
      "learning_rate": 3.541356441996646e-07,
      "loss": 0.4031,
      "step": 29182
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6490063667297363,
      "learning_rate": 3.538670711866132e-07,
      "loss": 0.4445,
      "step": 29183
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5442290306091309,
      "learning_rate": 3.5359859821938237e-07,
      "loss": 0.358,
      "step": 29184
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4976799488067627,
      "learning_rate": 3.53330225300762e-07,
      "loss": 0.3911,
      "step": 29185
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5226837396621704,
      "learning_rate": 3.5306195243353105e-07,
      "loss": 0.3872,
      "step": 29186
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.573116660118103,
      "learning_rate": 3.52793779620475e-07,
      "loss": 0.4143,
      "step": 29187
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4571324586868286,
      "learning_rate": 3.525257068643728e-07,
      "loss": 0.3623,
      "step": 29188
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6357393264770508,
      "learning_rate": 3.522577341680067e-07,
      "loss": 0.4063,
      "step": 29189
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5223479270935059,
      "learning_rate": 3.519898615341566e-07,
      "loss": 0.3931,
      "step": 29190
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5820724964141846,
      "learning_rate": 3.517220889655992e-07,
      "loss": 0.4326,
      "step": 29191
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4674041271209717,
      "learning_rate": 3.5145441646511126e-07,
      "loss": 0.3557,
      "step": 29192
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5444389581680298,
      "learning_rate": 3.5118684403546934e-07,
      "loss": 0.4693,
      "step": 29193
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4743684530258179,
      "learning_rate": 3.5091937167945014e-07,
      "loss": 0.3725,
      "step": 29194
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5283153057098389,
      "learning_rate": 3.5065199939982695e-07,
      "loss": 0.3981,
      "step": 29195
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5141404867172241,
      "learning_rate": 3.503847271993721e-07,
      "loss": 0.3685,
      "step": 29196
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5813895463943481,
      "learning_rate": 3.5011755508085775e-07,
      "loss": 0.4015,
      "step": 29197
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5590991973876953,
      "learning_rate": 3.4985048304705615e-07,
      "loss": 0.3586,
      "step": 29198
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.524859070777893,
      "learning_rate": 3.4958351110073616e-07,
      "loss": 0.407,
      "step": 29199
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5916898250579834,
      "learning_rate": 3.4931663924466673e-07,
      "loss": 0.3805,
      "step": 29200
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5689846277236938,
      "learning_rate": 3.4904986748161675e-07,
      "loss": 0.4156,
      "step": 29201
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4906266927719116,
      "learning_rate": 3.4878319581435063e-07,
      "loss": 0.3511,
      "step": 29202
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6383004188537598,
      "learning_rate": 3.4851662424563726e-07,
      "loss": 0.4365,
      "step": 29203
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4752802848815918,
      "learning_rate": 3.482501527782389e-07,
      "loss": 0.3681,
      "step": 29204
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6468206644058228,
      "learning_rate": 3.479837814149212e-07,
      "loss": 0.4207,
      "step": 29205
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.482845664024353,
      "learning_rate": 3.4771751015844626e-07,
      "loss": 0.3595,
      "step": 29206
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5644359588623047,
      "learning_rate": 3.4745133901157634e-07,
      "loss": 0.4029,
      "step": 29207
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5092527866363525,
      "learning_rate": 3.471852679770693e-07,
      "loss": 0.3768,
      "step": 29208
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5168883800506592,
      "learning_rate": 3.4691929705768737e-07,
      "loss": 0.3935,
      "step": 29209
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.497775673866272,
      "learning_rate": 3.4665342625618716e-07,
      "loss": 0.3672,
      "step": 29210
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5749025344848633,
      "learning_rate": 3.4638765557532985e-07,
      "loss": 0.4221,
      "step": 29211
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4892303943634033,
      "learning_rate": 3.461219850178688e-07,
      "loss": 0.3552,
      "step": 29212
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.518305778503418,
      "learning_rate": 3.458564145865595e-07,
      "loss": 0.4059,
      "step": 29213
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6359424591064453,
      "learning_rate": 3.455909442841565e-07,
      "loss": 0.3865,
      "step": 29214
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7095803022384644,
      "learning_rate": 3.453255741134143e-07,
      "loss": 0.444,
      "step": 29215
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.58579683303833,
      "learning_rate": 3.450603040770839e-07,
      "loss": 0.3965,
      "step": 29216
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5505985021591187,
      "learning_rate": 3.447951341779177e-07,
      "loss": 0.4313,
      "step": 29217
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4734032154083252,
      "learning_rate": 3.4453006441866556e-07,
      "loss": 0.3849,
      "step": 29218
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.727797031402588,
      "learning_rate": 3.442650948020765e-07,
      "loss": 0.4025,
      "step": 29219
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5716708898544312,
      "learning_rate": 3.4400022533089827e-07,
      "loss": 0.3729,
      "step": 29220
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5144764184951782,
      "learning_rate": 3.4373545600787874e-07,
      "loss": 0.4005,
      "step": 29221
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5464541912078857,
      "learning_rate": 3.434707868357634e-07,
      "loss": 0.3939,
      "step": 29222
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.51694655418396,
      "learning_rate": 3.43206217817299e-07,
      "loss": 0.4233,
      "step": 29223
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4958831071853638,
      "learning_rate": 3.429417489552256e-07,
      "loss": 0.356,
      "step": 29224
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5350489616394043,
      "learning_rate": 3.426773802522898e-07,
      "loss": 0.4071,
      "step": 29225
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5536739826202393,
      "learning_rate": 3.4241311171123283e-07,
      "loss": 0.3625,
      "step": 29226
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5955208539962769,
      "learning_rate": 3.4214894333479576e-07,
      "loss": 0.4163,
      "step": 29227
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.545838713645935,
      "learning_rate": 3.4188487512571646e-07,
      "loss": 0.3816,
      "step": 29228
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6090883016586304,
      "learning_rate": 3.416209070867349e-07,
      "loss": 0.4928,
      "step": 29229
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4801738262176514,
      "learning_rate": 3.4135703922059e-07,
      "loss": 0.3762,
      "step": 29230
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5072498321533203,
      "learning_rate": 3.4109327153001747e-07,
      "loss": 0.3842,
      "step": 29231
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6675211191177368,
      "learning_rate": 3.4082960401775387e-07,
      "loss": 0.3944,
      "step": 29232
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6564455032348633,
      "learning_rate": 3.4056603668653154e-07,
      "loss": 0.3797,
      "step": 29233
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.464184045791626,
      "learning_rate": 3.40302569539086e-07,
      "loss": 0.3619,
      "step": 29234
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5469622611999512,
      "learning_rate": 3.400392025781507e-07,
      "loss": 0.4019,
      "step": 29235
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4859447479248047,
      "learning_rate": 3.397759358064545e-07,
      "loss": 0.3594,
      "step": 29236
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5686203241348267,
      "learning_rate": 3.395127692267297e-07,
      "loss": 0.4065,
      "step": 29237
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.571494460105896,
      "learning_rate": 3.392497028417052e-07,
      "loss": 0.3597,
      "step": 29238
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5579124689102173,
      "learning_rate": 3.3898673665411e-07,
      "loss": 0.3886,
      "step": 29239
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5339298248291016,
      "learning_rate": 3.387238706666707e-07,
      "loss": 0.4079,
      "step": 29240
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5774617195129395,
      "learning_rate": 3.3846110488211404e-07,
      "loss": 0.3875,
      "step": 29241
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5054302215576172,
      "learning_rate": 3.3819843930316564e-07,
      "loss": 0.3498,
      "step": 29242
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5610419511795044,
      "learning_rate": 3.3793587393255e-07,
      "loss": 0.4188,
      "step": 29243
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5891222953796387,
      "learning_rate": 3.3767340877298936e-07,
      "loss": 0.3659,
      "step": 29244
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.527633786201477,
      "learning_rate": 3.374110438272049e-07,
      "loss": 0.4027,
      "step": 29245
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5790348052978516,
      "learning_rate": 3.3714877909791996e-07,
      "loss": 0.3928,
      "step": 29246
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5517584085464478,
      "learning_rate": 3.368866145878547e-07,
      "loss": 0.455,
      "step": 29247
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4941357374191284,
      "learning_rate": 3.3662455029972784e-07,
      "loss": 0.385,
      "step": 29248
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5413833856582642,
      "learning_rate": 3.3636258623625516e-07,
      "loss": 0.3988,
      "step": 29249
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4722412824630737,
      "learning_rate": 3.3610072240015554e-07,
      "loss": 0.3405,
      "step": 29250
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6410092115402222,
      "learning_rate": 3.3583895879414686e-07,
      "loss": 0.4444,
      "step": 29251
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5752882957458496,
      "learning_rate": 3.355772954209402e-07,
      "loss": 0.3829,
      "step": 29252
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6425753831863403,
      "learning_rate": 3.353157322832512e-07,
      "loss": 0.4938,
      "step": 29253
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.465716004371643,
      "learning_rate": 3.350542693837933e-07,
      "loss": 0.356,
      "step": 29254
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4678112268447876,
      "learning_rate": 3.347929067252775e-07,
      "loss": 0.4219,
      "step": 29255
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5389429330825806,
      "learning_rate": 3.3453164431041407e-07,
      "loss": 0.3754,
      "step": 29256
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5259921550750732,
      "learning_rate": 3.3427048214191406e-07,
      "loss": 0.4367,
      "step": 29257
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4786467552185059,
      "learning_rate": 3.340094202224853e-07,
      "loss": 0.3714,
      "step": 29258
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6129287481307983,
      "learning_rate": 3.3374845855483564e-07,
      "loss": 0.4367,
      "step": 29259
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4866809844970703,
      "learning_rate": 3.334875971416718e-07,
      "loss": 0.3635,
      "step": 29260
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5814450979232788,
      "learning_rate": 3.3322683598569826e-07,
      "loss": 0.4201,
      "step": 29261
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.502057671546936,
      "learning_rate": 3.329661750896207e-07,
      "loss": 0.3571,
      "step": 29262
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5580648183822632,
      "learning_rate": 3.327056144561436e-07,
      "loss": 0.4466,
      "step": 29263
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5507482290267944,
      "learning_rate": 3.3244515408796584e-07,
      "loss": 0.3939,
      "step": 29264
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5178217887878418,
      "learning_rate": 3.32184793987792e-07,
      "loss": 0.4105,
      "step": 29265
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5740649700164795,
      "learning_rate": 3.3192453415832105e-07,
      "loss": 0.3579,
      "step": 29266
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.605652093887329,
      "learning_rate": 3.316643746022541e-07,
      "loss": 0.4324,
      "step": 29267
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5150032043457031,
      "learning_rate": 3.314043153222879e-07,
      "loss": 0.3813,
      "step": 29268
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5671683549880981,
      "learning_rate": 3.311443563211181e-07,
      "loss": 0.3995,
      "step": 29269
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4972609281539917,
      "learning_rate": 3.308844976014436e-07,
      "loss": 0.3792,
      "step": 29270
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5456836223602295,
      "learning_rate": 3.306247391659589e-07,
      "loss": 0.4324,
      "step": 29271
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4961169958114624,
      "learning_rate": 3.303650810173564e-07,
      "loss": 0.3662,
      "step": 29272
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5726295709609985,
      "learning_rate": 3.301055231583328e-07,
      "loss": 0.4397,
      "step": 29273
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5982240438461304,
      "learning_rate": 3.2984606559157583e-07,
      "loss": 0.4394,
      "step": 29274
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.561220407485962,
      "learning_rate": 3.2958670831978013e-07,
      "loss": 0.4439,
      "step": 29275
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5221534967422485,
      "learning_rate": 3.293274513456324e-07,
      "loss": 0.3898,
      "step": 29276
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5422170162200928,
      "learning_rate": 3.290682946718238e-07,
      "loss": 0.4288,
      "step": 29277
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5075429677963257,
      "learning_rate": 3.2880923830104216e-07,
      "loss": 0.3479,
      "step": 29278
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5013760328292847,
      "learning_rate": 3.285502822359743e-07,
      "loss": 0.3943,
      "step": 29279
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5474661588668823,
      "learning_rate": 3.2829142647930357e-07,
      "loss": 0.3679,
      "step": 29280
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5486730337142944,
      "learning_rate": 3.2803267103371785e-07,
      "loss": 0.4135,
      "step": 29281
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.528485894203186,
      "learning_rate": 3.2777401590189827e-07,
      "loss": 0.3543,
      "step": 29282
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5217608213424683,
      "learning_rate": 3.275154610865305e-07,
      "loss": 0.3966,
      "step": 29283
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.494868516921997,
      "learning_rate": 3.272570065902947e-07,
      "loss": 0.366,
      "step": 29284
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5499744415283203,
      "learning_rate": 3.269986524158697e-07,
      "loss": 0.43,
      "step": 29285
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5156105756759644,
      "learning_rate": 3.2674039856593674e-07,
      "loss": 0.3693,
      "step": 29286
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5670045614242554,
      "learning_rate": 3.2648224504317484e-07,
      "loss": 0.4426,
      "step": 29287
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4532634019851685,
      "learning_rate": 3.2622419185026065e-07,
      "loss": 0.3788,
      "step": 29288
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.516209602355957,
      "learning_rate": 3.25966238989871e-07,
      "loss": 0.3765,
      "step": 29289
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4698281288146973,
      "learning_rate": 3.257083864646804e-07,
      "loss": 0.3624,
      "step": 29290
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6322870254516602,
      "learning_rate": 3.2545063427736445e-07,
      "loss": 0.47,
      "step": 29291
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5358269214630127,
      "learning_rate": 3.251929824305955e-07,
      "loss": 0.3817,
      "step": 29292
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.576436996459961,
      "learning_rate": 3.2493543092704584e-07,
      "loss": 0.452,
      "step": 29293
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6255030632019043,
      "learning_rate": 3.2467797976938776e-07,
      "loss": 0.3775,
      "step": 29294
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5115476846694946,
      "learning_rate": 3.2442062896029024e-07,
      "loss": 0.4327,
      "step": 29295
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4252092838287354,
      "learning_rate": 3.241633785024223e-07,
      "loss": 0.3561,
      "step": 29296
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6125805377960205,
      "learning_rate": 3.239062283984529e-07,
      "loss": 0.4299,
      "step": 29297
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.49799382686615,
      "learning_rate": 3.2364917865104874e-07,
      "loss": 0.3566,
      "step": 29298
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.450536847114563,
      "learning_rate": 3.233922292628777e-07,
      "loss": 0.3984,
      "step": 29299
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6968852281570435,
      "learning_rate": 3.2313538023659996e-07,
      "loss": 0.3856,
      "step": 29300
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5506137609481812,
      "learning_rate": 3.228786315748833e-07,
      "loss": 0.3952,
      "step": 29301
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.469316840171814,
      "learning_rate": 3.226219832803901e-07,
      "loss": 0.3482,
      "step": 29302
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5701054334640503,
      "learning_rate": 3.223654353557826e-07,
      "loss": 0.3918,
      "step": 29303
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5486563444137573,
      "learning_rate": 3.221089878037209e-07,
      "loss": 0.3847,
      "step": 29304
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5507915019989014,
      "learning_rate": 3.2185264062686297e-07,
      "loss": 0.4599,
      "step": 29305
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.492806077003479,
      "learning_rate": 3.215963938278699e-07,
      "loss": 0.3958,
      "step": 29306
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.517242670059204,
      "learning_rate": 3.213402474093996e-07,
      "loss": 0.3971,
      "step": 29307
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4864012002944946,
      "learning_rate": 3.210842013741078e-07,
      "loss": 0.373,
      "step": 29308
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5574414730072021,
      "learning_rate": 3.2082825572465004e-07,
      "loss": 0.3892,
      "step": 29309
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.49053156375885,
      "learning_rate": 3.2057241046367984e-07,
      "loss": 0.3568,
      "step": 29310
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.510035753250122,
      "learning_rate": 3.2031666559385387e-07,
      "loss": 0.4017,
      "step": 29311
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.612439513206482,
      "learning_rate": 3.2006102111782124e-07,
      "loss": 0.3893,
      "step": 29312
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5385661125183105,
      "learning_rate": 3.198054770382353e-07,
      "loss": 0.4721,
      "step": 29313
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.512745976448059,
      "learning_rate": 3.1955003335774615e-07,
      "loss": 0.401,
      "step": 29314
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5754910707473755,
      "learning_rate": 3.1929469007900394e-07,
      "loss": 0.4084,
      "step": 29315
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5554845333099365,
      "learning_rate": 3.190394472046543e-07,
      "loss": 0.3842,
      "step": 29316
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5287097692489624,
      "learning_rate": 3.1878430473734625e-07,
      "loss": 0.4041,
      "step": 29317
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5148690938949585,
      "learning_rate": 3.1852926267972653e-07,
      "loss": 0.3945,
      "step": 29318
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5256422758102417,
      "learning_rate": 3.182743210344408e-07,
      "loss": 0.4198,
      "step": 29319
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5412942171096802,
      "learning_rate": 3.180194798041314e-07,
      "loss": 0.3675,
      "step": 29320
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5199247598648071,
      "learning_rate": 3.177647389914418e-07,
      "loss": 0.3837,
      "step": 29321
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4199347496032715,
      "learning_rate": 3.1751009859901426e-07,
      "loss": 0.3526,
      "step": 29322
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.551313877105713,
      "learning_rate": 3.172555586294912e-07,
      "loss": 0.4242,
      "step": 29323
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4884353876113892,
      "learning_rate": 3.1700111908551155e-07,
      "loss": 0.3774,
      "step": 29324
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7007042169570923,
      "learning_rate": 3.167467799697144e-07,
      "loss": 0.4066,
      "step": 29325
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5056326389312744,
      "learning_rate": 3.1649254128473526e-07,
      "loss": 0.3632,
      "step": 29326
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5366376638412476,
      "learning_rate": 3.162384030332155e-07,
      "loss": 0.4217,
      "step": 29327
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.450605869293213,
      "learning_rate": 3.159843652177874e-07,
      "loss": 0.364,
      "step": 29328
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6571542024612427,
      "learning_rate": 3.1573042784108667e-07,
      "loss": 0.4322,
      "step": 29329
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5322763919830322,
      "learning_rate": 3.154765909057489e-07,
      "loss": 0.3499,
      "step": 29330
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.831370234489441,
      "learning_rate": 3.152228544144054e-07,
      "loss": 0.4075,
      "step": 29331
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4845237731933594,
      "learning_rate": 3.149692183696862e-07,
      "loss": 0.3582,
      "step": 29332
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5757174491882324,
      "learning_rate": 3.147156827742226e-07,
      "loss": 0.3822,
      "step": 29333
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5236009359359741,
      "learning_rate": 3.144622476306458e-07,
      "loss": 0.3472,
      "step": 29334
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6543811559677124,
      "learning_rate": 3.1420891294158594e-07,
      "loss": 0.4107,
      "step": 29335
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4791704416275024,
      "learning_rate": 3.1395567870966536e-07,
      "loss": 0.3481,
      "step": 29336
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8527584075927734,
      "learning_rate": 3.137025449375131e-07,
      "loss": 0.4435,
      "step": 29337
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4492064714431763,
      "learning_rate": 3.134495116277536e-07,
      "loss": 0.3587,
      "step": 29338
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5737833976745605,
      "learning_rate": 3.131965787830149e-07,
      "loss": 0.4452,
      "step": 29339
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5569440126419067,
      "learning_rate": 3.1294374640591707e-07,
      "loss": 0.3788,
      "step": 29340
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.531140923500061,
      "learning_rate": 3.1269101449908246e-07,
      "loss": 0.439,
      "step": 29341
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5929083824157715,
      "learning_rate": 3.124383830651323e-07,
      "loss": 0.4277,
      "step": 29342
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5126633644104004,
      "learning_rate": 3.1218585210668786e-07,
      "loss": 0.39,
      "step": 29343
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5331066846847534,
      "learning_rate": 3.11933421626367e-07,
      "loss": 0.3974,
      "step": 29344
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5596932172775269,
      "learning_rate": 3.1168109162678986e-07,
      "loss": 0.3912,
      "step": 29345
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.452787160873413,
      "learning_rate": 3.1142886211057097e-07,
      "loss": 0.3366,
      "step": 29346
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.675744652748108,
      "learning_rate": 3.111767330803295e-07,
      "loss": 0.415,
      "step": 29347
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5087294578552246,
      "learning_rate": 3.1092470453867653e-07,
      "loss": 0.3947,
      "step": 29348
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5821253061294556,
      "learning_rate": 3.106727764882278e-07,
      "loss": 0.4346,
      "step": 29349
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5829213857650757,
      "learning_rate": 3.104209489315979e-07,
      "loss": 0.3682,
      "step": 29350
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.511390209197998,
      "learning_rate": 3.1016922187139807e-07,
      "loss": 0.4031,
      "step": 29351
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.549077033996582,
      "learning_rate": 3.099175953102362e-07,
      "loss": 0.3801,
      "step": 29352
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5357675552368164,
      "learning_rate": 3.0966606925072363e-07,
      "loss": 0.4014,
      "step": 29353
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5298327207565308,
      "learning_rate": 3.0941464369547035e-07,
      "loss": 0.346,
      "step": 29354
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5439441204071045,
      "learning_rate": 3.0916331864708327e-07,
      "loss": 0.4294,
      "step": 29355
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.527597188949585,
      "learning_rate": 3.0891209410817024e-07,
      "loss": 0.3938,
      "step": 29356
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6675379276275635,
      "learning_rate": 3.086609700813337e-07,
      "loss": 0.9664,
      "step": 29357
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4695665836334229,
      "learning_rate": 3.0840994656918143e-07,
      "loss": 1.0149,
      "step": 29358
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6183865070343018,
      "learning_rate": 3.0815902357431484e-07,
      "loss": 0.4392,
      "step": 29359
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4742441177368164,
      "learning_rate": 3.0790820109933727e-07,
      "loss": 0.3621,
      "step": 29360
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5533828735351562,
      "learning_rate": 3.0765747914685116e-07,
      "loss": 0.4295,
      "step": 29361
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4722471237182617,
      "learning_rate": 3.0740685771945444e-07,
      "loss": 0.3631,
      "step": 29362
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6533514261245728,
      "learning_rate": 3.0715633681974944e-07,
      "loss": 0.3993,
      "step": 29363
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4496753215789795,
      "learning_rate": 3.0690591645033184e-07,
      "loss": 0.3585,
      "step": 29364
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5612571239471436,
      "learning_rate": 3.066555966137996e-07,
      "loss": 0.3981,
      "step": 29365
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5208197832107544,
      "learning_rate": 3.0640537731274957e-07,
      "loss": 0.3769,
      "step": 29366
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5235718488693237,
      "learning_rate": 3.0615525854977736e-07,
      "loss": 0.4454,
      "step": 29367
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.493779182434082,
      "learning_rate": 3.059052403274754e-07,
      "loss": 0.3672,
      "step": 29368
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5143967866897583,
      "learning_rate": 3.056553226484382e-07,
      "loss": 0.4178,
      "step": 29369
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.487205982208252,
      "learning_rate": 3.0540550551525605e-07,
      "loss": 0.3745,
      "step": 29370
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5190598964691162,
      "learning_rate": 3.0515578893052343e-07,
      "loss": 0.4134,
      "step": 29371
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4752144813537598,
      "learning_rate": 3.0490617289682724e-07,
      "loss": 0.3623,
      "step": 29372
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.751561164855957,
      "learning_rate": 3.0465665741675755e-07,
      "loss": 0.4216,
      "step": 29373
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6406315565109253,
      "learning_rate": 3.044072424929001e-07,
      "loss": 0.381,
      "step": 29374
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.495232343673706,
      "learning_rate": 3.0415792812784617e-07,
      "loss": 0.4417,
      "step": 29375
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5190335512161255,
      "learning_rate": 3.039087143241781e-07,
      "loss": 0.3844,
      "step": 29376
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6079102754592896,
      "learning_rate": 3.036596010844817e-07,
      "loss": 0.3954,
      "step": 29377
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4643385410308838,
      "learning_rate": 3.034105884113392e-07,
      "loss": 0.3597,
      "step": 29378
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5400711297988892,
      "learning_rate": 3.031616763073353e-07,
      "loss": 0.4315,
      "step": 29379
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4997172355651855,
      "learning_rate": 3.029128647750512e-07,
      "loss": 0.3639,
      "step": 29380
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6266428232192993,
      "learning_rate": 3.0266415381706716e-07,
      "loss": 0.4708,
      "step": 29381
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5572686195373535,
      "learning_rate": 3.0241554343596215e-07,
      "loss": 0.3709,
      "step": 29382
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.548857569694519,
      "learning_rate": 3.0216703363431523e-07,
      "loss": 0.4047,
      "step": 29383
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4960099458694458,
      "learning_rate": 3.0191862441470434e-07,
      "loss": 0.369,
      "step": 29384
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4789496660232544,
      "learning_rate": 3.0167031577970405e-07,
      "loss": 0.4024,
      "step": 29385
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.471140742301941,
      "learning_rate": 3.014221077318924e-07,
      "loss": 0.3614,
      "step": 29386
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6930837631225586,
      "learning_rate": 3.0117400027384167e-07,
      "loss": 0.4087,
      "step": 29387
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.538371205329895,
      "learning_rate": 3.0092599340812547e-07,
      "loss": 0.4155,
      "step": 29388
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4521781206130981,
      "learning_rate": 3.006780871373172e-07,
      "loss": 0.4041,
      "step": 29389
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4910001754760742,
      "learning_rate": 3.004302814639859e-07,
      "loss": 0.3858,
      "step": 29390
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5919687747955322,
      "learning_rate": 3.00182576390704e-07,
      "loss": 0.3955,
      "step": 29391
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5830345153808594,
      "learning_rate": 2.999349719200406e-07,
      "loss": 0.3775,
      "step": 29392
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5573859214782715,
      "learning_rate": 2.996874680545603e-07,
      "loss": 0.4377,
      "step": 29393
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5197147130966187,
      "learning_rate": 2.994400647968332e-07,
      "loss": 0.3792,
      "step": 29394
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5210129022598267,
      "learning_rate": 2.9919276214942506e-07,
      "loss": 0.4319,
      "step": 29395
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5477676391601562,
      "learning_rate": 2.989455601149005e-07,
      "loss": 0.379,
      "step": 29396
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4455111026763916,
      "learning_rate": 2.9869845869582413e-07,
      "loss": 0.3929,
      "step": 29397
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.53559410572052,
      "learning_rate": 2.984514578947562e-07,
      "loss": 0.3657,
      "step": 29398
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5158023834228516,
      "learning_rate": 2.9820455771426114e-07,
      "loss": 0.4309,
      "step": 29399
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5614887475967407,
      "learning_rate": 2.979577581568993e-07,
      "loss": 0.3608,
      "step": 29400
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.581707000732422,
      "learning_rate": 2.9771105922522856e-07,
      "loss": 0.4181,
      "step": 29401
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8976491689682007,
      "learning_rate": 2.974644609218091e-07,
      "loss": 0.3811,
      "step": 29402
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5419312715530396,
      "learning_rate": 2.972179632491989e-07,
      "loss": 0.3966,
      "step": 29403
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5937772989273071,
      "learning_rate": 2.969715662099537e-07,
      "loss": 0.378,
      "step": 29404
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.538669466972351,
      "learning_rate": 2.967252698066292e-07,
      "loss": 0.4337,
      "step": 29405
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5108275413513184,
      "learning_rate": 2.9647907404178e-07,
      "loss": 0.3703,
      "step": 29406
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6796112060546875,
      "learning_rate": 2.962329789179597e-07,
      "loss": 0.4105,
      "step": 29407
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.454641580581665,
      "learning_rate": 2.9598698443772165e-07,
      "loss": 0.3684,
      "step": 29408
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5357509851455688,
      "learning_rate": 2.95741090603614e-07,
      "loss": 0.3975,
      "step": 29409
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5299708843231201,
      "learning_rate": 2.954952974181902e-07,
      "loss": 0.3371,
      "step": 29410
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6286934614181519,
      "learning_rate": 2.9524960488399703e-07,
      "loss": 0.4164,
      "step": 29411
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4596071243286133,
      "learning_rate": 2.9500401300358695e-07,
      "loss": 0.3492,
      "step": 29412
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5799607038497925,
      "learning_rate": 2.9475852177950235e-07,
      "loss": 0.4383,
      "step": 29413
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5339375734329224,
      "learning_rate": 2.9451313121429127e-07,
      "loss": 0.3791,
      "step": 29414
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5582621097564697,
      "learning_rate": 2.9426784131049936e-07,
      "loss": 0.4633,
      "step": 29415
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4884583950042725,
      "learning_rate": 2.9402265207067015e-07,
      "loss": 0.363,
      "step": 29416
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6428329944610596,
      "learning_rate": 2.9377756349734834e-07,
      "loss": 0.4628,
      "step": 29417
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5407564640045166,
      "learning_rate": 2.935325755930718e-07,
      "loss": 0.3655,
      "step": 29418
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7147061824798584,
      "learning_rate": 2.932876883603841e-07,
      "loss": 0.42,
      "step": 29419
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5284632444381714,
      "learning_rate": 2.930429018018255e-07,
      "loss": 0.3582,
      "step": 29420
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5009474754333496,
      "learning_rate": 2.9279821591993386e-07,
      "loss": 0.4715,
      "step": 29421
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5449477434158325,
      "learning_rate": 2.925536307172483e-07,
      "loss": 0.3849,
      "step": 29422
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.486963152885437,
      "learning_rate": 2.9230914619630346e-07,
      "loss": 0.4142,
      "step": 29423
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.426392912864685,
      "learning_rate": 2.920647623596373e-07,
      "loss": 0.3513,
      "step": 29424
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.487152099609375,
      "learning_rate": 2.918204792097823e-07,
      "loss": 0.3972,
      "step": 29425
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4872897863388062,
      "learning_rate": 2.91576296749273e-07,
      "loss": 0.3861,
      "step": 29426
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5714904069900513,
      "learning_rate": 2.91332214980643e-07,
      "loss": 0.4312,
      "step": 29427
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.503993272781372,
      "learning_rate": 2.910882339064236e-07,
      "loss": 0.3662,
      "step": 29428
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.497226357460022,
      "learning_rate": 2.9084435352914274e-07,
      "loss": 0.4103,
      "step": 29429
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4998899698257446,
      "learning_rate": 2.906005738513318e-07,
      "loss": 0.3884,
      "step": 29430
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.607601523399353,
      "learning_rate": 2.9035689487551864e-07,
      "loss": 0.4565,
      "step": 29431
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4521400928497314,
      "learning_rate": 2.901133166042325e-07,
      "loss": 0.3657,
      "step": 29432
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4958161115646362,
      "learning_rate": 2.898698390399979e-07,
      "loss": 0.4096,
      "step": 29433
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.473468542098999,
      "learning_rate": 2.896264621853395e-07,
      "loss": 0.3577,
      "step": 29434
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6406999826431274,
      "learning_rate": 2.8938318604278314e-07,
      "loss": 0.4209,
      "step": 29435
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.491020917892456,
      "learning_rate": 2.891400106148512e-07,
      "loss": 0.3703,
      "step": 29436
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6265132427215576,
      "learning_rate": 2.88896935904065e-07,
      "loss": 0.4162,
      "step": 29437
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4624865055084229,
      "learning_rate": 2.8865396191294805e-07,
      "loss": 0.3541,
      "step": 29438
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5817426443099976,
      "learning_rate": 2.8841108864401615e-07,
      "loss": 0.4221,
      "step": 29439
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4944134950637817,
      "learning_rate": 2.8816831609979387e-07,
      "loss": 0.3807,
      "step": 29440
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6060409545898438,
      "learning_rate": 2.879256442827938e-07,
      "loss": 0.44,
      "step": 29441
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4909666776657104,
      "learning_rate": 2.87683073195536e-07,
      "loss": 0.3484,
      "step": 29442
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.583868384361267,
      "learning_rate": 2.874406028405352e-07,
      "loss": 0.4251,
      "step": 29443
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5401363372802734,
      "learning_rate": 2.8719823322030827e-07,
      "loss": 0.3565,
      "step": 29444
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4965379238128662,
      "learning_rate": 2.869559643373654e-07,
      "loss": 0.3926,
      "step": 29445
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6110960245132446,
      "learning_rate": 2.867137961942201e-07,
      "loss": 0.4108,
      "step": 29446
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7907438278198242,
      "learning_rate": 2.8647172879338603e-07,
      "loss": 0.3811,
      "step": 29447
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4741634130477905,
      "learning_rate": 2.862297621373744e-07,
      "loss": 0.349,
      "step": 29448
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.499041199684143,
      "learning_rate": 2.8598789622869215e-07,
      "loss": 0.3657,
      "step": 29449
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4930461645126343,
      "learning_rate": 2.857461310698495e-07,
      "loss": 0.3586,
      "step": 29450
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.499487280845642,
      "learning_rate": 2.8550446666335217e-07,
      "loss": 0.417,
      "step": 29451
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5257638692855835,
      "learning_rate": 2.852629030117082e-07,
      "loss": 0.3754,
      "step": 29452
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5982472896575928,
      "learning_rate": 2.850214401174234e-07,
      "loss": 0.4095,
      "step": 29453
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6524057388305664,
      "learning_rate": 2.847800779830001e-07,
      "loss": 0.4155,
      "step": 29454
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5270907878875732,
      "learning_rate": 2.845388166109431e-07,
      "loss": 0.3983,
      "step": 29455
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5113741159439087,
      "learning_rate": 2.842976560037558e-07,
      "loss": 0.3507,
      "step": 29456
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5456172227859497,
      "learning_rate": 2.8405659616393635e-07,
      "loss": 0.4388,
      "step": 29457
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5917394161224365,
      "learning_rate": 2.8381563709398705e-07,
      "loss": 0.3874,
      "step": 29458
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6147780418395996,
      "learning_rate": 2.8357477879640606e-07,
      "loss": 0.3902,
      "step": 29459
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4755336046218872,
      "learning_rate": 2.8333402127369346e-07,
      "loss": 0.3634,
      "step": 29460
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.534092664718628,
      "learning_rate": 2.830933645283429e-07,
      "loss": 0.4034,
      "step": 29461
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5273847579956055,
      "learning_rate": 2.828528085628535e-07,
      "loss": 0.3835,
      "step": 29462
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5308200120925903,
      "learning_rate": 2.826123533797187e-07,
      "loss": 0.4042,
      "step": 29463
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4884151220321655,
      "learning_rate": 2.823719989814333e-07,
      "loss": 0.3469,
      "step": 29464
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7608968019485474,
      "learning_rate": 2.821317453704886e-07,
      "loss": 0.4156,
      "step": 29465
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5191797018051147,
      "learning_rate": 2.8189159254937704e-07,
      "loss": 0.3615,
      "step": 29466
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5888320207595825,
      "learning_rate": 2.8165154052058996e-07,
      "loss": 0.4906,
      "step": 29467
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5039947032928467,
      "learning_rate": 2.814115892866176e-07,
      "loss": 0.3819,
      "step": 29468
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.575602650642395,
      "learning_rate": 2.81171738849948e-07,
      "loss": 0.4628,
      "step": 29469
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4568064212799072,
      "learning_rate": 2.80931989213068e-07,
      "loss": 0.3417,
      "step": 29470
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.576894998550415,
      "learning_rate": 2.806923403784645e-07,
      "loss": 0.4281,
      "step": 29471
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4799187183380127,
      "learning_rate": 2.8045279234862443e-07,
      "loss": 0.378,
      "step": 29472
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5449832677841187,
      "learning_rate": 2.8021334512603025e-07,
      "loss": 0.4441,
      "step": 29473
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5586469173431396,
      "learning_rate": 2.799739987131678e-07,
      "loss": 0.3683,
      "step": 29474
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5917445421218872,
      "learning_rate": 2.7973475311251606e-07,
      "loss": 0.4182,
      "step": 29475
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5412555932998657,
      "learning_rate": 2.79495608326561e-07,
      "loss": 0.3738,
      "step": 29476
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.157855272293091,
      "learning_rate": 2.7925656435777825e-07,
      "loss": 0.47,
      "step": 29477
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5266040563583374,
      "learning_rate": 2.790176212086493e-07,
      "loss": 0.3567,
      "step": 29478
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5964815616607666,
      "learning_rate": 2.7877877888165317e-07,
      "loss": 0.3768,
      "step": 29479
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.519452691078186,
      "learning_rate": 2.785400373792668e-07,
      "loss": 0.3687,
      "step": 29480
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4746372699737549,
      "learning_rate": 2.783013967039638e-07,
      "loss": 0.4119,
      "step": 29481
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4448797702789307,
      "learning_rate": 2.780628568582211e-07,
      "loss": 0.3785,
      "step": 29482
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6282957792282104,
      "learning_rate": 2.778244178445133e-07,
      "loss": 0.4593,
      "step": 29483
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5341300964355469,
      "learning_rate": 2.7758607966531294e-07,
      "loss": 0.3732,
      "step": 29484
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4529774188995361,
      "learning_rate": 2.7734784232309356e-07,
      "loss": 0.3773,
      "step": 29485
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5152788162231445,
      "learning_rate": 2.7710970582032204e-07,
      "loss": 0.3647,
      "step": 29486
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5493048429489136,
      "learning_rate": 2.7687167015947093e-07,
      "loss": 0.4234,
      "step": 29487
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5169260501861572,
      "learning_rate": 2.766337353430093e-07,
      "loss": 0.3763,
      "step": 29488
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.549998164176941,
      "learning_rate": 2.763959013734041e-07,
      "loss": 0.4097,
      "step": 29489
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.501172661781311,
      "learning_rate": 2.761581682531223e-07,
      "loss": 0.3451,
      "step": 29490
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4793511629104614,
      "learning_rate": 2.759205359846284e-07,
      "loss": 0.3829,
      "step": 29491
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.533078908920288,
      "learning_rate": 2.7568300457038953e-07,
      "loss": 0.3793,
      "step": 29492
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5515170097351074,
      "learning_rate": 2.75445574012867e-07,
      "loss": 0.4326,
      "step": 29493
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5065375566482544,
      "learning_rate": 2.7520824431452543e-07,
      "loss": 0.3663,
      "step": 29494
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5428153276443481,
      "learning_rate": 2.74971015477824e-07,
      "loss": 0.4323,
      "step": 29495
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7240523099899292,
      "learning_rate": 2.7473388750522523e-07,
      "loss": 0.3973,
      "step": 29496
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4943453073501587,
      "learning_rate": 2.74496860399186e-07,
      "loss": 0.4084,
      "step": 29497
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6228715181350708,
      "learning_rate": 2.7425993416216657e-07,
      "loss": 0.395,
      "step": 29498
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5032309293746948,
      "learning_rate": 2.7402310879662497e-07,
      "loss": 0.4098,
      "step": 29499
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5544981956481934,
      "learning_rate": 2.73786384305017e-07,
      "loss": 0.3852,
      "step": 29500
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5026404857635498,
      "learning_rate": 2.7354976068979524e-07,
      "loss": 0.4035,
      "step": 29501
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6036802530288696,
      "learning_rate": 2.7331323795341647e-07,
      "loss": 0.3788,
      "step": 29502
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.519116759300232,
      "learning_rate": 2.7307681609833216e-07,
      "loss": 0.4286,
      "step": 29503
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6081386804580688,
      "learning_rate": 2.7284049512699697e-07,
      "loss": 0.3598,
      "step": 29504
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4581230878829956,
      "learning_rate": 2.7260427504186115e-07,
      "loss": 0.4016,
      "step": 29505
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5012999773025513,
      "learning_rate": 2.723681558453717e-07,
      "loss": 0.3492,
      "step": 29506
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.532170295715332,
      "learning_rate": 2.7213213753998103e-07,
      "loss": 0.447,
      "step": 29507
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.50456702709198,
      "learning_rate": 2.71896220128135e-07,
      "loss": 0.3704,
      "step": 29508
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.546515703201294,
      "learning_rate": 2.7166040361228165e-07,
      "loss": 0.4303,
      "step": 29509
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5523799657821655,
      "learning_rate": 2.7142468799486676e-07,
      "loss": 0.3788,
      "step": 29510
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5052984952926636,
      "learning_rate": 2.71189073278334e-07,
      "loss": 0.4193,
      "step": 29511
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.532145380973816,
      "learning_rate": 2.7095355946512805e-07,
      "loss": 0.3747,
      "step": 29512
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5687568187713623,
      "learning_rate": 2.7071814655769134e-07,
      "loss": 0.4124,
      "step": 29513
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4980067014694214,
      "learning_rate": 2.7048283455846426e-07,
      "loss": 0.3747,
      "step": 29514
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5013786554336548,
      "learning_rate": 2.702476234698892e-07,
      "loss": 0.3971,
      "step": 29515
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4799997806549072,
      "learning_rate": 2.700125132944065e-07,
      "loss": 0.3451,
      "step": 29516
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5376837253570557,
      "learning_rate": 2.697775040344508e-07,
      "loss": 0.4481,
      "step": 29517
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5410003662109375,
      "learning_rate": 2.6954259569246246e-07,
      "loss": 0.3639,
      "step": 29518
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5259358882904053,
      "learning_rate": 2.693077882708761e-07,
      "loss": 0.4008,
      "step": 29519
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4678115844726562,
      "learning_rate": 2.690730817721299e-07,
      "loss": 0.388,
      "step": 29520
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6095727682113647,
      "learning_rate": 2.6883847619865733e-07,
      "loss": 0.4085,
      "step": 29521
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.639032006263733,
      "learning_rate": 2.6860397155288875e-07,
      "loss": 0.4462,
      "step": 29522
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6465070247650146,
      "learning_rate": 2.6836956783725774e-07,
      "loss": 0.415,
      "step": 29523
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4995497465133667,
      "learning_rate": 2.68135265054198e-07,
      "loss": 0.3688,
      "step": 29524
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6419014930725098,
      "learning_rate": 2.6790106320613516e-07,
      "loss": 0.4416,
      "step": 29525
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5021544694900513,
      "learning_rate": 2.6766696229550307e-07,
      "loss": 0.3846,
      "step": 29526
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6293190717697144,
      "learning_rate": 2.674329623247263e-07,
      "loss": 0.4263,
      "step": 29527
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.541675090789795,
      "learning_rate": 2.671990632962329e-07,
      "loss": 0.383,
      "step": 29528
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6125438213348389,
      "learning_rate": 2.669652652124488e-07,
      "loss": 0.4732,
      "step": 29529
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5257973670959473,
      "learning_rate": 2.667315680757998e-07,
      "loss": 0.3771,
      "step": 29530
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5307316780090332,
      "learning_rate": 2.664979718887073e-07,
      "loss": 0.4172,
      "step": 29531
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4897537231445312,
      "learning_rate": 2.66264476653596e-07,
      "loss": 0.3534,
      "step": 29532
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5330018997192383,
      "learning_rate": 2.660310823728873e-07,
      "loss": 0.4354,
      "step": 29533
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5417572259902954,
      "learning_rate": 2.657977890490004e-07,
      "loss": 0.3759,
      "step": 29534
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5288360118865967,
      "learning_rate": 2.6556459668435783e-07,
      "loss": 0.438,
      "step": 29535
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5301684141159058,
      "learning_rate": 2.653315052813765e-07,
      "loss": 0.3765,
      "step": 29536
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.576206088066101,
      "learning_rate": 2.650985148424734e-07,
      "loss": 0.427,
      "step": 29537
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5244580507278442,
      "learning_rate": 2.6486562537006544e-07,
      "loss": 0.3866,
      "step": 29538
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6518645286560059,
      "learning_rate": 2.6463283686656735e-07,
      "loss": 0.4504,
      "step": 29539
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4705342054367065,
      "learning_rate": 2.644001493343962e-07,
      "loss": 0.3584,
      "step": 29540
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5165424346923828,
      "learning_rate": 2.6416756277596324e-07,
      "loss": 0.3858,
      "step": 29541
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4947515726089478,
      "learning_rate": 2.6393507719367996e-07,
      "loss": 0.3633,
      "step": 29542
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5291205644607544,
      "learning_rate": 2.6370269258995885e-07,
      "loss": 0.3714,
      "step": 29543
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.664526104927063,
      "learning_rate": 2.634704089672113e-07,
      "loss": 0.4254,
      "step": 29544
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5967351198196411,
      "learning_rate": 2.6323822632784325e-07,
      "loss": 0.4229,
      "step": 29545
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6077890396118164,
      "learning_rate": 2.630061446742671e-07,
      "loss": 0.3868,
      "step": 29546
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4758573770523071,
      "learning_rate": 2.6277416400888546e-07,
      "loss": 0.404,
      "step": 29547
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5055322647094727,
      "learning_rate": 2.6254228433410743e-07,
      "loss": 0.3624,
      "step": 29548
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.61272394657135,
      "learning_rate": 2.623105056523356e-07,
      "loss": 0.4547,
      "step": 29549
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.588623046875,
      "learning_rate": 2.620788279659758e-07,
      "loss": 0.4232,
      "step": 29550
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5529439449310303,
      "learning_rate": 2.618472512774306e-07,
      "loss": 0.4439,
      "step": 29551
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.666353702545166,
      "learning_rate": 2.6161577558910137e-07,
      "loss": 0.3707,
      "step": 29552
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5703011751174927,
      "learning_rate": 2.6138440090338944e-07,
      "loss": 0.4629,
      "step": 29553
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5754843950271606,
      "learning_rate": 2.6115312722269306e-07,
      "loss": 0.3739,
      "step": 29554
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.613154411315918,
      "learning_rate": 2.609219545494113e-07,
      "loss": 0.3958,
      "step": 29555
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4977235794067383,
      "learning_rate": 2.606908828859445e-07,
      "loss": 0.3575,
      "step": 29556
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.664844036102295,
      "learning_rate": 2.6045991223468736e-07,
      "loss": 0.4728,
      "step": 29557
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.492573857307434,
      "learning_rate": 2.602290425980336e-07,
      "loss": 0.375,
      "step": 29558
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5142537355422974,
      "learning_rate": 2.5999827397837905e-07,
      "loss": 0.3902,
      "step": 29559
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6182730197906494,
      "learning_rate": 2.597676063781196e-07,
      "loss": 0.3831,
      "step": 29560
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7410376071929932,
      "learning_rate": 2.5953703979964327e-07,
      "loss": 0.8977,
      "step": 29561
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.43705415725708,
      "learning_rate": 2.5930657424534486e-07,
      "loss": 0.9381,
      "step": 29562
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4940862655639648,
      "learning_rate": 2.590762097176136e-07,
      "loss": 0.3798,
      "step": 29563
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6679437160491943,
      "learning_rate": 2.588459462188386e-07,
      "loss": 0.3595,
      "step": 29564
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5966883897781372,
      "learning_rate": 2.586157837514069e-07,
      "loss": 0.4309,
      "step": 29565
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.719283103942871,
      "learning_rate": 2.5838572231770773e-07,
      "loss": 0.4044,
      "step": 29566
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5158500671386719,
      "learning_rate": 2.581557619201258e-07,
      "loss": 0.3845,
      "step": 29567
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5162439346313477,
      "learning_rate": 2.579259025610481e-07,
      "loss": 0.3785,
      "step": 29568
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.537184238433838,
      "learning_rate": 2.5769614424285496e-07,
      "loss": 0.4282,
      "step": 29569
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5288645029067993,
      "learning_rate": 2.574664869679322e-07,
      "loss": 0.376,
      "step": 29570
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5426416397094727,
      "learning_rate": 2.572369307386613e-07,
      "loss": 0.4125,
      "step": 29571
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5078603029251099,
      "learning_rate": 2.5700747555742364e-07,
      "loss": 0.3619,
      "step": 29572
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.545385479927063,
      "learning_rate": 2.567781214265974e-07,
      "loss": 0.4373,
      "step": 29573
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4884110689163208,
      "learning_rate": 2.5654886834856176e-07,
      "loss": 0.3796,
      "step": 29574
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.551034688949585,
      "learning_rate": 2.5631971632569585e-07,
      "loss": 0.402,
      "step": 29575
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5552366971969604,
      "learning_rate": 2.560906653603756e-07,
      "loss": 0.3737,
      "step": 29576
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.5209593772888184,
      "learning_rate": 2.5586171545497585e-07,
      "loss": 0.4304,
      "step": 29577
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5224028825759888,
      "learning_rate": 2.5563286661187234e-07,
      "loss": 0.3728,
      "step": 29578
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6462963819503784,
      "learning_rate": 2.554041188334366e-07,
      "loss": 0.4264,
      "step": 29579
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5269947052001953,
      "learning_rate": 2.551754721220445e-07,
      "loss": 0.3888,
      "step": 29580
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6325757503509521,
      "learning_rate": 2.5494692648006414e-07,
      "loss": 0.4122,
      "step": 29581
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.486267328262329,
      "learning_rate": 2.54718481909868e-07,
      "loss": 0.3897,
      "step": 29582
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5253582000732422,
      "learning_rate": 2.544901384138243e-07,
      "loss": 0.4428,
      "step": 29583
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5002237558364868,
      "learning_rate": 2.5426189599430217e-07,
      "loss": 0.4025,
      "step": 29584
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4700983762741089,
      "learning_rate": 2.540337546536675e-07,
      "loss": 0.405,
      "step": 29585
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4769808053970337,
      "learning_rate": 2.5380571439428736e-07,
      "loss": 0.3719,
      "step": 29586
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.545416235923767,
      "learning_rate": 2.535777752185287e-07,
      "loss": 0.3982,
      "step": 29587
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.468835711479187,
      "learning_rate": 2.533499371287529e-07,
      "loss": 0.3554,
      "step": 29588
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6003260612487793,
      "learning_rate": 2.531222001273237e-07,
      "loss": 0.4217,
      "step": 29589
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4459609985351562,
      "learning_rate": 2.5289456421660255e-07,
      "loss": 0.3712,
      "step": 29590
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6092661619186401,
      "learning_rate": 2.5266702939895196e-07,
      "loss": 0.4326,
      "step": 29591
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.522581934928894,
      "learning_rate": 2.5243959567673117e-07,
      "loss": 0.3777,
      "step": 29592
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6158177852630615,
      "learning_rate": 2.522122630522994e-07,
      "loss": 0.4831,
      "step": 29593
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5187681913375854,
      "learning_rate": 2.519850315280126e-07,
      "loss": 0.3624,
      "step": 29594
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.536160945892334,
      "learning_rate": 2.517579011062299e-07,
      "loss": 0.398,
      "step": 29595
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5966259241104126,
      "learning_rate": 2.5153087178930613e-07,
      "loss": 0.4094,
      "step": 29596
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5175102949142456,
      "learning_rate": 2.51303943579595e-07,
      "loss": 0.3897,
      "step": 29597
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6035467386245728,
      "learning_rate": 2.510771164794512e-07,
      "loss": 0.363,
      "step": 29598
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5663772821426392,
      "learning_rate": 2.5085039049122737e-07,
      "loss": 0.4114,
      "step": 29599
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.453973650932312,
      "learning_rate": 2.50623765617275e-07,
      "loss": 0.3473,
      "step": 29600
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5205078125,
      "learning_rate": 2.503972418599432e-07,
      "loss": 0.3955,
      "step": 29601
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5395138263702393,
      "learning_rate": 2.5017081922158237e-07,
      "loss": 0.3771,
      "step": 29602
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.540661334991455,
      "learning_rate": 2.4994449770454176e-07,
      "loss": 0.3958,
      "step": 29603
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4642276763916016,
      "learning_rate": 2.497182773111684e-07,
      "loss": 0.3677,
      "step": 29604
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5975080728530884,
      "learning_rate": 2.49492158043807e-07,
      "loss": 0.412,
      "step": 29605
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4852465391159058,
      "learning_rate": 2.4926613990480354e-07,
      "loss": 0.3636,
      "step": 29606
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5534427165985107,
      "learning_rate": 2.4904022289650276e-07,
      "loss": 0.3971,
      "step": 29607
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5291261672973633,
      "learning_rate": 2.4881440702124837e-07,
      "loss": 0.3699,
      "step": 29608
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5705329179763794,
      "learning_rate": 2.4858869228137963e-07,
      "loss": 0.384,
      "step": 29609
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5909335613250732,
      "learning_rate": 2.4836307867924014e-07,
      "loss": 0.4026,
      "step": 29610
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.642690658569336,
      "learning_rate": 2.481375662171692e-07,
      "loss": 0.3911,
      "step": 29611
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.610971212387085,
      "learning_rate": 2.479121548975061e-07,
      "loss": 0.4105,
      "step": 29612
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5082980394363403,
      "learning_rate": 2.476868447225889e-07,
      "loss": 0.4207,
      "step": 29613
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.562975287437439,
      "learning_rate": 2.4746163569475344e-07,
      "loss": 0.4187,
      "step": 29614
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4569480419158936,
      "learning_rate": 2.4723652781633466e-07,
      "loss": 0.4034,
      "step": 29615
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4925516843795776,
      "learning_rate": 2.470115210896706e-07,
      "loss": 0.3722,
      "step": 29616
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.486722469329834,
      "learning_rate": 2.467866155170906e-07,
      "loss": 0.4143,
      "step": 29617
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.57881498336792,
      "learning_rate": 2.4656181110093156e-07,
      "loss": 0.38,
      "step": 29618
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6075401306152344,
      "learning_rate": 2.4633710784352174e-07,
      "loss": 0.3983,
      "step": 29619
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5119037628173828,
      "learning_rate": 2.461125057471947e-07,
      "loss": 0.3792,
      "step": 29620
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5007330179214478,
      "learning_rate": 2.4588800481427644e-07,
      "loss": 0.3883,
      "step": 29621
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4870400428771973,
      "learning_rate": 2.456636050470973e-07,
      "loss": 0.3702,
      "step": 29622
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.517006754875183,
      "learning_rate": 2.4543930644798543e-07,
      "loss": 0.4021,
      "step": 29623
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5190784931182861,
      "learning_rate": 2.4521510901926673e-07,
      "loss": 0.386,
      "step": 29624
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8301708698272705,
      "learning_rate": 2.4499101276326487e-07,
      "loss": 0.4597,
      "step": 29625
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.514365553855896,
      "learning_rate": 2.447670176823058e-07,
      "loss": 0.3904,
      "step": 29626
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6154862642288208,
      "learning_rate": 2.4454312377871105e-07,
      "loss": 0.4472,
      "step": 29627
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5738588571548462,
      "learning_rate": 2.443193310548053e-07,
      "loss": 0.3689,
      "step": 29628
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5099948644638062,
      "learning_rate": 2.440956395129079e-07,
      "loss": 0.3782,
      "step": 29629
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5663676261901855,
      "learning_rate": 2.438720491553392e-07,
      "loss": 0.4031,
      "step": 29630
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5097812414169312,
      "learning_rate": 2.436485599844174e-07,
      "loss": 0.4372,
      "step": 29631
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5341366529464722,
      "learning_rate": 2.4342517200246276e-07,
      "loss": 0.3829,
      "step": 29632
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.482723355293274,
      "learning_rate": 2.43201885211789e-07,
      "loss": 0.4314,
      "step": 29633
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.556929349899292,
      "learning_rate": 2.4297869961471544e-07,
      "loss": 0.3949,
      "step": 29634
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5764272212982178,
      "learning_rate": 2.427556152135535e-07,
      "loss": 0.4267,
      "step": 29635
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5847936868667603,
      "learning_rate": 2.4253263201061804e-07,
      "loss": 0.3642,
      "step": 29636
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6242808103561401,
      "learning_rate": 2.423097500082239e-07,
      "loss": 0.8756,
      "step": 29637
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4538497924804688,
      "learning_rate": 2.4208696920867915e-07,
      "loss": 0.9983,
      "step": 29638
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.498429536819458,
      "learning_rate": 2.4186428961429864e-07,
      "loss": 0.4219,
      "step": 29639
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.512352466583252,
      "learning_rate": 2.416417112273872e-07,
      "loss": 0.3922,
      "step": 29640
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5729985237121582,
      "learning_rate": 2.4141923405025746e-07,
      "loss": 0.3943,
      "step": 29641
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5362441539764404,
      "learning_rate": 2.411968580852131e-07,
      "loss": 0.3633,
      "step": 29642
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5840051174163818,
      "learning_rate": 2.4097458333456337e-07,
      "loss": 0.4522,
      "step": 29643
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4713547229766846,
      "learning_rate": 2.407524098006131e-07,
      "loss": 0.3809,
      "step": 29644
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.578230381011963,
      "learning_rate": 2.4053033748566604e-07,
      "loss": 0.4617,
      "step": 29645
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4895302057266235,
      "learning_rate": 2.403083663920247e-07,
      "loss": 0.4108,
      "step": 29646
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5374888181686401,
      "learning_rate": 2.400864965219929e-07,
      "loss": 0.3992,
      "step": 29647
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4740678071975708,
      "learning_rate": 2.3986472787787096e-07,
      "loss": 0.384,
      "step": 29648
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5514551401138306,
      "learning_rate": 2.396430604619593e-07,
      "loss": 0.3823,
      "step": 29649
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.431916356086731,
      "learning_rate": 2.39421494276556e-07,
      "loss": 0.3317,
      "step": 29650
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.101931095123291,
      "learning_rate": 2.3920002932395937e-07,
      "loss": 0.4186,
      "step": 29651
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5353341102600098,
      "learning_rate": 2.3897866560646745e-07,
      "loss": 0.3633,
      "step": 29652
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.74068021774292,
      "learning_rate": 2.387574031263762e-07,
      "loss": 0.5297,
      "step": 29653
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5106313228607178,
      "learning_rate": 2.3853624188597935e-07,
      "loss": 0.3516,
      "step": 29654
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5846341848373413,
      "learning_rate": 2.3831518188756953e-07,
      "loss": 0.4115,
      "step": 29655
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.675270438194275,
      "learning_rate": 2.3809422313344266e-07,
      "loss": 0.3978,
      "step": 29656
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5398122072219849,
      "learning_rate": 2.3787336562588803e-07,
      "loss": 0.3999,
      "step": 29657
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5371354818344116,
      "learning_rate": 2.3765260936719715e-07,
      "loss": 0.4296,
      "step": 29658
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5200729370117188,
      "learning_rate": 2.3743195435966038e-07,
      "loss": 0.4171,
      "step": 29659
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4918798208236694,
      "learning_rate": 2.372114006055637e-07,
      "loss": 0.3522,
      "step": 29660
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.3173317909240723,
      "learning_rate": 2.3699094810719748e-07,
      "loss": 0.4348,
      "step": 29661
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5291345119476318,
      "learning_rate": 2.3677059686684657e-07,
      "loss": 0.395,
      "step": 29662
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4793809652328491,
      "learning_rate": 2.365503468867958e-07,
      "loss": 0.4113,
      "step": 29663
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.495504379272461,
      "learning_rate": 2.3633019816933224e-07,
      "loss": 0.3763,
      "step": 29664
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4685126543045044,
      "learning_rate": 2.3611015071673737e-07,
      "loss": 0.4049,
      "step": 29665
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4989738464355469,
      "learning_rate": 2.3589020453129274e-07,
      "loss": 0.3877,
      "step": 29666
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5418418645858765,
      "learning_rate": 2.3567035961527983e-07,
      "loss": 0.3858,
      "step": 29667
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5296753644943237,
      "learning_rate": 2.3545061597097908e-07,
      "loss": 0.3852,
      "step": 29668
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7369284629821777,
      "learning_rate": 2.3523097360067083e-07,
      "loss": 0.415,
      "step": 29669
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5989437103271484,
      "learning_rate": 2.3501143250663216e-07,
      "loss": 0.3558,
      "step": 29670
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5922505855560303,
      "learning_rate": 2.3479199269113905e-07,
      "loss": 0.4533,
      "step": 29671
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5214194059371948,
      "learning_rate": 2.3457265415646857e-07,
      "loss": 0.3493,
      "step": 29672
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.612223744392395,
      "learning_rate": 2.3435341690489666e-07,
      "loss": 0.3931,
      "step": 29673
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5404314994812012,
      "learning_rate": 2.3413428093869372e-07,
      "loss": 0.3701,
      "step": 29674
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6138931512832642,
      "learning_rate": 2.3391524626013685e-07,
      "loss": 0.4062,
      "step": 29675
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.523194670677185,
      "learning_rate": 2.336963128714942e-07,
      "loss": 0.3531,
      "step": 29676
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6008365154266357,
      "learning_rate": 2.334774807750384e-07,
      "loss": 0.4244,
      "step": 29677
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5218809843063354,
      "learning_rate": 2.3325874997303765e-07,
      "loss": 0.3575,
      "step": 29678
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4857257604599,
      "learning_rate": 2.3304012046776236e-07,
      "loss": 0.3751,
      "step": 29679
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.431867003440857,
      "learning_rate": 2.328215922614796e-07,
      "loss": 0.3649,
      "step": 29680
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5034642219543457,
      "learning_rate": 2.3260316535645534e-07,
      "loss": 0.4076,
      "step": 29681
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5490243434906006,
      "learning_rate": 2.323848397549544e-07,
      "loss": 0.3785,
      "step": 29682
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6708413362503052,
      "learning_rate": 2.3216661545924167e-07,
      "loss": 0.4115,
      "step": 29683
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5425995588302612,
      "learning_rate": 2.319484924715798e-07,
      "loss": 0.3888,
      "step": 29684
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5085079669952393,
      "learning_rate": 2.3173047079423583e-07,
      "loss": 0.3961,
      "step": 29685
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.533882975578308,
      "learning_rate": 2.3151255042946352e-07,
      "loss": 0.3779,
      "step": 29686
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8381686210632324,
      "learning_rate": 2.3129473137952773e-07,
      "loss": 0.4208,
      "step": 29687
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8080813884735107,
      "learning_rate": 2.3107701364668556e-07,
      "loss": 0.3978,
      "step": 29688
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4931780099868774,
      "learning_rate": 2.308593972331985e-07,
      "loss": 0.4316,
      "step": 29689
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6373748779296875,
      "learning_rate": 2.3064188214132033e-07,
      "loss": 0.3919,
      "step": 29690
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0018863677978516,
      "learning_rate": 2.304244683733059e-07,
      "loss": 0.4063,
      "step": 29691
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4701778888702393,
      "learning_rate": 2.3020715593141341e-07,
      "loss": 0.3756,
      "step": 29692
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5899235010147095,
      "learning_rate": 2.2998994481789548e-07,
      "loss": 0.4342,
      "step": 29693
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4894970655441284,
      "learning_rate": 2.2977283503500481e-07,
      "loss": 0.3582,
      "step": 29694
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5418165922164917,
      "learning_rate": 2.2955582658499399e-07,
      "loss": 0.4258,
      "step": 29695
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5685495138168335,
      "learning_rate": 2.293389194701112e-07,
      "loss": 0.3608,
      "step": 29696
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5028955936431885,
      "learning_rate": 2.2912211369260917e-07,
      "loss": 0.3903,
      "step": 29697
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5406501293182373,
      "learning_rate": 2.2890540925473491e-07,
      "loss": 0.3747,
      "step": 29698
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5438785552978516,
      "learning_rate": 2.2868880615873556e-07,
      "loss": 0.4748,
      "step": 29699
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4631723165512085,
      "learning_rate": 2.2847230440686042e-07,
      "loss": 0.3562,
      "step": 29700
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6393004655838013,
      "learning_rate": 2.282559040013521e-07,
      "loss": 0.4705,
      "step": 29701
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5147143602371216,
      "learning_rate": 2.2803960494445554e-07,
      "loss": 0.382,
      "step": 29702
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4875293970108032,
      "learning_rate": 2.2782340723841334e-07,
      "loss": 0.3862,
      "step": 29703
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.56170654296875,
      "learning_rate": 2.2760731088546928e-07,
      "loss": 0.3797,
      "step": 29704
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5321245193481445,
      "learning_rate": 2.27391315887866e-07,
      "loss": 0.3944,
      "step": 29705
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8228732347488403,
      "learning_rate": 2.2717542224784172e-07,
      "loss": 0.3444,
      "step": 29706
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5141772031784058,
      "learning_rate": 2.2695962996763466e-07,
      "loss": 0.3812,
      "step": 29707
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.429555058479309,
      "learning_rate": 2.2674393904948411e-07,
      "loss": 0.3659,
      "step": 29708
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8886635303497314,
      "learning_rate": 2.265283494956272e-07,
      "loss": 0.976,
      "step": 29709
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4396495819091797,
      "learning_rate": 2.2631286130829987e-07,
      "loss": 0.9882,
      "step": 29710
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6086443662643433,
      "learning_rate": 2.2609747448973819e-07,
      "loss": 0.4702,
      "step": 29711
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.524146318435669,
      "learning_rate": 2.2588218904217362e-07,
      "loss": 0.3645,
      "step": 29712
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.934215784072876,
      "learning_rate": 2.256670049678422e-07,
      "loss": 0.4088,
      "step": 29713
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4701074361801147,
      "learning_rate": 2.2545192226897217e-07,
      "loss": 0.3651,
      "step": 29714
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6109113693237305,
      "learning_rate": 2.2523694094779613e-07,
      "loss": 0.4696,
      "step": 29715
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5616458654403687,
      "learning_rate": 2.2502206100654343e-07,
      "loss": 0.3994,
      "step": 29716
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5447598695755005,
      "learning_rate": 2.2480728244744455e-07,
      "loss": 0.4296,
      "step": 29717
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5900895595550537,
      "learning_rate": 2.245926052727243e-07,
      "loss": 0.3942,
      "step": 29718
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6115024089813232,
      "learning_rate": 2.2437802948460986e-07,
      "loss": 0.4155,
      "step": 29719
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5371599197387695,
      "learning_rate": 2.2416355508532716e-07,
      "loss": 0.359,
      "step": 29720
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5436698198318481,
      "learning_rate": 2.2394918207710336e-07,
      "loss": 0.4471,
      "step": 29721
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5065091848373413,
      "learning_rate": 2.2373491046215666e-07,
      "loss": 0.3522,
      "step": 29722
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5135705471038818,
      "learning_rate": 2.23520740242712e-07,
      "loss": 0.4165,
      "step": 29723
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4850265979766846,
      "learning_rate": 2.2330667142099194e-07,
      "loss": 0.3507,
      "step": 29724
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.524996042251587,
      "learning_rate": 2.2309270399921478e-07,
      "loss": 0.4101,
      "step": 29725
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5676265954971313,
      "learning_rate": 2.2287883797960097e-07,
      "loss": 0.3624,
      "step": 29726
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5187358856201172,
      "learning_rate": 2.226650733643676e-07,
      "loss": 0.4055,
      "step": 29727
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.513427734375,
      "learning_rate": 2.2245141015573068e-07,
      "loss": 0.4021,
      "step": 29728
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.503912329673767,
      "learning_rate": 2.2223784835590955e-07,
      "loss": 0.3711,
      "step": 29729
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4045321941375732,
      "learning_rate": 2.220243879671169e-07,
      "loss": 0.3442,
      "step": 29730
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5543469190597534,
      "learning_rate": 2.218110289915676e-07,
      "loss": 0.3752,
      "step": 29731
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5246378183364868,
      "learning_rate": 2.215977714314732e-07,
      "loss": 0.3562,
      "step": 29732
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.514507532119751,
      "learning_rate": 2.2138461528904753e-07,
      "loss": 0.4088,
      "step": 29733
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4546698331832886,
      "learning_rate": 2.211715605664988e-07,
      "loss": 0.3789,
      "step": 29734
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5914084911346436,
      "learning_rate": 2.2095860726603856e-07,
      "loss": 0.4484,
      "step": 29735
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4947545528411865,
      "learning_rate": 2.2074575538987507e-07,
      "loss": 0.3515,
      "step": 29736
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6128016710281372,
      "learning_rate": 2.2053300494021657e-07,
      "loss": 0.4281,
      "step": 29737
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5455820560455322,
      "learning_rate": 2.2032035591926794e-07,
      "loss": 0.361,
      "step": 29738
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6985012292861938,
      "learning_rate": 2.2010780832923627e-07,
      "loss": 0.9683,
      "step": 29739
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.52305006980896,
      "learning_rate": 2.1989536217232543e-07,
      "loss": 1.0428,
      "step": 29740
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.509906530380249,
      "learning_rate": 2.1968301745073916e-07,
      "loss": 0.4212,
      "step": 29741
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4528011083602905,
      "learning_rate": 2.1947077416667906e-07,
      "loss": 0.3297,
      "step": 29742
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5942602157592773,
      "learning_rate": 2.1925863232234668e-07,
      "loss": 0.4197,
      "step": 29743
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6152516603469849,
      "learning_rate": 2.190465919199425e-07,
      "loss": 0.3747,
      "step": 29744
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5664230585098267,
      "learning_rate": 2.1883465296166694e-07,
      "loss": 0.4346,
      "step": 29745
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5671128034591675,
      "learning_rate": 2.1862281544971497e-07,
      "loss": 0.3806,
      "step": 29746
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4970338344573975,
      "learning_rate": 2.1841107938628814e-07,
      "loss": 0.402,
      "step": 29747
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.513128638267517,
      "learning_rate": 2.1819944477357801e-07,
      "loss": 0.3823,
      "step": 29748
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.61142897605896,
      "learning_rate": 2.1798791161378175e-07,
      "loss": 1.0111,
      "step": 29749
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4086508750915527,
      "learning_rate": 2.1777647990909311e-07,
      "loss": 0.9419,
      "step": 29750
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5832624435424805,
      "learning_rate": 2.1756514966170483e-07,
      "loss": 0.4171,
      "step": 29751
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.503757119178772,
      "learning_rate": 2.1735392087380957e-07,
      "loss": 0.3735,
      "step": 29752
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5700726509094238,
      "learning_rate": 2.1714279354759782e-07,
      "loss": 0.4793,
      "step": 29753
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5406603813171387,
      "learning_rate": 2.169317676852578e-07,
      "loss": 0.3746,
      "step": 29754
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.571484088897705,
      "learning_rate": 2.167208432889789e-07,
      "loss": 0.4177,
      "step": 29755
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5223569869995117,
      "learning_rate": 2.1651002036094937e-07,
      "loss": 0.3897,
      "step": 29756
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5381689071655273,
      "learning_rate": 2.1629929890335632e-07,
      "loss": 0.4319,
      "step": 29757
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5081449747085571,
      "learning_rate": 2.160886789183847e-07,
      "loss": 0.3667,
      "step": 29758
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5742202997207642,
      "learning_rate": 2.1587816040821717e-07,
      "loss": 0.4332,
      "step": 29759
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.531148910522461,
      "learning_rate": 2.1566774337503982e-07,
      "loss": 0.3725,
      "step": 29760
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6417149305343628,
      "learning_rate": 2.1545742782103418e-07,
      "loss": 0.4653,
      "step": 29761
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.558481216430664,
      "learning_rate": 2.1524721374838188e-07,
      "loss": 0.3598,
      "step": 29762
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.1715729236602783,
      "learning_rate": 2.1503710115926114e-07,
      "loss": 0.414,
      "step": 29763
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6215156316757202,
      "learning_rate": 2.1482709005585246e-07,
      "loss": 0.4081,
      "step": 29764
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5978021621704102,
      "learning_rate": 2.146171804403352e-07,
      "loss": 0.4195,
      "step": 29765
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5183483362197876,
      "learning_rate": 2.1440737231488428e-07,
      "loss": 0.3777,
      "step": 29766
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.505142331123352,
      "learning_rate": 2.1419766568167798e-07,
      "loss": 0.4241,
      "step": 29767
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.507213830947876,
      "learning_rate": 2.1398806054288901e-07,
      "loss": 0.3657,
      "step": 29768
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5687963962554932,
      "learning_rate": 2.1377855690069339e-07,
      "loss": 0.4421,
      "step": 29769
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4353570938110352,
      "learning_rate": 2.135691547572627e-07,
      "loss": 0.355,
      "step": 29770
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5506778955459595,
      "learning_rate": 2.1335985411476968e-07,
      "loss": 0.4201,
      "step": 29771
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.502094030380249,
      "learning_rate": 2.131506549753848e-07,
      "loss": 0.3708,
      "step": 29772
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6569875478744507,
      "learning_rate": 2.1294155734127742e-07,
      "loss": 0.4269,
      "step": 29773
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.515208125114441,
      "learning_rate": 2.1273256121461583e-07,
      "loss": 0.3833,
      "step": 29774
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.641793966293335,
      "learning_rate": 2.1252366659756717e-07,
      "loss": 0.4922,
      "step": 29775
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4744675159454346,
      "learning_rate": 2.1231487349229973e-07,
      "loss": 0.3553,
      "step": 29776
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7935012578964233,
      "learning_rate": 2.1210618190097952e-07,
      "loss": 0.3973,
      "step": 29777
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4701616764068604,
      "learning_rate": 2.1189759182577042e-07,
      "loss": 0.376,
      "step": 29778
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.001779794692993,
      "learning_rate": 2.1168910326883284e-07,
      "loss": 0.4705,
      "step": 29779
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.508223295211792,
      "learning_rate": 2.114807162323329e-07,
      "loss": 0.3792,
      "step": 29780
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5251357555389404,
      "learning_rate": 2.1127243071843107e-07,
      "loss": 0.4097,
      "step": 29781
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.776435136795044,
      "learning_rate": 2.110642467292856e-07,
      "loss": 0.3609,
      "step": 29782
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5340080261230469,
      "learning_rate": 2.1085616426705925e-07,
      "loss": 0.4051,
      "step": 29783
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5124378204345703,
      "learning_rate": 2.106481833339058e-07,
      "loss": 0.3569,
      "step": 29784
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6247762441635132,
      "learning_rate": 2.104403039319869e-07,
      "loss": 0.4439,
      "step": 29785
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.52217435836792,
      "learning_rate": 2.1023252606345524e-07,
      "loss": 0.3663,
      "step": 29786
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7026560306549072,
      "learning_rate": 2.1002484973046577e-07,
      "loss": 0.438,
      "step": 29787
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6437067985534668,
      "learning_rate": 2.0981727493517567e-07,
      "loss": 0.381,
      "step": 29788
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.544674277305603,
      "learning_rate": 2.0960980167973545e-07,
      "loss": 0.4589,
      "step": 29789
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5345191955566406,
      "learning_rate": 2.0940242996629556e-07,
      "loss": 0.3593,
      "step": 29790
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4859217405319214,
      "learning_rate": 2.091951597970099e-07,
      "loss": 0.407,
      "step": 29791
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4401155710220337,
      "learning_rate": 2.0898799117402558e-07,
      "loss": 0.3608,
      "step": 29792
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5644968748092651,
      "learning_rate": 2.0878092409949314e-07,
      "loss": 0.3923,
      "step": 29793
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6508630514144897,
      "learning_rate": 2.0857395857555974e-07,
      "loss": 0.4199,
      "step": 29794
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.561574101448059,
      "learning_rate": 2.0836709460437143e-07,
      "loss": 0.4031,
      "step": 29795
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5671764612197876,
      "learning_rate": 2.0816033218807318e-07,
      "loss": 0.3777,
      "step": 29796
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.603365182876587,
      "learning_rate": 2.0795367132881105e-07,
      "loss": 0.4664,
      "step": 29797
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4828630685806274,
      "learning_rate": 2.077471120287289e-07,
      "loss": 0.3418,
      "step": 29798
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.571766972541809,
      "learning_rate": 2.0754065428996607e-07,
      "loss": 0.4257,
      "step": 29799
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5223946571350098,
      "learning_rate": 2.0733429811466533e-07,
      "loss": 0.3778,
      "step": 29800
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5377482175827026,
      "learning_rate": 2.071280435049683e-07,
      "loss": 0.4245,
      "step": 29801
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.489778757095337,
      "learning_rate": 2.0692189046301214e-07,
      "loss": 0.3573,
      "step": 29802
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6272062063217163,
      "learning_rate": 2.0671583899093628e-07,
      "loss": 0.4221,
      "step": 29803
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5592619180679321,
      "learning_rate": 2.0650988909087677e-07,
      "loss": 0.4042,
      "step": 29804
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5348414182662964,
      "learning_rate": 2.063040407649719e-07,
      "loss": 0.3888,
      "step": 29805
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5705928802490234,
      "learning_rate": 2.060982940153533e-07,
      "loss": 0.4057,
      "step": 29806
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5817574262619019,
      "learning_rate": 2.0589264884415706e-07,
      "loss": 0.4675,
      "step": 29807
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5599321126937866,
      "learning_rate": 2.0568710525351698e-07,
      "loss": 0.3933,
      "step": 29808
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5987765789031982,
      "learning_rate": 2.054816632455625e-07,
      "loss": 0.4433,
      "step": 29809
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8837754726409912,
      "learning_rate": 2.0527632282242526e-07,
      "loss": 0.3674,
      "step": 29810
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4939814805984497,
      "learning_rate": 2.0507108398623575e-07,
      "loss": 0.4349,
      "step": 29811
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4947750568389893,
      "learning_rate": 2.0486594673912007e-07,
      "loss": 0.4061,
      "step": 29812
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5505446195602417,
      "learning_rate": 2.0466091108320985e-07,
      "loss": 0.3961,
      "step": 29813
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.558131456375122,
      "learning_rate": 2.0445597702063003e-07,
      "loss": 0.3823,
      "step": 29814
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5295569896697998,
      "learning_rate": 2.0425114455350335e-07,
      "loss": 0.4214,
      "step": 29815
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5623257160186768,
      "learning_rate": 2.0404641368395704e-07,
      "loss": 0.3727,
      "step": 29816
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5144964456558228,
      "learning_rate": 2.0384178441411495e-07,
      "loss": 0.414,
      "step": 29817
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4351887702941895,
      "learning_rate": 2.0363725674609758e-07,
      "loss": 0.3417,
      "step": 29818
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.583482027053833,
      "learning_rate": 2.034328306820288e-07,
      "loss": 0.45,
      "step": 29819
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.36375617980957,
      "learning_rate": 2.0322850622402468e-07,
      "loss": 0.3554,
      "step": 29820
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5659939050674438,
      "learning_rate": 2.0302428337420799e-07,
      "loss": 0.4221,
      "step": 29821
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5455025434494019,
      "learning_rate": 2.028201621346959e-07,
      "loss": 0.3718,
      "step": 29822
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6779310703277588,
      "learning_rate": 2.0261614250760342e-07,
      "loss": 0.9851,
      "step": 29823
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.429309606552124,
      "learning_rate": 2.0241222449505104e-07,
      "loss": 1.0131,
      "step": 29824
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6274443864822388,
      "learning_rate": 2.0220840809914931e-07,
      "loss": 0.42,
      "step": 29825
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4986916780471802,
      "learning_rate": 2.0200469332201433e-07,
      "loss": 0.3712,
      "step": 29826
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5756553411483765,
      "learning_rate": 2.0180108016575772e-07,
      "loss": 0.4488,
      "step": 29827
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.518738031387329,
      "learning_rate": 2.0159756863249226e-07,
      "loss": 0.3829,
      "step": 29828
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5199347734451294,
      "learning_rate": 2.0139415872432956e-07,
      "loss": 0.4109,
      "step": 29829
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.510990023612976,
      "learning_rate": 2.0119085044337794e-07,
      "loss": 0.3668,
      "step": 29830
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6213346719741821,
      "learning_rate": 2.0098764379174573e-07,
      "loss": 0.4015,
      "step": 29831
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5267877578735352,
      "learning_rate": 2.0078453877154013e-07,
      "loss": 0.3663,
      "step": 29832
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5770461559295654,
      "learning_rate": 2.0058153538487057e-07,
      "loss": 0.4181,
      "step": 29833
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4958549737930298,
      "learning_rate": 2.003786336338387e-07,
      "loss": 0.3742,
      "step": 29834
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.519850254058838,
      "learning_rate": 2.0017583352055281e-07,
      "loss": 0.4267,
      "step": 29835
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5563982725143433,
      "learning_rate": 1.9997313504711234e-07,
      "loss": 0.3803,
      "step": 29836
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6648772954940796,
      "learning_rate": 1.997705382156223e-07,
      "loss": 0.9878,
      "step": 29837
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4370416402816772,
      "learning_rate": 1.9956804302818322e-07,
      "loss": 0.9501,
      "step": 29838
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.547702670097351,
      "learning_rate": 1.9936564948689562e-07,
      "loss": 0.399,
      "step": 29839
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5813579559326172,
      "learning_rate": 1.9916335759385786e-07,
      "loss": 0.4028,
      "step": 29840
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8040632009506226,
      "learning_rate": 1.9896116735116821e-07,
      "loss": 0.4828,
      "step": 29841
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4783895015716553,
      "learning_rate": 1.9875907876092393e-07,
      "loss": 0.3705,
      "step": 29842
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.491777777671814,
      "learning_rate": 1.9855709182521998e-07,
      "loss": 0.3973,
      "step": 29843
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.983686923980713,
      "learning_rate": 1.983552065461547e-07,
      "loss": 0.3882,
      "step": 29844
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6502918004989624,
      "learning_rate": 1.9815342292581864e-07,
      "loss": 0.4115,
      "step": 29845
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5450401306152344,
      "learning_rate": 1.9795174096630454e-07,
      "loss": 0.3737,
      "step": 29846
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5336095094680786,
      "learning_rate": 1.977501606697052e-07,
      "loss": 0.438,
      "step": 29847
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.467300534248352,
      "learning_rate": 1.9754868203811116e-07,
      "loss": 0.3446,
      "step": 29848
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6499797105789185,
      "learning_rate": 1.973473050736141e-07,
      "loss": 0.447,
      "step": 29849
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5146892070770264,
      "learning_rate": 1.9714602977829899e-07,
      "loss": 0.352,
      "step": 29850
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5079950094223022,
      "learning_rate": 1.969448561542553e-07,
      "loss": 0.3966,
      "step": 29851
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4860990047454834,
      "learning_rate": 1.967437842035691e-07,
      "loss": 0.3638,
      "step": 29852
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.506186842918396,
      "learning_rate": 1.9654281392832765e-07,
      "loss": 0.3853,
      "step": 29853
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5120213031768799,
      "learning_rate": 1.9634194533061256e-07,
      "loss": 0.3706,
      "step": 29854
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.53987717628479,
      "learning_rate": 1.961411784125089e-07,
      "loss": 0.4141,
      "step": 29855
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5047606229782104,
      "learning_rate": 1.959405131760972e-07,
      "loss": 0.3621,
      "step": 29856
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.439784049987793,
      "learning_rate": 1.9573994962346133e-07,
      "loss": 0.4031,
      "step": 29857
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6577143669128418,
      "learning_rate": 1.9553948775667852e-07,
      "loss": 0.3904,
      "step": 29858
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5766503810882568,
      "learning_rate": 1.9533912757782937e-07,
      "loss": 0.3898,
      "step": 29859
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.508100986480713,
      "learning_rate": 1.9513886908899326e-07,
      "loss": 0.3663,
      "step": 29860
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4845362901687622,
      "learning_rate": 1.9493871229224636e-07,
      "loss": 0.371,
      "step": 29861
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4939920902252197,
      "learning_rate": 1.9473865718966257e-07,
      "loss": 0.3564,
      "step": 29862
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5700763463974,
      "learning_rate": 1.9453870378331796e-07,
      "loss": 0.4284,
      "step": 29863
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4672346115112305,
      "learning_rate": 1.9433885207528646e-07,
      "loss": 0.3573,
      "step": 29864
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5573638677597046,
      "learning_rate": 1.9413910206764309e-07,
      "loss": 0.4204,
      "step": 29865
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.594323754310608,
      "learning_rate": 1.9393945376245727e-07,
      "loss": 0.3884,
      "step": 29866
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.579270601272583,
      "learning_rate": 1.9373990716179848e-07,
      "loss": 0.4246,
      "step": 29867
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5016297101974487,
      "learning_rate": 1.9354046226773836e-07,
      "loss": 0.3681,
      "step": 29868
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5434216260910034,
      "learning_rate": 1.9334111908234642e-07,
      "loss": 0.3881,
      "step": 29869
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5239131450653076,
      "learning_rate": 1.9314187760768656e-07,
      "loss": 0.3585,
      "step": 29870
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6009881496429443,
      "learning_rate": 1.929427378458293e-07,
      "loss": 0.4574,
      "step": 29871
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5550436973571777,
      "learning_rate": 1.9274369979883745e-07,
      "loss": 0.4054,
      "step": 29872
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6245825290679932,
      "learning_rate": 1.9254476346877603e-07,
      "loss": 0.4899,
      "step": 29873
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6633636951446533,
      "learning_rate": 1.9234592885770786e-07,
      "loss": 0.3675,
      "step": 29874
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.474989891052246,
      "learning_rate": 1.921471959676957e-07,
      "loss": 0.3799,
      "step": 29875
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.487276315689087,
      "learning_rate": 1.9194856480080125e-07,
      "loss": 0.3496,
      "step": 29876
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5288584232330322,
      "learning_rate": 1.9175003535908286e-07,
      "loss": 0.4263,
      "step": 29877
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.53062903881073,
      "learning_rate": 1.9155160764460223e-07,
      "loss": 0.3745,
      "step": 29878
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.656440258026123,
      "learning_rate": 1.9135328165941546e-07,
      "loss": 0.4115,
      "step": 29879
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4823602437973022,
      "learning_rate": 1.9115505740557982e-07,
      "loss": 0.3658,
      "step": 29880
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7303149700164795,
      "learning_rate": 1.9095693488515145e-07,
      "loss": 0.4894,
      "step": 29881
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6026262044906616,
      "learning_rate": 1.907589141001853e-07,
      "loss": 0.368,
      "step": 29882
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5139151811599731,
      "learning_rate": 1.9056099505273428e-07,
      "loss": 0.4477,
      "step": 29883
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4369494915008545,
      "learning_rate": 1.9036317774485224e-07,
      "loss": 0.3421,
      "step": 29884
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.638463020324707,
      "learning_rate": 1.9016546217859088e-07,
      "loss": 0.4582,
      "step": 29885
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4971823692321777,
      "learning_rate": 1.899678483560008e-07,
      "loss": 0.3935,
      "step": 29886
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5966496467590332,
      "learning_rate": 1.8977033627913032e-07,
      "loss": 0.4036,
      "step": 29887
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5719095468521118,
      "learning_rate": 1.8957292595002786e-07,
      "loss": 0.367,
      "step": 29888
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.517521619796753,
      "learning_rate": 1.8937561737074283e-07,
      "loss": 0.4226,
      "step": 29889
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.564397931098938,
      "learning_rate": 1.8917841054332143e-07,
      "loss": 0.3664,
      "step": 29890
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5661534070968628,
      "learning_rate": 1.8898130546980752e-07,
      "loss": 0.4545,
      "step": 29891
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5371934175491333,
      "learning_rate": 1.8878430215224508e-07,
      "loss": 0.3532,
      "step": 29892
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6234331130981445,
      "learning_rate": 1.885874005926791e-07,
      "loss": 0.4306,
      "step": 29893
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5010607242584229,
      "learning_rate": 1.883906007931524e-07,
      "loss": 0.3673,
      "step": 29894
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6389727592468262,
      "learning_rate": 1.8819390275570227e-07,
      "loss": 0.4249,
      "step": 29895
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5164481401443481,
      "learning_rate": 1.8799730648237368e-07,
      "loss": 0.3695,
      "step": 29896
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4754878282546997,
      "learning_rate": 1.8780081197520172e-07,
      "loss": 0.3831,
      "step": 29897
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4852206707000732,
      "learning_rate": 1.8760441923622584e-07,
      "loss": 0.3548,
      "step": 29898
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.55454421043396,
      "learning_rate": 1.8740812826748334e-07,
      "loss": 0.4285,
      "step": 29899
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.554438829421997,
      "learning_rate": 1.8721193907100922e-07,
      "loss": 0.3656,
      "step": 29900
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5729175806045532,
      "learning_rate": 1.8701585164883963e-07,
      "loss": 0.4108,
      "step": 29901
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5335063934326172,
      "learning_rate": 1.8681986600300738e-07,
      "loss": 0.3821,
      "step": 29902
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.518441915512085,
      "learning_rate": 1.8662398213554423e-07,
      "loss": 0.4113,
      "step": 29903
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5097543001174927,
      "learning_rate": 1.8642820004848294e-07,
      "loss": 0.3708,
      "step": 29904
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6391711235046387,
      "learning_rate": 1.8623251974385414e-07,
      "loss": 0.4471,
      "step": 29905
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.487240195274353,
      "learning_rate": 1.860369412236873e-07,
      "loss": 0.3558,
      "step": 29906
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5483695268630981,
      "learning_rate": 1.8584146449001083e-07,
      "loss": 0.434,
      "step": 29907
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4436208009719849,
      "learning_rate": 1.8564608954485198e-07,
      "loss": 0.34,
      "step": 29908
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5282810926437378,
      "learning_rate": 1.8545081639023687e-07,
      "loss": 0.4087,
      "step": 29909
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5211418867111206,
      "learning_rate": 1.852556450281917e-07,
      "loss": 0.3901,
      "step": 29910
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5703777074813843,
      "learning_rate": 1.8506057546073818e-07,
      "loss": 0.4298,
      "step": 29911
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.442623496055603,
      "learning_rate": 1.8486560768990358e-07,
      "loss": 0.3585,
      "step": 29912
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5359821319580078,
      "learning_rate": 1.8467074171770628e-07,
      "loss": 0.4652,
      "step": 29913
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.479865550994873,
      "learning_rate": 1.844759775461702e-07,
      "loss": 0.3653,
      "step": 29914
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.572984218597412,
      "learning_rate": 1.8428131517731373e-07,
      "loss": 0.404,
      "step": 29915
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5023430585861206,
      "learning_rate": 1.8408675461315527e-07,
      "loss": 0.348,
      "step": 29916
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5893925428390503,
      "learning_rate": 1.8389229585571432e-07,
      "loss": 0.4459,
      "step": 29917
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4591652154922485,
      "learning_rate": 1.8369793890700706e-07,
      "loss": 0.3718,
      "step": 29918
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5465188026428223,
      "learning_rate": 1.835036837690485e-07,
      "loss": 0.4716,
      "step": 29919
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5948306322097778,
      "learning_rate": 1.833095304438548e-07,
      "loss": 0.398,
      "step": 29920
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5734422206878662,
      "learning_rate": 1.8311547893343883e-07,
      "loss": 0.4139,
      "step": 29921
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.445482611656189,
      "learning_rate": 1.8292152923981455e-07,
      "loss": 0.3614,
      "step": 29922
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.511742115020752,
      "learning_rate": 1.827276813649903e-07,
      "loss": 0.3893,
      "step": 29923
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4990816116333008,
      "learning_rate": 1.8253393531097895e-07,
      "loss": 0.3647,
      "step": 29924
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.640764832496643,
      "learning_rate": 1.8234029107979e-07,
      "loss": 0.3859,
      "step": 29925
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.539279818534851,
      "learning_rate": 1.8214674867343186e-07,
      "loss": 0.3935,
      "step": 29926
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.532293677330017,
      "learning_rate": 1.8195330809391175e-07,
      "loss": 0.4244,
      "step": 29927
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8921149969100952,
      "learning_rate": 1.817599693432337e-07,
      "loss": 0.3762,
      "step": 29928
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6523696184158325,
      "learning_rate": 1.8156673242340605e-07,
      "loss": 0.497,
      "step": 29929
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1447455883026123,
      "learning_rate": 1.813735973364328e-07,
      "loss": 0.3812,
      "step": 29930
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5036113262176514,
      "learning_rate": 1.811805640843156e-07,
      "loss": 0.3939,
      "step": 29931
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.495490312576294,
      "learning_rate": 1.809876326690574e-07,
      "loss": 0.3628,
      "step": 29932
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.526418924331665,
      "learning_rate": 1.8079480309265762e-07,
      "loss": 0.4014,
      "step": 29933
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6230169534683228,
      "learning_rate": 1.8060207535711804e-07,
      "loss": 0.3877,
      "step": 29934
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5320343971252441,
      "learning_rate": 1.8040944946443706e-07,
      "loss": 0.3989,
      "step": 29935
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4672482013702393,
      "learning_rate": 1.8021692541661196e-07,
      "loss": 0.3638,
      "step": 29936
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8019968271255493,
      "learning_rate": 1.8002450321564114e-07,
      "loss": 0.4326,
      "step": 29937
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5921005010604858,
      "learning_rate": 1.7983218286351855e-07,
      "loss": 0.4008,
      "step": 29938
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4946526288986206,
      "learning_rate": 1.796399643622393e-07,
      "loss": 0.4008,
      "step": 29939
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5153478384017944,
      "learning_rate": 1.794478477137962e-07,
      "loss": 0.4014,
      "step": 29940
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.497619867324829,
      "learning_rate": 1.7925583292018322e-07,
      "loss": 0.3997,
      "step": 29941
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5463507175445557,
      "learning_rate": 1.7906391998339323e-07,
      "loss": 0.3774,
      "step": 29942
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4918794631958008,
      "learning_rate": 1.788721089054135e-07,
      "loss": 0.3876,
      "step": 29943
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4630967378616333,
      "learning_rate": 1.786803996882347e-07,
      "loss": 0.3557,
      "step": 29944
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5552347898483276,
      "learning_rate": 1.784887923338452e-07,
      "loss": 0.4413,
      "step": 29945
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4391181468963623,
      "learning_rate": 1.7829728684423343e-07,
      "loss": 0.3228,
      "step": 29946
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5724996328353882,
      "learning_rate": 1.7810588322138222e-07,
      "loss": 0.4599,
      "step": 29947
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4772779941558838,
      "learning_rate": 1.7791458146728113e-07,
      "loss": 0.3638,
      "step": 29948
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4994949102401733,
      "learning_rate": 1.7772338158391078e-07,
      "loss": 0.4153,
      "step": 29949
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4824708700180054,
      "learning_rate": 1.7753228357325737e-07,
      "loss": 0.3785,
      "step": 29950
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5508949756622314,
      "learning_rate": 1.7734128743729927e-07,
      "loss": 0.4705,
      "step": 29951
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4741013050079346,
      "learning_rate": 1.771503931780194e-07,
      "loss": 0.3611,
      "step": 29952
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5864510536193848,
      "learning_rate": 1.7695960079739837e-07,
      "loss": 0.4045,
      "step": 29953
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6176741123199463,
      "learning_rate": 1.767689102974146e-07,
      "loss": 0.4186,
      "step": 29954
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.591719627380371,
      "learning_rate": 1.765783216800443e-07,
      "loss": 0.4266,
      "step": 29955
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.535176396369934,
      "learning_rate": 1.7638783494726475e-07,
      "loss": 0.3819,
      "step": 29956
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6109840869903564,
      "learning_rate": 1.7619745010105327e-07,
      "loss": 0.41,
      "step": 29957
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.537969708442688,
      "learning_rate": 1.7600716714338385e-07,
      "loss": 0.3789,
      "step": 29958
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4940060377120972,
      "learning_rate": 1.7581698607622822e-07,
      "loss": 0.3944,
      "step": 29959
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4955687522888184,
      "learning_rate": 1.7562690690155926e-07,
      "loss": 0.3864,
      "step": 29960
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5958795547485352,
      "learning_rate": 1.7543692962134985e-07,
      "loss": 0.4117,
      "step": 29961
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5156793594360352,
      "learning_rate": 1.752470542375706e-07,
      "loss": 0.3747,
      "step": 29962
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4791440963745117,
      "learning_rate": 1.7505728075218997e-07,
      "loss": 0.4067,
      "step": 29963
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5113595724105835,
      "learning_rate": 1.748676091671764e-07,
      "loss": 0.3632,
      "step": 29964
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.546927809715271,
      "learning_rate": 1.7467803948449602e-07,
      "loss": 0.412,
      "step": 29965
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5746898651123047,
      "learning_rate": 1.744885717061162e-07,
      "loss": 0.385,
      "step": 29966
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.602105975151062,
      "learning_rate": 1.7429920583400094e-07,
      "loss": 0.4106,
      "step": 29967
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4490920305252075,
      "learning_rate": 1.7410994187011643e-07,
      "loss": 0.3482,
      "step": 29968
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5258543491363525,
      "learning_rate": 1.739207798164233e-07,
      "loss": 0.4362,
      "step": 29969
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.470967411994934,
      "learning_rate": 1.7373171967488557e-07,
      "loss": 0.3532,
      "step": 29970
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6214982271194458,
      "learning_rate": 1.7354276144746052e-07,
      "loss": 0.4399,
      "step": 29971
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5796858072280884,
      "learning_rate": 1.7335390513611216e-07,
      "loss": 0.3553,
      "step": 29972
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5970556735992432,
      "learning_rate": 1.7316515074279672e-07,
      "loss": 0.4075,
      "step": 29973
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4549839496612549,
      "learning_rate": 1.7297649826947372e-07,
      "loss": 0.3555,
      "step": 29974
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4983493089675903,
      "learning_rate": 1.7278794771809716e-07,
      "loss": 0.4085,
      "step": 29975
    },
    {
      "epoch": 0.94,
      "grad_norm": 3.2536041736602783,
      "learning_rate": 1.7259949909062435e-07,
      "loss": 0.3802,
      "step": 29976
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7730419635772705,
      "learning_rate": 1.724111523890104e-07,
      "loss": 0.4017,
      "step": 29977
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5055798292160034,
      "learning_rate": 1.722229076152071e-07,
      "loss": 0.3808,
      "step": 29978
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5081919431686401,
      "learning_rate": 1.7203476477116842e-07,
      "loss": 0.4217,
      "step": 29979
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6306439638137817,
      "learning_rate": 1.7184672385884394e-07,
      "loss": 0.3707,
      "step": 29980
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.520774245262146,
      "learning_rate": 1.716587848801854e-07,
      "loss": 0.3839,
      "step": 29981
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.879717469215393,
      "learning_rate": 1.7147094783714236e-07,
      "loss": 0.4383,
      "step": 29982
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.919981837272644,
      "learning_rate": 1.7128321273166104e-07,
      "loss": 0.433,
      "step": 29983
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5600782632827759,
      "learning_rate": 1.71095579565691e-07,
      "loss": 0.367,
      "step": 29984
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5994056463241577,
      "learning_rate": 1.7090804834117513e-07,
      "loss": 0.4397,
      "step": 29985
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4925771951675415,
      "learning_rate": 1.7072061906006187e-07,
      "loss": 0.3943,
      "step": 29986
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.553472638130188,
      "learning_rate": 1.70533291724293e-07,
      "loss": 0.4168,
      "step": 29987
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5028847455978394,
      "learning_rate": 1.7034606633581252e-07,
      "loss": 0.3639,
      "step": 29988
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5358664989471436,
      "learning_rate": 1.701589428965611e-07,
      "loss": 0.4088,
      "step": 29989
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6360282897949219,
      "learning_rate": 1.6997192140848163e-07,
      "loss": 0.4007,
      "step": 29990
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5451921224594116,
      "learning_rate": 1.6978500187351034e-07,
      "loss": 0.3886,
      "step": 29991
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5069605112075806,
      "learning_rate": 1.6959818429358788e-07,
      "loss": 0.3654,
      "step": 29992
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.569097876548767,
      "learning_rate": 1.6941146867065383e-07,
      "loss": 0.4409,
      "step": 29993
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4733697175979614,
      "learning_rate": 1.692248550066422e-07,
      "loss": 0.3607,
      "step": 29994
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4670313596725464,
      "learning_rate": 1.6903834330348813e-07,
      "loss": 0.3633,
      "step": 29995
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.556190013885498,
      "learning_rate": 1.6885193356312667e-07,
      "loss": 0.4063,
      "step": 29996
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4846128225326538,
      "learning_rate": 1.686656257874919e-07,
      "loss": 0.3844,
      "step": 29997
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5957342386245728,
      "learning_rate": 1.684794199785156e-07,
      "loss": 0.3745,
      "step": 29998
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5783002376556396,
      "learning_rate": 1.682933161381295e-07,
      "loss": 0.4327,
      "step": 29999
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4940046072006226,
      "learning_rate": 1.681073142682632e-07,
      "loss": 0.3691,
      "step": 30000
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5756053924560547,
      "learning_rate": 1.679214143708463e-07,
      "loss": 0.4431,
      "step": 30001
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4492077827453613,
      "learning_rate": 1.6773561644780612e-07,
      "loss": 0.3626,
      "step": 30002
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.481514573097229,
      "learning_rate": 1.6754992050107e-07,
      "loss": 0.4043,
      "step": 30003
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.016007900238037,
      "learning_rate": 1.673643265325653e-07,
      "loss": 0.3568,
      "step": 30004
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5844206809997559,
      "learning_rate": 1.671788345442149e-07,
      "loss": 0.4198,
      "step": 30005
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.626570701599121,
      "learning_rate": 1.6699344453794397e-07,
      "loss": 0.4137,
      "step": 30006
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.625385046005249,
      "learning_rate": 1.6680815651567427e-07,
      "loss": 0.4429,
      "step": 30007
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5605602264404297,
      "learning_rate": 1.6662297047932763e-07,
      "loss": 0.4058,
      "step": 30008
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7134867906570435,
      "learning_rate": 1.6643788643082582e-07,
      "loss": 0.4011,
      "step": 30009
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5639901161193848,
      "learning_rate": 1.6625290437208841e-07,
      "loss": 0.3968,
      "step": 30010
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5006784200668335,
      "learning_rate": 1.6606802430503166e-07,
      "loss": 0.415,
      "step": 30011
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5108208656311035,
      "learning_rate": 1.6588324623157515e-07,
      "loss": 0.3673,
      "step": 30012
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6239603757858276,
      "learning_rate": 1.656985701536351e-07,
      "loss": 0.4815,
      "step": 30013
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5677529573440552,
      "learning_rate": 1.655139960731278e-07,
      "loss": 0.3805,
      "step": 30014
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.517378330230713,
      "learning_rate": 1.653295239919661e-07,
      "loss": 0.411,
      "step": 30015
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5944218635559082,
      "learning_rate": 1.65145153912063e-07,
      "loss": 0.4155,
      "step": 30016
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5794849395751953,
      "learning_rate": 1.6496088583533132e-07,
      "loss": 0.4204,
      "step": 30017
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4974281787872314,
      "learning_rate": 1.6477671976368404e-07,
      "loss": 0.3453,
      "step": 30018
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5472599267959595,
      "learning_rate": 1.6459265569902738e-07,
      "loss": 0.4026,
      "step": 30019
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4643168449401855,
      "learning_rate": 1.6440869364327316e-07,
      "loss": 0.3563,
      "step": 30020
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.500335693359375,
      "learning_rate": 1.6422483359832875e-07,
      "loss": 0.4188,
      "step": 30021
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4907655715942383,
      "learning_rate": 1.640410755661015e-07,
      "loss": 0.3726,
      "step": 30022
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5711356401443481,
      "learning_rate": 1.6385741954849655e-07,
      "loss": 0.5072,
      "step": 30023
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4761964082717896,
      "learning_rate": 1.6367386554741794e-07,
      "loss": 0.3589,
      "step": 30024
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5402220487594604,
      "learning_rate": 1.634904135647719e-07,
      "loss": 0.3996,
      "step": 30025
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4567757844924927,
      "learning_rate": 1.6330706360245918e-07,
      "loss": 0.357,
      "step": 30026
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5934630632400513,
      "learning_rate": 1.6312381566238155e-07,
      "loss": 0.4076,
      "step": 30027
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5184874534606934,
      "learning_rate": 1.6294066974643973e-07,
      "loss": 0.3546,
      "step": 30028
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5448068380355835,
      "learning_rate": 1.6275762585653332e-07,
      "loss": 0.4394,
      "step": 30029
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.581452488899231,
      "learning_rate": 1.625746839945619e-07,
      "loss": 0.4059,
      "step": 30030
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6364803314208984,
      "learning_rate": 1.623918441624217e-07,
      "loss": 0.432,
      "step": 30031
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.505040168762207,
      "learning_rate": 1.6220910636200794e-07,
      "loss": 0.3696,
      "step": 30032
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6766737699508667,
      "learning_rate": 1.6202647059521792e-07,
      "loss": 0.4639,
      "step": 30033
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5842444896697998,
      "learning_rate": 1.6184393686394572e-07,
      "loss": 0.384,
      "step": 30034
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4909588098526,
      "learning_rate": 1.6166150517008317e-07,
      "loss": 0.3992,
      "step": 30035
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5313410758972168,
      "learning_rate": 1.6147917551552316e-07,
      "loss": 0.3652,
      "step": 30036
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5456202030181885,
      "learning_rate": 1.6129694790215643e-07,
      "loss": 0.3907,
      "step": 30037
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5494663715362549,
      "learning_rate": 1.611148223318737e-07,
      "loss": 0.403,
      "step": 30038
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5439434051513672,
      "learning_rate": 1.6093279880656342e-07,
      "loss": 0.3949,
      "step": 30039
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6730269193649292,
      "learning_rate": 1.607508773281141e-07,
      "loss": 0.388,
      "step": 30040
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5466755628585815,
      "learning_rate": 1.6056905789841093e-07,
      "loss": 0.4272,
      "step": 30041
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4673607349395752,
      "learning_rate": 1.6038734051934124e-07,
      "loss": 0.3631,
      "step": 30042
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.561132550239563,
      "learning_rate": 1.602057251927891e-07,
      "loss": 0.4547,
      "step": 30043
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4800944328308105,
      "learning_rate": 1.6002421192063744e-07,
      "loss": 0.3661,
      "step": 30044
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.53047776222229,
      "learning_rate": 1.5984280070477032e-07,
      "loss": 0.4276,
      "step": 30045
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6099255084991455,
      "learning_rate": 1.5966149154706957e-07,
      "loss": 0.3944,
      "step": 30046
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5293599367141724,
      "learning_rate": 1.5948028444941254e-07,
      "loss": 0.3982,
      "step": 30047
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5166120529174805,
      "learning_rate": 1.5929917941368223e-07,
      "loss": 0.3756,
      "step": 30048
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5892066955566406,
      "learning_rate": 1.5911817644175377e-07,
      "loss": 0.4102,
      "step": 30049
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5192123651504517,
      "learning_rate": 1.5893727553550787e-07,
      "loss": 0.3784,
      "step": 30050
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5541216135025024,
      "learning_rate": 1.587564766968197e-07,
      "loss": 0.4129,
      "step": 30051
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5576119422912598,
      "learning_rate": 1.5857577992756224e-07,
      "loss": 0.4116,
      "step": 30052
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6155803203582764,
      "learning_rate": 1.5839518522961173e-07,
      "loss": 0.4496,
      "step": 30053
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.474357008934021,
      "learning_rate": 1.5821469260484112e-07,
      "loss": 0.3462,
      "step": 30054
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.518344521522522,
      "learning_rate": 1.5803430205512115e-07,
      "loss": 0.4159,
      "step": 30055
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5351853370666504,
      "learning_rate": 1.5785401358232366e-07,
      "loss": 0.3521,
      "step": 30056
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6636053323745728,
      "learning_rate": 1.5767382718831937e-07,
      "loss": 0.9928,
      "step": 30057
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3810691833496094,
      "learning_rate": 1.5749374287497565e-07,
      "loss": 0.9704,
      "step": 30058
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5867565870285034,
      "learning_rate": 1.5731376064415993e-07,
      "loss": 0.4739,
      "step": 30059
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4515835046768188,
      "learning_rate": 1.571338804977407e-07,
      "loss": 0.375,
      "step": 30060
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.555092453956604,
      "learning_rate": 1.5695410243758203e-07,
      "loss": 0.4391,
      "step": 30061
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5227956771850586,
      "learning_rate": 1.567744264655502e-07,
      "loss": 0.3591,
      "step": 30062
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7651575803756714,
      "learning_rate": 1.5659485258350705e-07,
      "loss": 0.4013,
      "step": 30063
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5536108016967773,
      "learning_rate": 1.5641538079331554e-07,
      "loss": 0.3533,
      "step": 30064
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6160165071487427,
      "learning_rate": 1.562360110968364e-07,
      "loss": 0.3988,
      "step": 30065
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.477805495262146,
      "learning_rate": 1.560567434959337e-07,
      "loss": 0.3594,
      "step": 30066
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6082597970962524,
      "learning_rate": 1.558775779924615e-07,
      "loss": 0.4568,
      "step": 30067
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5434495210647583,
      "learning_rate": 1.5569851458828056e-07,
      "loss": 0.3691,
      "step": 30068
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.412158489227295,
      "learning_rate": 1.5551955328524716e-07,
      "loss": 0.39,
      "step": 30069
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.531672716140747,
      "learning_rate": 1.553406940852198e-07,
      "loss": 0.3681,
      "step": 30070
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5055292844772339,
      "learning_rate": 1.5516193699005145e-07,
      "loss": 0.3915,
      "step": 30071
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4679516553878784,
      "learning_rate": 1.5498328200159507e-07,
      "loss": 0.3493,
      "step": 30072
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4734086990356445,
      "learning_rate": 1.5480472912170586e-07,
      "loss": 0.4277,
      "step": 30073
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5176151990890503,
      "learning_rate": 1.5462627835223566e-07,
      "loss": 0.3773,
      "step": 30074
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5490412712097168,
      "learning_rate": 1.544479296950341e-07,
      "loss": 0.4259,
      "step": 30075
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4839550256729126,
      "learning_rate": 1.5426968315195078e-07,
      "loss": 0.3599,
      "step": 30076
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4673936367034912,
      "learning_rate": 1.540915387248354e-07,
      "loss": 0.4054,
      "step": 30077
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6351401805877686,
      "learning_rate": 1.5391349641553531e-07,
      "loss": 0.3553,
      "step": 30078
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.461214303970337,
      "learning_rate": 1.5373555622589686e-07,
      "loss": 0.4061,
      "step": 30079
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5580220222473145,
      "learning_rate": 1.5355771815776522e-07,
      "loss": 0.388,
      "step": 30080
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5415149927139282,
      "learning_rate": 1.5337998221298667e-07,
      "loss": 0.4517,
      "step": 30081
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.466684103012085,
      "learning_rate": 1.5320234839340198e-07,
      "loss": 0.3779,
      "step": 30082
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4799984693527222,
      "learning_rate": 1.5302481670085523e-07,
      "loss": 0.3981,
      "step": 30083
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.709694504737854,
      "learning_rate": 1.5284738713718606e-07,
      "loss": 0.3916,
      "step": 30084
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6794565916061401,
      "learning_rate": 1.5267005970423633e-07,
      "loss": 0.4465,
      "step": 30085
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4790663719177246,
      "learning_rate": 1.5249283440384677e-07,
      "loss": 0.3711,
      "step": 30086
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.557226538658142,
      "learning_rate": 1.523157112378515e-07,
      "loss": 0.4233,
      "step": 30087
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4617600440979004,
      "learning_rate": 1.5213869020809013e-07,
      "loss": 0.3622,
      "step": 30088
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4850037097930908,
      "learning_rate": 1.519617713163979e-07,
      "loss": 0.3836,
      "step": 30089
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5201969146728516,
      "learning_rate": 1.5178495456460995e-07,
      "loss": 0.3914,
      "step": 30090
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5670331716537476,
      "learning_rate": 1.5160823995456043e-07,
      "loss": 0.4064,
      "step": 30091
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4910496473312378,
      "learning_rate": 1.5143162748808116e-07,
      "loss": 0.355,
      "step": 30092
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5653337240219116,
      "learning_rate": 1.5125511716700513e-07,
      "loss": 0.3917,
      "step": 30093
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.490973949432373,
      "learning_rate": 1.5107870899316312e-07,
      "loss": 0.3666,
      "step": 30094
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5131295919418335,
      "learning_rate": 1.5090240296838255e-07,
      "loss": 0.3993,
      "step": 30095
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.56929612159729,
      "learning_rate": 1.5072619909449415e-07,
      "loss": 0.3966,
      "step": 30096
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6021665334701538,
      "learning_rate": 1.5055009737332537e-07,
      "loss": 0.4518,
      "step": 30097
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5310038328170776,
      "learning_rate": 1.5037409780670253e-07,
      "loss": 0.383,
      "step": 30098
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6746684312820435,
      "learning_rate": 1.501982003964497e-07,
      "loss": 0.4158,
      "step": 30099
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6784756183624268,
      "learning_rate": 1.5002240514439215e-07,
      "loss": 0.3713,
      "step": 30100
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.660294532775879,
      "learning_rate": 1.498467120523528e-07,
      "loss": 0.4292,
      "step": 30101
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5240931510925293,
      "learning_rate": 1.4967112112215576e-07,
      "loss": 0.3714,
      "step": 30102
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.652627944946289,
      "learning_rate": 1.494956323556196e-07,
      "loss": 0.4729,
      "step": 30103
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4787378311157227,
      "learning_rate": 1.4932024575456504e-07,
      "loss": 0.3546,
      "step": 30104
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5688670873641968,
      "learning_rate": 1.491449613208118e-07,
      "loss": 0.4139,
      "step": 30105
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4300059080123901,
      "learning_rate": 1.489697790561795e-07,
      "loss": 0.3491,
      "step": 30106
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.57837975025177,
      "learning_rate": 1.4879469896248112e-07,
      "loss": 0.4039,
      "step": 30107
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5089093446731567,
      "learning_rate": 1.4861972104153522e-07,
      "loss": 0.3859,
      "step": 30108
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.514359712600708,
      "learning_rate": 1.484448452951559e-07,
      "loss": 0.3982,
      "step": 30109
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5412156581878662,
      "learning_rate": 1.4827007172515728e-07,
      "loss": 0.3716,
      "step": 30110
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4945813417434692,
      "learning_rate": 1.4809540033335124e-07,
      "loss": 0.3811,
      "step": 30111
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.458800196647644,
      "learning_rate": 1.4792083112155075e-07,
      "loss": 0.3723,
      "step": 30112
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5381009578704834,
      "learning_rate": 1.477463640915644e-07,
      "loss": 0.408,
      "step": 30113
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5234508514404297,
      "learning_rate": 1.4757199924520292e-07,
      "loss": 0.3781,
      "step": 30114
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5539827346801758,
      "learning_rate": 1.473977365842749e-07,
      "loss": 0.4533,
      "step": 30115
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5554441213607788,
      "learning_rate": 1.472235761105878e-07,
      "loss": 0.4022,
      "step": 30116
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5125670433044434,
      "learning_rate": 1.470495178259479e-07,
      "loss": 0.3821,
      "step": 30117
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.528229832649231,
      "learning_rate": 1.4687556173215934e-07,
      "loss": 0.3691,
      "step": 30118
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5469154119491577,
      "learning_rate": 1.4670170783102734e-07,
      "loss": 0.4285,
      "step": 30119
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5057852268218994,
      "learning_rate": 1.465279561243549e-07,
      "loss": 0.3666,
      "step": 30120
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6845613718032837,
      "learning_rate": 1.463543066139439e-07,
      "loss": 0.5123,
      "step": 30121
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4790672063827515,
      "learning_rate": 1.4618075930159625e-07,
      "loss": 0.3586,
      "step": 30122
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.591341495513916,
      "learning_rate": 1.4600731418911053e-07,
      "loss": 0.4308,
      "step": 30123
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4858404397964478,
      "learning_rate": 1.4583397127828636e-07,
      "loss": 0.361,
      "step": 30124
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5198991298675537,
      "learning_rate": 1.4566073057092122e-07,
      "loss": 0.4103,
      "step": 30125
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4662095308303833,
      "learning_rate": 1.4548759206881257e-07,
      "loss": 0.3663,
      "step": 30126
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5697354078292847,
      "learning_rate": 1.4531455577375563e-07,
      "loss": 0.441,
      "step": 30127
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4385143518447876,
      "learning_rate": 1.4514162168754563e-07,
      "loss": 0.3464,
      "step": 30128
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.593409538269043,
      "learning_rate": 1.4496878981197559e-07,
      "loss": 0.4386,
      "step": 30129
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4991284608840942,
      "learning_rate": 1.447960601488374e-07,
      "loss": 0.396,
      "step": 30130
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6273090839385986,
      "learning_rate": 1.4462343269992407e-07,
      "loss": 0.4734,
      "step": 30131
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.496024250984192,
      "learning_rate": 1.4445090746702528e-07,
      "loss": 0.3482,
      "step": 30132
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4757940769195557,
      "learning_rate": 1.4427848445193072e-07,
      "loss": 0.3806,
      "step": 30133
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.486513376235962,
      "learning_rate": 1.441061636564267e-07,
      "loss": 0.3719,
      "step": 30134
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.54503333568573,
      "learning_rate": 1.4393394508230408e-07,
      "loss": 0.4085,
      "step": 30135
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.531211018562317,
      "learning_rate": 1.4376182873134582e-07,
      "loss": 0.368,
      "step": 30136
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.556740641593933,
      "learning_rate": 1.435898146053394e-07,
      "loss": 0.4632,
      "step": 30137
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5926406383514404,
      "learning_rate": 1.4341790270606672e-07,
      "loss": 0.3952,
      "step": 30138
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.544447898864746,
      "learning_rate": 1.43246093035313e-07,
      "loss": 0.4549,
      "step": 30139
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.475537657737732,
      "learning_rate": 1.4307438559485797e-07,
      "loss": 0.3677,
      "step": 30140
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4734102487564087,
      "learning_rate": 1.4290278038648463e-07,
      "loss": 0.3971,
      "step": 30141
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6485882997512817,
      "learning_rate": 1.4273127741197047e-07,
      "loss": 0.4223,
      "step": 30142
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5329166650772095,
      "learning_rate": 1.4255987667309624e-07,
      "loss": 0.3962,
      "step": 30143
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4209507703781128,
      "learning_rate": 1.4238857817163942e-07,
      "loss": 0.3448,
      "step": 30144
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4450767040252686,
      "learning_rate": 1.4221738190937527e-07,
      "loss": 0.379,
      "step": 30145
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4967979192733765,
      "learning_rate": 1.4204628788808128e-07,
      "loss": 0.4088,
      "step": 30146
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6209083795547485,
      "learning_rate": 1.4187529610953045e-07,
      "loss": 0.4442,
      "step": 30147
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5928771495819092,
      "learning_rate": 1.417044065754969e-07,
      "loss": 0.3972,
      "step": 30148
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5405194759368896,
      "learning_rate": 1.4153361928775255e-07,
      "loss": 0.4135,
      "step": 30149
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4894596338272095,
      "learning_rate": 1.4136293424806823e-07,
      "loss": 0.3715,
      "step": 30150
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6705983877182007,
      "learning_rate": 1.4119235145821697e-07,
      "loss": 0.4196,
      "step": 30151
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5086005926132202,
      "learning_rate": 1.41021870919964e-07,
      "loss": 0.3517,
      "step": 30152
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.828320860862732,
      "learning_rate": 1.4085149263508125e-07,
      "loss": 0.9897,
      "step": 30153
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4673584699630737,
      "learning_rate": 1.4068121660533284e-07,
      "loss": 0.9884,
      "step": 30154
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5987493991851807,
      "learning_rate": 1.405110428324874e-07,
      "loss": 0.4353,
      "step": 30155
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6205605268478394,
      "learning_rate": 1.4034097131830794e-07,
      "loss": 0.4,
      "step": 30156
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4973410367965698,
      "learning_rate": 1.4017100206455857e-07,
      "loss": 0.4021,
      "step": 30157
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5719174146652222,
      "learning_rate": 1.4000113507300238e-07,
      "loss": 0.4102,
      "step": 30158
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5893478393554688,
      "learning_rate": 1.398313703454024e-07,
      "loss": 0.4003,
      "step": 30159
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4852449893951416,
      "learning_rate": 1.396617078835172e-07,
      "loss": 0.3805,
      "step": 30160
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.56486177444458,
      "learning_rate": 1.3949214768910868e-07,
      "loss": 0.4234,
      "step": 30161
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4837604761123657,
      "learning_rate": 1.3932268976393326e-07,
      "loss": 0.3635,
      "step": 30162
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6169888973236084,
      "learning_rate": 1.3915333410975064e-07,
      "loss": 0.4181,
      "step": 30163
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.558562159538269,
      "learning_rate": 1.3898408072831603e-07,
      "loss": 0.3834,
      "step": 30164
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4850492477416992,
      "learning_rate": 1.3881492962138366e-07,
      "loss": 0.391,
      "step": 30165
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.52797532081604,
      "learning_rate": 1.3864588079071094e-07,
      "loss": 0.3736,
      "step": 30166
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5706231594085693,
      "learning_rate": 1.3847693423804877e-07,
      "loss": 0.4341,
      "step": 30167
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5650324821472168,
      "learning_rate": 1.383080899651501e-07,
      "loss": 0.362,
      "step": 30168
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4779452085494995,
      "learning_rate": 1.3813934797376695e-07,
      "loss": 0.439,
      "step": 30169
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4386104345321655,
      "learning_rate": 1.3797070826564783e-07,
      "loss": 0.3603,
      "step": 30170
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4666149616241455,
      "learning_rate": 1.3780217084254366e-07,
      "loss": 0.3827,
      "step": 30171
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5000396966934204,
      "learning_rate": 1.3763373570620077e-07,
      "loss": 0.362,
      "step": 30172
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5240411758422852,
      "learning_rate": 1.374654028583655e-07,
      "loss": 0.4162,
      "step": 30173
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6679129600524902,
      "learning_rate": 1.3729717230078765e-07,
      "loss": 0.3704,
      "step": 30174
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4680448770523071,
      "learning_rate": 1.37129044035208e-07,
      "loss": 0.3852,
      "step": 30175
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.44343900680542,
      "learning_rate": 1.369610180633707e-07,
      "loss": 0.3653,
      "step": 30176
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6209334135055542,
      "learning_rate": 1.3679309438702105e-07,
      "loss": 0.4185,
      "step": 30177
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.4077985286712646,
      "learning_rate": 1.3662527300789763e-07,
      "loss": 0.3949,
      "step": 30178
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5086954832077026,
      "learning_rate": 1.3645755392774352e-07,
      "loss": 0.3808,
      "step": 30179
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4947590827941895,
      "learning_rate": 1.362899371482973e-07,
      "loss": 0.3483,
      "step": 30180
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4940837621688843,
      "learning_rate": 1.3612242267129648e-07,
      "loss": 0.3956,
      "step": 30181
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4756463766098022,
      "learning_rate": 1.359550104984797e-07,
      "loss": 0.3546,
      "step": 30182
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6116080284118652,
      "learning_rate": 1.357877006315833e-07,
      "loss": 0.4387,
      "step": 30183
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4758639335632324,
      "learning_rate": 1.356204930723426e-07,
      "loss": 0.3509,
      "step": 30184
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5863035917282104,
      "learning_rate": 1.3545338782248952e-07,
      "loss": 0.4567,
      "step": 30185
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6345794200897217,
      "learning_rate": 1.3528638488376046e-07,
      "loss": 0.38,
      "step": 30186
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5802093744277954,
      "learning_rate": 1.3511948425788513e-07,
      "loss": 0.4171,
      "step": 30187
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.518449306488037,
      "learning_rate": 1.349526859465955e-07,
      "loss": 0.3505,
      "step": 30188
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5587788820266724,
      "learning_rate": 1.347859899516224e-07,
      "loss": 0.4161,
      "step": 30189
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5428454875946045,
      "learning_rate": 1.3461939627469334e-07,
      "loss": 0.4083,
      "step": 30190
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5726789236068726,
      "learning_rate": 1.3445290491753694e-07,
      "loss": 0.4237,
      "step": 30191
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.587431788444519,
      "learning_rate": 1.3428651588187957e-07,
      "loss": 0.375,
      "step": 30192
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.551560878753662,
      "learning_rate": 1.3412022916944655e-07,
      "loss": 0.4293,
      "step": 30193
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.571934700012207,
      "learning_rate": 1.3395404478196428e-07,
      "loss": 0.3838,
      "step": 30194
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6251250505447388,
      "learning_rate": 1.337879627211547e-07,
      "loss": 0.4808,
      "step": 30195
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.571825385093689,
      "learning_rate": 1.3362198298874086e-07,
      "loss": 0.3653,
      "step": 30196
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.593894600868225,
      "learning_rate": 1.3345610558644362e-07,
      "loss": 0.3951,
      "step": 30197
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5864728689193726,
      "learning_rate": 1.3329033051598383e-07,
      "loss": 0.3788,
      "step": 30198
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4754998683929443,
      "learning_rate": 1.3312465777908235e-07,
      "loss": 0.3668,
      "step": 30199
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5586416721343994,
      "learning_rate": 1.3295908737745555e-07,
      "loss": 0.3748,
      "step": 30200
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.536466121673584,
      "learning_rate": 1.3279361931281987e-07,
      "loss": 0.3619,
      "step": 30201
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5417546033859253,
      "learning_rate": 1.3262825358689392e-07,
      "loss": 0.3568,
      "step": 30202
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5043052434921265,
      "learning_rate": 1.3246299020139185e-07,
      "loss": 0.3996,
      "step": 30203
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5441075563430786,
      "learning_rate": 1.322978291580268e-07,
      "loss": 0.3897,
      "step": 30204
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4850994348526,
      "learning_rate": 1.3213277045851402e-07,
      "loss": 0.3799,
      "step": 30205
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4670287370681763,
      "learning_rate": 1.3196781410456216e-07,
      "loss": 0.3532,
      "step": 30206
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5499716997146606,
      "learning_rate": 1.318029600978854e-07,
      "loss": 0.4528,
      "step": 30207
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6005637645721436,
      "learning_rate": 1.3163820844019126e-07,
      "loss": 0.389,
      "step": 30208
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5590293407440186,
      "learning_rate": 1.3147355913318838e-07,
      "loss": 0.4167,
      "step": 30209
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5453171730041504,
      "learning_rate": 1.3130901217858649e-07,
      "loss": 0.3973,
      "step": 30210
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4580758810043335,
      "learning_rate": 1.311445675780909e-07,
      "loss": 0.3777,
      "step": 30211
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4672484397888184,
      "learning_rate": 1.3098022533340693e-07,
      "loss": 0.3518,
      "step": 30212
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6428686380386353,
      "learning_rate": 1.3081598544623874e-07,
      "loss": 0.4291,
      "step": 30213
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5096356868743896,
      "learning_rate": 1.3065184791829055e-07,
      "loss": 0.3603,
      "step": 30214
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7669721841812134,
      "learning_rate": 1.3048781275126542e-07,
      "loss": 0.3687,
      "step": 30215
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4682698249816895,
      "learning_rate": 1.3032387994686423e-07,
      "loss": 0.3617,
      "step": 30216
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5551296472549438,
      "learning_rate": 1.3016004950678562e-07,
      "loss": 0.4369,
      "step": 30217
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.532691478729248,
      "learning_rate": 1.2999632143273044e-07,
      "loss": 0.38,
      "step": 30218
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.567599892616272,
      "learning_rate": 1.2983269572639624e-07,
      "loss": 0.3637,
      "step": 30219
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4880073070526123,
      "learning_rate": 1.296691723894805e-07,
      "loss": 0.3493,
      "step": 30220
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5159766674041748,
      "learning_rate": 1.295057514236786e-07,
      "loss": 0.4156,
      "step": 30221
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4788312911987305,
      "learning_rate": 1.2934243283068582e-07,
      "loss": 0.3658,
      "step": 30222
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4831533432006836,
      "learning_rate": 1.2917921661219636e-07,
      "loss": 0.3811,
      "step": 30223
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5203858613967896,
      "learning_rate": 1.2901610276990217e-07,
      "loss": 0.3528,
      "step": 30224
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5825773477554321,
      "learning_rate": 1.2885309130549528e-07,
      "loss": 0.4076,
      "step": 30225
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5313503742218018,
      "learning_rate": 1.2869018222066653e-07,
      "loss": 0.3733,
      "step": 30226
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5628553628921509,
      "learning_rate": 1.2852737551710682e-07,
      "loss": 0.4309,
      "step": 30227
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5144091844558716,
      "learning_rate": 1.2836467119650253e-07,
      "loss": 0.3878,
      "step": 30228
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5584977865219116,
      "learning_rate": 1.2820206926054234e-07,
      "loss": 0.4546,
      "step": 30229
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.547927975654602,
      "learning_rate": 1.2803956971091269e-07,
      "loss": 0.3653,
      "step": 30230
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8415896892547607,
      "learning_rate": 1.2787717254929887e-07,
      "loss": 0.4411,
      "step": 30231
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5829284191131592,
      "learning_rate": 1.27714877777384e-07,
      "loss": 0.3865,
      "step": 30232
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.569105863571167,
      "learning_rate": 1.2755268539685338e-07,
      "loss": 0.4432,
      "step": 30233
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0144569873809814,
      "learning_rate": 1.2739059540938793e-07,
      "loss": 0.3757,
      "step": 30234
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5826748609542847,
      "learning_rate": 1.2722860781666958e-07,
      "loss": 0.4432,
      "step": 30235
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5015698671340942,
      "learning_rate": 1.270667226203781e-07,
      "loss": 0.3607,
      "step": 30236
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5249511003494263,
      "learning_rate": 1.2690493982219109e-07,
      "loss": 0.4068,
      "step": 30237
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5395920276641846,
      "learning_rate": 1.267432594237883e-07,
      "loss": 0.3864,
      "step": 30238
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5552363395690918,
      "learning_rate": 1.2658168142684724e-07,
      "loss": 0.3957,
      "step": 30239
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4896502494812012,
      "learning_rate": 1.2642020583304104e-07,
      "loss": 0.3563,
      "step": 30240
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5539902448654175,
      "learning_rate": 1.2625883264404615e-07,
      "loss": 0.4084,
      "step": 30241
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5174901485443115,
      "learning_rate": 1.2609756186153676e-07,
      "loss": 0.3387,
      "step": 30242
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5569199323654175,
      "learning_rate": 1.259363934871849e-07,
      "loss": 0.4555,
      "step": 30243
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.524551510810852,
      "learning_rate": 1.257753275226614e-07,
      "loss": 0.3762,
      "step": 30244
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5714927911758423,
      "learning_rate": 1.2561436396963723e-07,
      "loss": 0.4583,
      "step": 30245
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.527361512184143,
      "learning_rate": 1.254535028297832e-07,
      "loss": 0.3822,
      "step": 30246
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5567306280136108,
      "learning_rate": 1.2529274410476578e-07,
      "loss": 0.4175,
      "step": 30247
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5473017692565918,
      "learning_rate": 1.251320877962525e-07,
      "loss": 0.3963,
      "step": 30248
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.565431833267212,
      "learning_rate": 1.24971533905911e-07,
      "loss": 0.4461,
      "step": 30249
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5142508745193481,
      "learning_rate": 1.248110824354054e-07,
      "loss": 0.3725,
      "step": 30250
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6071697473526,
      "learning_rate": 1.246507333864e-07,
      "loss": 0.3942,
      "step": 30251
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.493779182434082,
      "learning_rate": 1.2449048676055897e-07,
      "loss": 0.3667,
      "step": 30252
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6906406879425049,
      "learning_rate": 1.2433034255954213e-07,
      "loss": 0.9659,
      "step": 30253
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4789021015167236,
      "learning_rate": 1.2417030078501146e-07,
      "loss": 1.0654,
      "step": 30254
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7805126905441284,
      "learning_rate": 1.2401036143862788e-07,
      "loss": 0.9492,
      "step": 30255
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4436761140823364,
      "learning_rate": 1.2385052452204894e-07,
      "loss": 1.0164,
      "step": 30256
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5243122577667236,
      "learning_rate": 1.2369079003693328e-07,
      "loss": 0.4573,
      "step": 30257
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5314052104949951,
      "learning_rate": 1.235311579849363e-07,
      "loss": 0.3604,
      "step": 30258
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.575871229171753,
      "learning_rate": 1.2337162836771553e-07,
      "loss": 0.4095,
      "step": 30259
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4765998125076294,
      "learning_rate": 1.2321220118692412e-07,
      "loss": 0.3407,
      "step": 30260
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9183697700500488,
      "learning_rate": 1.2305287644421516e-07,
      "loss": 0.4011,
      "step": 30261
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4999533891677856,
      "learning_rate": 1.2289365414124288e-07,
      "loss": 0.3574,
      "step": 30262
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.544736385345459,
      "learning_rate": 1.227345342796582e-07,
      "loss": 0.3996,
      "step": 30263
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5212016105651855,
      "learning_rate": 1.225755168611098e-07,
      "loss": 0.352,
      "step": 30264
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6662737131118774,
      "learning_rate": 1.2241660188724746e-07,
      "loss": 0.4203,
      "step": 30265
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5189872980117798,
      "learning_rate": 1.2225778935972098e-07,
      "loss": 0.3399,
      "step": 30266
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8483405113220215,
      "learning_rate": 1.2209907928017794e-07,
      "loss": 0.4646,
      "step": 30267
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5949863195419312,
      "learning_rate": 1.2194047165026034e-07,
      "loss": 0.3654,
      "step": 30268
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4413291215896606,
      "learning_rate": 1.217819664716169e-07,
      "loss": 0.3965,
      "step": 30269
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5441522598266602,
      "learning_rate": 1.2162356374589068e-07,
      "loss": 0.3797,
      "step": 30270
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5118385553359985,
      "learning_rate": 1.2146526347472487e-07,
      "loss": 0.3842,
      "step": 30271
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5089738368988037,
      "learning_rate": 1.2130706565976147e-07,
      "loss": 0.3665,
      "step": 30272
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5477336645126343,
      "learning_rate": 1.2114897030263917e-07,
      "loss": 0.3998,
      "step": 30273
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5204333066940308,
      "learning_rate": 1.2099097740499887e-07,
      "loss": 0.3619,
      "step": 30274
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.771854281425476,
      "learning_rate": 1.208330869684804e-07,
      "loss": 1.0187,
      "step": 30275
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.398382306098938,
      "learning_rate": 1.206752989947202e-07,
      "loss": 1.001,
      "step": 30276
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7678223848342896,
      "learning_rate": 1.2051761348535585e-07,
      "loss": 0.9392,
      "step": 30277
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4791438579559326,
      "learning_rate": 1.203600304420205e-07,
      "loss": 0.9996,
      "step": 30278
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5715138912200928,
      "learning_rate": 1.202025498663506e-07,
      "loss": 0.3816,
      "step": 30279
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.496607780456543,
      "learning_rate": 1.2004517175997932e-07,
      "loss": 0.3805,
      "step": 30280
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5563782453536987,
      "learning_rate": 1.1988789612453755e-07,
      "loss": 0.4194,
      "step": 30281
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5714967250823975,
      "learning_rate": 1.1973072296165845e-07,
      "loss": 0.3827,
      "step": 30282
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5509737730026245,
      "learning_rate": 1.1957365227297068e-07,
      "loss": 0.4525,
      "step": 30283
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.56125009059906,
      "learning_rate": 1.1941668406010298e-07,
      "loss": 0.3976,
      "step": 30284
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.644171953201294,
      "learning_rate": 1.1925981832468403e-07,
      "loss": 0.4506,
      "step": 30285
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4468332529067993,
      "learning_rate": 1.1910305506834142e-07,
      "loss": 0.3564,
      "step": 30286
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5410319566726685,
      "learning_rate": 1.1894639429270049e-07,
      "loss": 0.4038,
      "step": 30287
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6357333660125732,
      "learning_rate": 1.1878983599938665e-07,
      "loss": 0.3975,
      "step": 30288
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5960144996643066,
      "learning_rate": 1.1863338019002302e-07,
      "loss": 0.4243,
      "step": 30289
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.52512526512146,
      "learning_rate": 1.1847702686623163e-07,
      "loss": 0.3723,
      "step": 30290
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6170966625213623,
      "learning_rate": 1.1832077602963455e-07,
      "loss": 0.4285,
      "step": 30291
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.486981987953186,
      "learning_rate": 1.1816462768185378e-07,
      "loss": 0.3594,
      "step": 30292
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9414584636688232,
      "learning_rate": 1.1800858182450692e-07,
      "loss": 0.451,
      "step": 30293
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5038321018218994,
      "learning_rate": 1.1785263845921269e-07,
      "loss": 0.3625,
      "step": 30294
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5659419298171997,
      "learning_rate": 1.1769679758759089e-07,
      "loss": 0.416,
      "step": 30295
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5040698051452637,
      "learning_rate": 1.1754105921125359e-07,
      "loss": 0.3759,
      "step": 30296
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6264292001724243,
      "learning_rate": 1.1738542333181945e-07,
      "loss": 0.4404,
      "step": 30297
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.590341567993164,
      "learning_rate": 1.1722988995090168e-07,
      "loss": 0.3938,
      "step": 30298
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6291873455047607,
      "learning_rate": 1.170744590701134e-07,
      "loss": 0.4128,
      "step": 30299
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.554649829864502,
      "learning_rate": 1.1691913069106553e-07,
      "loss": 0.4021,
      "step": 30300
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6050959825515747,
      "learning_rate": 1.1676390481537125e-07,
      "loss": 0.452,
      "step": 30301
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4574717283248901,
      "learning_rate": 1.1660878144463927e-07,
      "loss": 0.3323,
      "step": 30302
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.585969090461731,
      "learning_rate": 1.164537605804783e-07,
      "loss": 0.4585,
      "step": 30303
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4856175184249878,
      "learning_rate": 1.1629884222449594e-07,
      "loss": 0.3518,
      "step": 30304
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5259414911270142,
      "learning_rate": 1.161440263782998e-07,
      "loss": 0.373,
      "step": 30305
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.567476749420166,
      "learning_rate": 1.1598931304349525e-07,
      "loss": 0.3847,
      "step": 30306
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4942171573638916,
      "learning_rate": 1.158347022216877e-07,
      "loss": 0.4597,
      "step": 30307
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.515312671661377,
      "learning_rate": 1.1568019391447915e-07,
      "loss": 0.3621,
      "step": 30308
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6732264757156372,
      "learning_rate": 1.1552578812347281e-07,
      "loss": 0.425,
      "step": 30309
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.55875563621521,
      "learning_rate": 1.1537148485026961e-07,
      "loss": 0.4009,
      "step": 30310
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5487439632415771,
      "learning_rate": 1.1521728409647048e-07,
      "loss": 0.4285,
      "step": 30311
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6022189855575562,
      "learning_rate": 1.1506318586367526e-07,
      "loss": 0.3628,
      "step": 30312
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5325744152069092,
      "learning_rate": 1.1490919015348156e-07,
      "loss": 0.4346,
      "step": 30313
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5045288801193237,
      "learning_rate": 1.1475529696748589e-07,
      "loss": 0.3537,
      "step": 30314
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5523605346679688,
      "learning_rate": 1.1460150630728695e-07,
      "loss": 0.388,
      "step": 30315
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4153082370758057,
      "learning_rate": 1.144478181744757e-07,
      "loss": 0.408,
      "step": 30316
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5673940181732178,
      "learning_rate": 1.1429423257064975e-07,
      "loss": 0.4045,
      "step": 30317
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4564777612686157,
      "learning_rate": 1.1414074949740007e-07,
      "loss": 0.3535,
      "step": 30318
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.564570426940918,
      "learning_rate": 1.1398736895631978e-07,
      "loss": 0.4747,
      "step": 30319
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4986417293548584,
      "learning_rate": 1.1383409094899878e-07,
      "loss": 0.3659,
      "step": 30320
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5381501913070679,
      "learning_rate": 1.1368091547702575e-07,
      "loss": 0.4077,
      "step": 30321
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5398982763290405,
      "learning_rate": 1.1352784254199168e-07,
      "loss": 0.3883,
      "step": 30322
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5518290996551514,
      "learning_rate": 1.1337487214548304e-07,
      "loss": 0.4245,
      "step": 30323
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4946950674057007,
      "learning_rate": 1.1322200428908747e-07,
      "loss": 0.3515,
      "step": 30324
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5127248764038086,
      "learning_rate": 1.1306923897438816e-07,
      "loss": 0.4304,
      "step": 30325
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5241544246673584,
      "learning_rate": 1.1291657620297158e-07,
      "loss": 0.3471,
      "step": 30326
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8118066787719727,
      "learning_rate": 1.1276401597641984e-07,
      "loss": 0.9116,
      "step": 30327
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4086780548095703,
      "learning_rate": 1.126115582963161e-07,
      "loss": 0.9272,
      "step": 30328
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6730682849884033,
      "learning_rate": 1.124592031642413e-07,
      "loss": 0.4116,
      "step": 30329
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4750953912734985,
      "learning_rate": 1.1230695058177532e-07,
      "loss": 0.3834,
      "step": 30330
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.714512586593628,
      "learning_rate": 1.1215480055049799e-07,
      "loss": 0.4499,
      "step": 30331
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5074384212493896,
      "learning_rate": 1.1200275307198583e-07,
      "loss": 0.3699,
      "step": 30332
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8239046335220337,
      "learning_rate": 1.1185080814781756e-07,
      "loss": 1.0472,
      "step": 30333
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4330488443374634,
      "learning_rate": 1.1169896577956752e-07,
      "loss": 1.0004,
      "step": 30334
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4829310178756714,
      "learning_rate": 1.1154722596881329e-07,
      "loss": 0.388,
      "step": 30335
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5222713947296143,
      "learning_rate": 1.1139558871712475e-07,
      "loss": 0.3735,
      "step": 30336
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.771303653717041,
      "learning_rate": 1.112440540260773e-07,
      "loss": 0.4039,
      "step": 30337
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4975661039352417,
      "learning_rate": 1.110926218972419e-07,
      "loss": 0.3809,
      "step": 30338
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6018840074539185,
      "learning_rate": 1.1094129233219064e-07,
      "loss": 0.4446,
      "step": 30339
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4680776596069336,
      "learning_rate": 1.1079006533249004e-07,
      "loss": 0.3581,
      "step": 30340
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5805293321609497,
      "learning_rate": 1.1063894089970995e-07,
      "loss": 0.431,
      "step": 30341
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5015158653259277,
      "learning_rate": 1.10487919035418e-07,
      "loss": 0.343,
      "step": 30342
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4823116064071655,
      "learning_rate": 1.1033699974118072e-07,
      "loss": 0.4044,
      "step": 30343
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.53861665725708,
      "learning_rate": 1.1018618301856354e-07,
      "loss": 0.3572,
      "step": 30344
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5145063400268555,
      "learning_rate": 1.1003546886912964e-07,
      "loss": 0.3925,
      "step": 30345
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5335872173309326,
      "learning_rate": 1.0988485729444331e-07,
      "loss": 0.3566,
      "step": 30346
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5874887704849243,
      "learning_rate": 1.0973434829606556e-07,
      "loss": 0.4113,
      "step": 30347
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5486414432525635,
      "learning_rate": 1.0958394187555843e-07,
      "loss": 0.371,
      "step": 30348
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.553253412246704,
      "learning_rate": 1.0943363803448182e-07,
      "loss": 0.4355,
      "step": 30349
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.425581693649292,
      "learning_rate": 1.0928343677439335e-07,
      "loss": 0.3754,
      "step": 30350
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6541876792907715,
      "learning_rate": 1.0913333809685178e-07,
      "loss": 0.4806,
      "step": 30351
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.508512258529663,
      "learning_rate": 1.0898334200341365e-07,
      "loss": 0.3663,
      "step": 30352
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5806517601013184,
      "learning_rate": 1.0883344849563437e-07,
      "loss": 0.4934,
      "step": 30353
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4561551809310913,
      "learning_rate": 1.0868365757507049e-07,
      "loss": 0.3525,
      "step": 30354
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.349433422088623,
      "learning_rate": 1.0853396924327409e-07,
      "loss": 0.4313,
      "step": 30355
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4685488939285278,
      "learning_rate": 1.0838438350179615e-07,
      "loss": 0.3673,
      "step": 30356
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5568898916244507,
      "learning_rate": 1.0823490035218986e-07,
      "loss": 0.4194,
      "step": 30357
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5506261587142944,
      "learning_rate": 1.0808551979600623e-07,
      "loss": 0.3951,
      "step": 30358
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5339621305465698,
      "learning_rate": 1.07936241834794e-07,
      "loss": 0.3721,
      "step": 30359
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4930357933044434,
      "learning_rate": 1.0778706647010085e-07,
      "loss": 0.3734,
      "step": 30360
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.626530408859253,
      "learning_rate": 1.0763799370347328e-07,
      "loss": 0.442,
      "step": 30361
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.528328537940979,
      "learning_rate": 1.0748902353645895e-07,
      "loss": 0.3746,
      "step": 30362
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5958315134048462,
      "learning_rate": 1.0734015597060222e-07,
      "loss": 0.4554,
      "step": 30363
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5917541980743408,
      "learning_rate": 1.0719139100744846e-07,
      "loss": 0.3922,
      "step": 30364
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4737385511398315,
      "learning_rate": 1.0704272864853871e-07,
      "loss": 0.3747,
      "step": 30365
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5066663026809692,
      "learning_rate": 1.0689416889541504e-07,
      "loss": 0.3681,
      "step": 30366
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5942445993423462,
      "learning_rate": 1.0674571174961845e-07,
      "loss": 0.4099,
      "step": 30367
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4960243701934814,
      "learning_rate": 1.0659735721268993e-07,
      "loss": 0.3792,
      "step": 30368
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5024477243423462,
      "learning_rate": 1.0644910528616714e-07,
      "loss": 0.3804,
      "step": 30369
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5237252712249756,
      "learning_rate": 1.0630095597158774e-07,
      "loss": 0.3646,
      "step": 30370
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5405257940292358,
      "learning_rate": 1.0615290927048716e-07,
      "loss": 0.4069,
      "step": 30371
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5271459817886353,
      "learning_rate": 1.0600496518440306e-07,
      "loss": 0.3872,
      "step": 30372
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6431267261505127,
      "learning_rate": 1.0585712371486757e-07,
      "loss": 0.4829,
      "step": 30373
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.52471125125885,
      "learning_rate": 1.0570938486341609e-07,
      "loss": 0.3776,
      "step": 30374
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5557399988174438,
      "learning_rate": 1.0556174863158075e-07,
      "loss": 0.4177,
      "step": 30375
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6106677055358887,
      "learning_rate": 1.0541421502089144e-07,
      "loss": 0.3845,
      "step": 30376
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.514923095703125,
      "learning_rate": 1.0526678403287804e-07,
      "loss": 0.4052,
      "step": 30377
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5127477645874023,
      "learning_rate": 1.0511945566907156e-07,
      "loss": 0.3834,
      "step": 30378
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5244230031967163,
      "learning_rate": 1.0497222993099853e-07,
      "loss": 0.4233,
      "step": 30379
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5037254095077515,
      "learning_rate": 1.0482510682018776e-07,
      "loss": 0.3649,
      "step": 30380
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5505075454711914,
      "learning_rate": 1.0467808633816246e-07,
      "loss": 0.4089,
      "step": 30381
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.485716462135315,
      "learning_rate": 1.0453116848644918e-07,
      "loss": 0.3638,
      "step": 30382
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7534152269363403,
      "learning_rate": 1.0438435326657114e-07,
      "loss": 0.4435,
      "step": 30383
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4592701196670532,
      "learning_rate": 1.042376406800516e-07,
      "loss": 0.3407,
      "step": 30384
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5501110553741455,
      "learning_rate": 1.0409103072841153e-07,
      "loss": 0.4178,
      "step": 30385
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5721153020858765,
      "learning_rate": 1.0394452341317196e-07,
      "loss": 0.3769,
      "step": 30386
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5253005027770996,
      "learning_rate": 1.0379811873585166e-07,
      "loss": 0.3952,
      "step": 30387
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5297874212265015,
      "learning_rate": 1.0365181669797164e-07,
      "loss": 0.3789,
      "step": 30388
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5253621339797974,
      "learning_rate": 1.0350561730104513e-07,
      "loss": 0.3887,
      "step": 30389
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4583300352096558,
      "learning_rate": 1.0335952054659204e-07,
      "loss": 0.3668,
      "step": 30390
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5223937034606934,
      "learning_rate": 1.0321352643612559e-07,
      "loss": 0.4214,
      "step": 30391
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5319253206253052,
      "learning_rate": 1.0306763497116123e-07,
      "loss": 0.3629,
      "step": 30392
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5080662965774536,
      "learning_rate": 1.029218461532111e-07,
      "loss": 0.4144,
      "step": 30393
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4835115671157837,
      "learning_rate": 1.0277615998378732e-07,
      "loss": 0.3634,
      "step": 30394
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.609663724899292,
      "learning_rate": 1.02630576464402e-07,
      "loss": 0.4434,
      "step": 30395
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5902854204177856,
      "learning_rate": 1.0248509559656395e-07,
      "loss": 0.3776,
      "step": 30396
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4716883897781372,
      "learning_rate": 1.0233971738178195e-07,
      "loss": 0.3793,
      "step": 30397
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6747944355010986,
      "learning_rate": 1.0219444182156479e-07,
      "loss": 0.3914,
      "step": 30398
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5412198305130005,
      "learning_rate": 1.0204926891741795e-07,
      "loss": 0.4176,
      "step": 30399
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.486022710800171,
      "learning_rate": 1.019041986708491e-07,
      "loss": 0.356,
      "step": 30400
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4618700742721558,
      "learning_rate": 1.0175923108336039e-07,
      "loss": 0.4047,
      "step": 30401
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.47629714012146,
      "learning_rate": 1.0161436615645726e-07,
      "loss": 0.3754,
      "step": 30402
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5227667093276978,
      "learning_rate": 1.0146960389164073e-07,
      "loss": 0.4641,
      "step": 30403
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6292357444763184,
      "learning_rate": 1.0132494429041406e-07,
      "loss": 0.39,
      "step": 30404
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5104007720947266,
      "learning_rate": 1.0118038735427493e-07,
      "loss": 0.3901,
      "step": 30405
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5167245864868164,
      "learning_rate": 1.0103593308472548e-07,
      "loss": 0.3769,
      "step": 30406
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.55183744430542,
      "learning_rate": 1.0089158148326228e-07,
      "loss": 0.4611,
      "step": 30407
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5475142002105713,
      "learning_rate": 1.0074733255138303e-07,
      "loss": 0.3641,
      "step": 30408
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.71196448802948,
      "learning_rate": 1.0060318629058319e-07,
      "loss": 0.4301,
      "step": 30409
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5760798454284668,
      "learning_rate": 1.0045914270235824e-07,
      "loss": 0.3692,
      "step": 30410
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6251968145370483,
      "learning_rate": 1.0031520178820254e-07,
      "loss": 0.4622,
      "step": 30411
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5360119342803955,
      "learning_rate": 1.0017136354960933e-07,
      "loss": 0.3818,
      "step": 30412
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6305017471313477,
      "learning_rate": 1.0002762798806854e-07,
      "loss": 0.394,
      "step": 30413
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8652342557907104,
      "learning_rate": 9.98839951050723e-08,
      "loss": 0.3987,
      "step": 30414
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7241588830947876,
      "learning_rate": 9.974046490210942e-08,
      "loss": 0.4335,
      "step": 30415
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.515167474746704,
      "learning_rate": 9.959703738067094e-08,
      "loss": 0.3647,
      "step": 30416
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5798348188400269,
      "learning_rate": 9.945371254224234e-08,
      "loss": 0.4484,
      "step": 30417
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.636236548423767,
      "learning_rate": 9.931049038830908e-08,
      "loss": 0.3554,
      "step": 30418
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8342605829238892,
      "learning_rate": 9.916737092035889e-08,
      "loss": 0.9713,
      "step": 30419
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5386329889297485,
      "learning_rate": 9.90243541398761e-08,
      "loss": 1.0237,
      "step": 30420
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7170896530151367,
      "learning_rate": 9.888144004834288e-08,
      "loss": 0.4084,
      "step": 30421
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5078173875808716,
      "learning_rate": 9.873862864724138e-08,
      "loss": 0.3737,
      "step": 30422
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.486685037612915,
      "learning_rate": 9.859591993805262e-08,
      "loss": 0.3939,
      "step": 30423
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5100526809692383,
      "learning_rate": 9.845331392225876e-08,
      "loss": 0.367,
      "step": 30424
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.53400719165802,
      "learning_rate": 9.83108106013364e-08,
      "loss": 0.4122,
      "step": 30425
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5654317140579224,
      "learning_rate": 9.816840997676435e-08,
      "loss": 0.4117,
      "step": 30426
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5840144157409668,
      "learning_rate": 9.802611205002033e-08,
      "loss": 0.4351,
      "step": 30427
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5217576026916504,
      "learning_rate": 9.78839168225798e-08,
      "loss": 0.3936,
      "step": 30428
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5158571004867554,
      "learning_rate": 9.774182429591606e-08,
      "loss": 0.3993,
      "step": 30429
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5482985973358154,
      "learning_rate": 9.759983447150567e-08,
      "loss": 0.362,
      "step": 30430
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5193489789962769,
      "learning_rate": 9.745794735081859e-08,
      "loss": 0.4157,
      "step": 30431
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.583225965499878,
      "learning_rate": 9.73161629353292e-08,
      "loss": 0.3863,
      "step": 30432
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6042766571044922,
      "learning_rate": 9.717448122650408e-08,
      "loss": 0.4136,
      "step": 30433
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5379482507705688,
      "learning_rate": 9.703290222581652e-08,
      "loss": 0.3703,
      "step": 30434
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6186511516571045,
      "learning_rate": 9.689142593473311e-08,
      "loss": 0.4242,
      "step": 30435
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4386732578277588,
      "learning_rate": 9.675005235472156e-08,
      "loss": 0.3586,
      "step": 30436
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6241466999053955,
      "learning_rate": 9.66087814872485e-08,
      "loss": 0.4418,
      "step": 30437
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4939576387405396,
      "learning_rate": 9.646761333377718e-08,
      "loss": 0.3753,
      "step": 30438
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5832916498184204,
      "learning_rate": 9.632654789577422e-08,
      "loss": 0.4159,
      "step": 30439
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.567552089691162,
      "learning_rate": 9.618558517470178e-08,
      "loss": 0.3783,
      "step": 30440
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.621335744857788,
      "learning_rate": 9.604472517202201e-08,
      "loss": 0.4159,
      "step": 30441
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4811182022094727,
      "learning_rate": 9.590396788919598e-08,
      "loss": 0.3993,
      "step": 30442
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5871260166168213,
      "learning_rate": 9.576331332768252e-08,
      "loss": 0.4117,
      "step": 30443
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.56148362159729,
      "learning_rate": 9.562276148894157e-08,
      "loss": 0.3821,
      "step": 30444
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9702235460281372,
      "learning_rate": 9.548231237442972e-08,
      "loss": 0.9954,
      "step": 30445
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4452379941940308,
      "learning_rate": 9.534196598560474e-08,
      "loss": 0.964,
      "step": 30446
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5183533430099487,
      "learning_rate": 9.52017223239221e-08,
      "loss": 0.4015,
      "step": 30447
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.456452488899231,
      "learning_rate": 9.506158139083732e-08,
      "loss": 0.407,
      "step": 30448
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.610160231590271,
      "learning_rate": 9.492154318780145e-08,
      "loss": 0.398,
      "step": 30449
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4871597290039062,
      "learning_rate": 9.478160771626776e-08,
      "loss": 0.375,
      "step": 30450
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6309932470321655,
      "learning_rate": 9.464177497768845e-08,
      "loss": 0.4514,
      "step": 30451
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5130815505981445,
      "learning_rate": 9.450204497351456e-08,
      "loss": 0.3624,
      "step": 30452
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.602264165878296,
      "learning_rate": 9.436241770519273e-08,
      "loss": 0.4242,
      "step": 30453
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5069668292999268,
      "learning_rate": 9.422289317417288e-08,
      "loss": 0.3646,
      "step": 30454
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.506463646888733,
      "learning_rate": 9.408347138190166e-08,
      "loss": 0.4444,
      "step": 30455
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4697997570037842,
      "learning_rate": 9.394415232982568e-08,
      "loss": 0.3623,
      "step": 30456
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5408132076263428,
      "learning_rate": 9.380493601939044e-08,
      "loss": 0.3828,
      "step": 30457
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.538979172706604,
      "learning_rate": 9.366582245203704e-08,
      "loss": 0.3873,
      "step": 30458
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4375423192977905,
      "learning_rate": 9.352681162920985e-08,
      "loss": 0.3824,
      "step": 30459
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4831143617630005,
      "learning_rate": 9.338790355235216e-08,
      "loss": 0.3581,
      "step": 30460
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.571264386177063,
      "learning_rate": 9.324909822290285e-08,
      "loss": 0.4083,
      "step": 30461
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5882680416107178,
      "learning_rate": 9.311039564230295e-08,
      "loss": 0.383,
      "step": 30462
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5606579780578613,
      "learning_rate": 9.297179581198912e-08,
      "loss": 0.4214,
      "step": 30463
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4541703462600708,
      "learning_rate": 9.283329873340019e-08,
      "loss": 0.3739,
      "step": 30464
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.542932152748108,
      "learning_rate": 9.269490440797169e-08,
      "loss": 0.3997,
      "step": 30465
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5240120887756348,
      "learning_rate": 9.255661283714023e-08,
      "loss": 0.3911,
      "step": 30466
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7683566808700562,
      "learning_rate": 9.241842402234025e-08,
      "loss": 0.3952,
      "step": 30467
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6202375888824463,
      "learning_rate": 9.22803379650039e-08,
      "loss": 0.3605,
      "step": 30468
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4839072227478027,
      "learning_rate": 9.21423546665623e-08,
      "loss": 0.395,
      "step": 30469
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4445677995681763,
      "learning_rate": 9.20044741284487e-08,
      "loss": 0.3546,
      "step": 30470
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5270456075668335,
      "learning_rate": 9.186669635209311e-08,
      "loss": 0.3769,
      "step": 30471
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5997191667556763,
      "learning_rate": 9.172902133892325e-08,
      "loss": 0.3674,
      "step": 30472
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4979385137557983,
      "learning_rate": 9.159144909036799e-08,
      "loss": 0.3957,
      "step": 30473
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5461558103561401,
      "learning_rate": 9.145397960785285e-08,
      "loss": 0.3405,
      "step": 30474
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4641749858856201,
      "learning_rate": 9.131661289280447e-08,
      "loss": 0.4388,
      "step": 30475
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4954278469085693,
      "learning_rate": 9.117934894664948e-08,
      "loss": 0.3695,
      "step": 30476
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5872552394866943,
      "learning_rate": 9.104218777080786e-08,
      "loss": 0.4079,
      "step": 30477
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4967344999313354,
      "learning_rate": 9.090512936670404e-08,
      "loss": 0.3537,
      "step": 30478
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5261528491973877,
      "learning_rate": 9.076817373576019e-08,
      "loss": 0.4116,
      "step": 30479
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5213943719863892,
      "learning_rate": 9.06313208793963e-08,
      "loss": 0.3688,
      "step": 30480
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5244446992874146,
      "learning_rate": 9.049457079903013e-08,
      "loss": 0.3912,
      "step": 30481
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5522193908691406,
      "learning_rate": 9.035792349608163e-08,
      "loss": 0.3663,
      "step": 30482
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5366612672805786,
      "learning_rate": 9.022137897196859e-08,
      "loss": 0.4189,
      "step": 30483
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4772754907608032,
      "learning_rate": 9.008493722810651e-08,
      "loss": 0.3542,
      "step": 30484
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5279922485351562,
      "learning_rate": 8.994859826590985e-08,
      "loss": 0.4098,
      "step": 30485
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5797195434570312,
      "learning_rate": 8.9812362086793e-08,
      "loss": 0.3671,
      "step": 30486
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.638657569885254,
      "learning_rate": 8.96762286921693e-08,
      "loss": 0.4526,
      "step": 30487
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4840381145477295,
      "learning_rate": 8.954019808345094e-08,
      "loss": 0.3487,
      "step": 30488
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6145583391189575,
      "learning_rate": 8.9404270262049e-08,
      "loss": 0.4384,
      "step": 30489
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.438212513923645,
      "learning_rate": 8.926844522937128e-08,
      "loss": 0.3554,
      "step": 30490
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6807793378829956,
      "learning_rate": 8.913272298682773e-08,
      "loss": 0.4386,
      "step": 30491
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.590720534324646,
      "learning_rate": 8.899710353582725e-08,
      "loss": 0.3805,
      "step": 30492
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5330395698547363,
      "learning_rate": 8.886158687777536e-08,
      "loss": 0.4192,
      "step": 30493
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5161688327789307,
      "learning_rate": 8.872617301407649e-08,
      "loss": 0.364,
      "step": 30494
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.551724910736084,
      "learning_rate": 8.85908619461362e-08,
      "loss": 0.3913,
      "step": 30495
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4550787210464478,
      "learning_rate": 8.845565367535891e-08,
      "loss": 0.359,
      "step": 30496
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6053502559661865,
      "learning_rate": 8.832054820314462e-08,
      "loss": 0.4497,
      "step": 30497
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.565625786781311,
      "learning_rate": 8.818554553089664e-08,
      "loss": 0.3676,
      "step": 30498
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.574733018875122,
      "learning_rate": 8.805064566001386e-08,
      "loss": 0.4338,
      "step": 30499
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6192532777786255,
      "learning_rate": 8.791584859189628e-08,
      "loss": 0.3985,
      "step": 30500
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5747276544570923,
      "learning_rate": 8.778115432794166e-08,
      "loss": 0.4181,
      "step": 30501
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4750698804855347,
      "learning_rate": 8.764656286954665e-08,
      "loss": 0.3586,
      "step": 30502
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4788492918014526,
      "learning_rate": 8.751207421810682e-08,
      "loss": 0.3717,
      "step": 30503
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5262919664382935,
      "learning_rate": 8.73776883750188e-08,
      "loss": 0.3804,
      "step": 30504
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5800215005874634,
      "learning_rate": 8.724340534167486e-08,
      "loss": 0.423,
      "step": 30505
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4683918952941895,
      "learning_rate": 8.710922511946718e-08,
      "loss": 0.3635,
      "step": 30506
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.528645396232605,
      "learning_rate": 8.69751477097891e-08,
      "loss": 0.4311,
      "step": 30507
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.572100043296814,
      "learning_rate": 8.684117311403061e-08,
      "loss": 0.3824,
      "step": 30508
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5644813776016235,
      "learning_rate": 8.670730133358173e-08,
      "loss": 0.4232,
      "step": 30509
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5781348943710327,
      "learning_rate": 8.657353236982913e-08,
      "loss": 0.3793,
      "step": 30510
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7398773431777954,
      "learning_rate": 8.643986622416167e-08,
      "loss": 0.4154,
      "step": 30511
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5228221416473389,
      "learning_rate": 8.630630289796605e-08,
      "loss": 0.3531,
      "step": 30512
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5421850681304932,
      "learning_rate": 8.617284239262558e-08,
      "loss": 0.4246,
      "step": 30513
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5427155494689941,
      "learning_rate": 8.603948470952694e-08,
      "loss": 0.3753,
      "step": 30514
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4894715547561646,
      "learning_rate": 8.590622985005126e-08,
      "loss": 0.3891,
      "step": 30515
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5170549154281616,
      "learning_rate": 8.577307781558186e-08,
      "loss": 0.3745,
      "step": 30516
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5778405666351318,
      "learning_rate": 8.564002860749875e-08,
      "loss": 0.4158,
      "step": 30517
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4952868223190308,
      "learning_rate": 8.550708222718307e-08,
      "loss": 0.375,
      "step": 30518
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.540579080581665,
      "learning_rate": 8.537423867601257e-08,
      "loss": 0.4364,
      "step": 30519
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5290135145187378,
      "learning_rate": 8.524149795536618e-08,
      "loss": 0.3815,
      "step": 30520
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5369529724121094,
      "learning_rate": 8.510886006661945e-08,
      "loss": 0.4012,
      "step": 30521
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.469482183456421,
      "learning_rate": 8.497632501114905e-08,
      "loss": 0.3646,
      "step": 30522
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5687026977539062,
      "learning_rate": 8.484389279032835e-08,
      "loss": 0.4071,
      "step": 30523
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.530265212059021,
      "learning_rate": 8.471156340553177e-08,
      "loss": 0.3747,
      "step": 30524
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5714539289474487,
      "learning_rate": 8.457933685813269e-08,
      "loss": 0.4154,
      "step": 30525
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.48281991481781,
      "learning_rate": 8.44472131495e-08,
      "loss": 0.3818,
      "step": 30526
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.415267825126648,
      "learning_rate": 8.431519228100593e-08,
      "loss": 0.3704,
      "step": 30527
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6316553354263306,
      "learning_rate": 8.418327425401939e-08,
      "loss": 0.3942,
      "step": 30528
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.709257960319519,
      "learning_rate": 8.40514590699082e-08,
      "loss": 0.9873,
      "step": 30529
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4343324899673462,
      "learning_rate": 8.391974673004011e-08,
      "loss": 1.0158,
      "step": 30530
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5813006162643433,
      "learning_rate": 8.378813723578072e-08,
      "loss": 0.4544,
      "step": 30531
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5020418167114258,
      "learning_rate": 8.36566305884956e-08,
      "loss": 0.3761,
      "step": 30532
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6463242769241333,
      "learning_rate": 8.35252267895481e-08,
      "loss": 0.4168,
      "step": 30533
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6541212797164917,
      "learning_rate": 8.33939258403016e-08,
      "loss": 0.3874,
      "step": 30534
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7279871702194214,
      "learning_rate": 8.326272774211608e-08,
      "loss": 0.4635,
      "step": 30535
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4768389463424683,
      "learning_rate": 8.313163249635492e-08,
      "loss": 0.3692,
      "step": 30536
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.495907187461853,
      "learning_rate": 8.300064010437703e-08,
      "loss": 0.4256,
      "step": 30537
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.582025408744812,
      "learning_rate": 8.28697505675391e-08,
      "loss": 0.4419,
      "step": 30538
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5689352750778198,
      "learning_rate": 8.273896388720115e-08,
      "loss": 0.4284,
      "step": 30539
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6245609521865845,
      "learning_rate": 8.260828006471877e-08,
      "loss": 0.4011,
      "step": 30540
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5076066255569458,
      "learning_rate": 8.247769910144753e-08,
      "loss": 0.387,
      "step": 30541
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5740710496902466,
      "learning_rate": 8.23472209987397e-08,
      "loss": 0.3955,
      "step": 30542
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5109513998031616,
      "learning_rate": 8.221684575795197e-08,
      "loss": 0.4217,
      "step": 30543
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5149441957473755,
      "learning_rate": 8.208657338043435e-08,
      "loss": 0.3449,
      "step": 30544
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5221061706542969,
      "learning_rate": 8.195640386753911e-08,
      "loss": 0.4207,
      "step": 30545
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5220227241516113,
      "learning_rate": 8.182633722061517e-08,
      "loss": 0.3525,
      "step": 30546
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5667954683303833,
      "learning_rate": 8.169637344101145e-08,
      "loss": 0.4425,
      "step": 30547
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4520039558410645,
      "learning_rate": 8.156651253007796e-08,
      "loss": 0.3844,
      "step": 30548
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4595654010772705,
      "learning_rate": 8.143675448915922e-08,
      "loss": 0.4222,
      "step": 30549
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4682284593582153,
      "learning_rate": 8.13070993196019e-08,
      "loss": 0.3763,
      "step": 30550
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5769068002700806,
      "learning_rate": 8.11775470227505e-08,
      "loss": 0.4472,
      "step": 30551
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5018327236175537,
      "learning_rate": 8.104809759994947e-08,
      "loss": 0.3521,
      "step": 30552
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.611708402633667,
      "learning_rate": 8.09187510525411e-08,
      "loss": 0.4412,
      "step": 30553
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5937751531600952,
      "learning_rate": 8.078950738186541e-08,
      "loss": 0.3938,
      "step": 30554
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.480608344078064,
      "learning_rate": 8.066036658926579e-08,
      "loss": 0.3608,
      "step": 30555
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5096735954284668,
      "learning_rate": 8.053132867608004e-08,
      "loss": 0.3841,
      "step": 30556
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5166586637496948,
      "learning_rate": 8.040239364364488e-08,
      "loss": 0.3941,
      "step": 30557
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.51716947555542,
      "learning_rate": 8.027356149329923e-08,
      "loss": 0.3489,
      "step": 30558
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5084459781646729,
      "learning_rate": 8.01448322263798e-08,
      "loss": 0.3931,
      "step": 30559
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.480205774307251,
      "learning_rate": 8.001620584422221e-08,
      "loss": 0.3562,
      "step": 30560
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5126206874847412,
      "learning_rate": 7.98876823481587e-08,
      "loss": 0.4195,
      "step": 30561
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4539592266082764,
      "learning_rate": 7.975926173952264e-08,
      "loss": 0.3922,
      "step": 30562
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6442784070968628,
      "learning_rate": 7.963094401964522e-08,
      "loss": 0.487,
      "step": 30563
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5013383626937866,
      "learning_rate": 7.950272918985979e-08,
      "loss": 0.3672,
      "step": 30564
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.584317684173584,
      "learning_rate": 7.937461725149531e-08,
      "loss": 0.4235,
      "step": 30565
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5787477493286133,
      "learning_rate": 7.924660820587848e-08,
      "loss": 0.3769,
      "step": 30566
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5860482454299927,
      "learning_rate": 7.911870205433936e-08,
      "loss": 0.4234,
      "step": 30567
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4825711250305176,
      "learning_rate": 7.899089879820355e-08,
      "loss": 0.3553,
      "step": 30568
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5782052278518677,
      "learning_rate": 7.886319843879553e-08,
      "loss": 0.4065,
      "step": 30569
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5663233995437622,
      "learning_rate": 7.873560097744204e-08,
      "loss": 0.3771,
      "step": 30570
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.512471079826355,
      "learning_rate": 7.860810641546534e-08,
      "loss": 0.3925,
      "step": 30571
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5789556503295898,
      "learning_rate": 7.848071475418773e-08,
      "loss": 0.3978,
      "step": 30572
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5315793752670288,
      "learning_rate": 7.835342599492923e-08,
      "loss": 0.4597,
      "step": 30573
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4658691883087158,
      "learning_rate": 7.822624013901215e-08,
      "loss": 0.3707,
      "step": 30574
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.506351113319397,
      "learning_rate": 7.809915718775429e-08,
      "loss": 0.4064,
      "step": 30575
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4881616830825806,
      "learning_rate": 7.797217714247462e-08,
      "loss": 0.4046,
      "step": 30576
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5469200611114502,
      "learning_rate": 7.784530000448875e-08,
      "loss": 0.3929,
      "step": 30577
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.478270173072815,
      "learning_rate": 7.771852577511341e-08,
      "loss": 0.3672,
      "step": 30578
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.015061140060425,
      "learning_rate": 7.759185445566419e-08,
      "loss": 0.4215,
      "step": 30579
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.488487720489502,
      "learning_rate": 7.746528604745451e-08,
      "loss": 0.3597,
      "step": 30580
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5283393859863281,
      "learning_rate": 7.733882055179665e-08,
      "loss": 0.368,
      "step": 30581
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5090159177780151,
      "learning_rate": 7.721245797000177e-08,
      "loss": 0.3452,
      "step": 30582
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4890278577804565,
      "learning_rate": 7.708619830338105e-08,
      "loss": 0.3781,
      "step": 30583
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5234156847000122,
      "learning_rate": 7.696004155324454e-08,
      "loss": 0.3851,
      "step": 30584
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6312780380249023,
      "learning_rate": 7.683398772089901e-08,
      "loss": 0.4107,
      "step": 30585
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5348604917526245,
      "learning_rate": 7.670803680765449e-08,
      "loss": 0.3661,
      "step": 30586
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5076875686645508,
      "learning_rate": 7.658218881481438e-08,
      "loss": 0.3642,
      "step": 30587
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5325156450271606,
      "learning_rate": 7.645644374368655e-08,
      "loss": 0.358,
      "step": 30588
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6050045490264893,
      "learning_rate": 7.633080159557327e-08,
      "loss": 0.4271,
      "step": 30589
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5146183967590332,
      "learning_rate": 7.620526237177794e-08,
      "loss": 0.3714,
      "step": 30590
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5570640563964844,
      "learning_rate": 7.607982607360398e-08,
      "loss": 0.3901,
      "step": 30591
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5076216459274292,
      "learning_rate": 7.595449270235034e-08,
      "loss": 0.3591,
      "step": 30592
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.545420527458191,
      "learning_rate": 7.582926225931819e-08,
      "loss": 0.4119,
      "step": 30593
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4539493322372437,
      "learning_rate": 7.570413474580651e-08,
      "loss": 0.3576,
      "step": 30594
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5498977899551392,
      "learning_rate": 7.557911016311203e-08,
      "loss": 0.4441,
      "step": 30595
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5433776378631592,
      "learning_rate": 7.545418851253372e-08,
      "loss": 0.3906,
      "step": 30596
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6002171039581299,
      "learning_rate": 7.532936979536498e-08,
      "loss": 0.4683,
      "step": 30597
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6123900413513184,
      "learning_rate": 7.520465401290033e-08,
      "loss": 0.4019,
      "step": 30598
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5841608047485352,
      "learning_rate": 7.508004116643431e-08,
      "loss": 0.4369,
      "step": 30599
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.462530493736267,
      "learning_rate": 7.495553125725807e-08,
      "loss": 0.3555,
      "step": 30600
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.607640027999878,
      "learning_rate": 7.483112428666506e-08,
      "loss": 0.4117,
      "step": 30601
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.612671136856079,
      "learning_rate": 7.470682025594533e-08,
      "loss": 0.3821,
      "step": 30602
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5933482646942139,
      "learning_rate": 7.458261916638565e-08,
      "loss": 0.4595,
      "step": 30603
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5221244096755981,
      "learning_rate": 7.445852101927609e-08,
      "loss": 0.3646,
      "step": 30604
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5303486585617065,
      "learning_rate": 7.43345258159034e-08,
      "loss": 0.4088,
      "step": 30605
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5052354335784912,
      "learning_rate": 7.421063355755431e-08,
      "loss": 0.3648,
      "step": 30606
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5260009765625,
      "learning_rate": 7.408684424551227e-08,
      "loss": 0.4051,
      "step": 30607
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4770357608795166,
      "learning_rate": 7.396315788106289e-08,
      "loss": 0.3703,
      "step": 30608
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5046247243881226,
      "learning_rate": 7.38395744654874e-08,
      "loss": 0.395,
      "step": 30609
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7269014120101929,
      "learning_rate": 7.371609400006807e-08,
      "loss": 0.3732,
      "step": 30610
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8086405992507935,
      "learning_rate": 7.359271648608613e-08,
      "loss": 0.5028,
      "step": 30611
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.557613492012024,
      "learning_rate": 7.346944192482053e-08,
      "loss": 0.3845,
      "step": 30612
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6220197677612305,
      "learning_rate": 7.334627031755137e-08,
      "loss": 0.4065,
      "step": 30613
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5412338972091675,
      "learning_rate": 7.32232016655532e-08,
      "loss": 0.362,
      "step": 30614
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5213996171951294,
      "learning_rate": 7.310023597010496e-08,
      "loss": 0.4093,
      "step": 30615
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.203960657119751,
      "learning_rate": 7.297737323248122e-08,
      "loss": 0.3367,
      "step": 30616
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.571818470954895,
      "learning_rate": 7.285461345395651e-08,
      "loss": 0.3873,
      "step": 30617
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.574948787689209,
      "learning_rate": 7.273195663580312e-08,
      "loss": 0.401,
      "step": 30618
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5117135047912598,
      "learning_rate": 7.260940277929452e-08,
      "loss": 0.3996,
      "step": 30619
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5023943185806274,
      "learning_rate": 7.248695188569966e-08,
      "loss": 0.3716,
      "step": 30620
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5544756650924683,
      "learning_rate": 7.236460395629197e-08,
      "loss": 0.4028,
      "step": 30621
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5966423749923706,
      "learning_rate": 7.224235899233712e-08,
      "loss": 0.3984,
      "step": 30622
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.568603515625,
      "learning_rate": 7.212021699510519e-08,
      "loss": 0.4086,
      "step": 30623
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4909000396728516,
      "learning_rate": 7.199817796586073e-08,
      "loss": 0.3802,
      "step": 30624
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6650028228759766,
      "learning_rate": 7.187624190587273e-08,
      "loss": 0.3936,
      "step": 30625
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5015255212783813,
      "learning_rate": 7.175440881640239e-08,
      "loss": 0.3706,
      "step": 30626
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5344620943069458,
      "learning_rate": 7.163267869871649e-08,
      "loss": 0.3914,
      "step": 30627
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.509222149848938,
      "learning_rate": 7.151105155407401e-08,
      "loss": 0.3582,
      "step": 30628
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5778132677078247,
      "learning_rate": 7.13895273837395e-08,
      "loss": 0.4357,
      "step": 30629
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5392563343048096,
      "learning_rate": 7.126810618897195e-08,
      "loss": 0.3948,
      "step": 30630
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4765031337738037,
      "learning_rate": 7.114678797103147e-08,
      "loss": 0.4315,
      "step": 30631
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4773775339126587,
      "learning_rate": 7.102557273117594e-08,
      "loss": 0.3881,
      "step": 30632
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6248501539230347,
      "learning_rate": 7.090446047066213e-08,
      "loss": 0.3723,
      "step": 30633
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5366809368133545,
      "learning_rate": 7.07834511907457e-08,
      "loss": 0.3628,
      "step": 30634
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.56947922706604,
      "learning_rate": 7.066254489268343e-08,
      "loss": 0.4235,
      "step": 30635
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5426661968231201,
      "learning_rate": 7.054174157772765e-08,
      "loss": 0.3864,
      "step": 30636
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5334709882736206,
      "learning_rate": 7.04210412471329e-08,
      "loss": 0.4338,
      "step": 30637
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5923423767089844,
      "learning_rate": 7.030044390215041e-08,
      "loss": 0.4004,
      "step": 30638
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4894646406173706,
      "learning_rate": 7.017994954402918e-08,
      "loss": 0.3652,
      "step": 30639
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5579276084899902,
      "learning_rate": 7.005955817402155e-08,
      "loss": 0.4056,
      "step": 30640
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5778800249099731,
      "learning_rate": 6.99392697933754e-08,
      "loss": 0.4251,
      "step": 30641
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5331445932388306,
      "learning_rate": 6.98190844033364e-08,
      "loss": 0.3937,
      "step": 30642
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4918216466903687,
      "learning_rate": 6.969900200515467e-08,
      "loss": 0.3972,
      "step": 30643
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5100739002227783,
      "learning_rate": 6.957902260007254e-08,
      "loss": 0.3756,
      "step": 30644
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5433428287506104,
      "learning_rate": 6.945914618933569e-08,
      "loss": 0.4157,
      "step": 30645
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6449837684631348,
      "learning_rate": 6.933937277418646e-08,
      "loss": 0.41,
      "step": 30646
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5798999071121216,
      "learning_rate": 6.921970235586828e-08,
      "loss": 0.3713,
      "step": 30647
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.644407033920288,
      "learning_rate": 6.910013493562351e-08,
      "loss": 0.3769,
      "step": 30648
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5186306238174438,
      "learning_rate": 6.898067051468892e-08,
      "loss": 0.4524,
      "step": 30649
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4244394302368164,
      "learning_rate": 6.886130909430577e-08,
      "loss": 0.3589,
      "step": 30650
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5217233896255493,
      "learning_rate": 6.874205067571082e-08,
      "loss": 0.4225,
      "step": 30651
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.512425184249878,
      "learning_rate": 6.86228952601431e-08,
      "loss": 0.3667,
      "step": 30652
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.556145429611206,
      "learning_rate": 6.850384284883715e-08,
      "loss": 0.4453,
      "step": 30653
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5126733779907227,
      "learning_rate": 6.838489344302646e-08,
      "loss": 0.3626,
      "step": 30654
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.549775242805481,
      "learning_rate": 6.826604704394557e-08,
      "loss": 0.4105,
      "step": 30655
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5269659757614136,
      "learning_rate": 6.814730365282795e-08,
      "loss": 0.4285,
      "step": 30656
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.510048747062683,
      "learning_rate": 6.802866327090486e-08,
      "loss": 0.4123,
      "step": 30657
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4807569980621338,
      "learning_rate": 6.791012589940638e-08,
      "loss": 0.3692,
      "step": 30658
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5199217796325684,
      "learning_rate": 6.779169153956267e-08,
      "loss": 0.4435,
      "step": 30659
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.504221796989441,
      "learning_rate": 6.767336019260052e-08,
      "loss": 0.3565,
      "step": 30660
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4992117881774902,
      "learning_rate": 6.755513185974893e-08,
      "loss": 0.3945,
      "step": 30661
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5184370279312134,
      "learning_rate": 6.743700654223251e-08,
      "loss": 0.4018,
      "step": 30662
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4995841979980469,
      "learning_rate": 6.731898424127914e-08,
      "loss": 0.4225,
      "step": 30663
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.545767068862915,
      "learning_rate": 6.720106495810896e-08,
      "loss": 0.3708,
      "step": 30664
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5385627746582031,
      "learning_rate": 6.708324869394878e-08,
      "loss": 0.388,
      "step": 30665
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5454530715942383,
      "learning_rate": 6.69655354500176e-08,
      "loss": 0.3674,
      "step": 30666
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5926754474639893,
      "learning_rate": 6.68479252275378e-08,
      "loss": 0.4451,
      "step": 30667
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5773059129714966,
      "learning_rate": 6.67304180277295e-08,
      "loss": 0.3698,
      "step": 30668
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5028581619262695,
      "learning_rate": 6.661301385181062e-08,
      "loss": 0.4076,
      "step": 30669
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5968520641326904,
      "learning_rate": 6.649571270099908e-08,
      "loss": 0.3883,
      "step": 30670
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5439653396606445,
      "learning_rate": 6.637851457651168e-08,
      "loss": 0.4206,
      "step": 30671
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5013492107391357,
      "learning_rate": 6.626141947956299e-08,
      "loss": 0.3715,
      "step": 30672
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5890274047851562,
      "learning_rate": 6.614442741136872e-08,
      "loss": 0.3998,
      "step": 30673
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5621824264526367,
      "learning_rate": 6.602753837314236e-08,
      "loss": 0.4098,
      "step": 30674
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.519309401512146,
      "learning_rate": 6.5910752366094e-08,
      "loss": 0.4258,
      "step": 30675
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4785637855529785,
      "learning_rate": 6.579406939143829e-08,
      "loss": 0.3714,
      "step": 30676
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5904768705368042,
      "learning_rate": 6.56774894503831e-08,
      "loss": 0.394,
      "step": 30677
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4728659391403198,
      "learning_rate": 6.556101254413749e-08,
      "loss": 0.3461,
      "step": 30678
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.612119436264038,
      "learning_rate": 6.54446386739116e-08,
      "loss": 0.4436,
      "step": 30679
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.438148021697998,
      "learning_rate": 6.532836784090891e-08,
      "loss": 0.3599,
      "step": 30680
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5589734315872192,
      "learning_rate": 6.521220004633954e-08,
      "loss": 0.4095,
      "step": 30681
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4892936944961548,
      "learning_rate": 6.50961352914048e-08,
      "loss": 0.3685,
      "step": 30682
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.562382459640503,
      "learning_rate": 6.498017357731035e-08,
      "loss": 0.4215,
      "step": 30683
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.56349515914917,
      "learning_rate": 6.486431490525858e-08,
      "loss": 0.3703,
      "step": 30684
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.645533561706543,
      "learning_rate": 6.474855927645074e-08,
      "loss": 0.4191,
      "step": 30685
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4660582542419434,
      "learning_rate": 6.4632906692087e-08,
      "loss": 0.3445,
      "step": 30686
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4735376834869385,
      "learning_rate": 6.451735715336748e-08,
      "loss": 0.3884,
      "step": 30687
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5779372453689575,
      "learning_rate": 6.440191066149237e-08,
      "loss": 0.381,
      "step": 30688
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4965952634811401,
      "learning_rate": 6.428656721765625e-08,
      "loss": 0.4043,
      "step": 30689
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.588547706604004,
      "learning_rate": 6.417132682305594e-08,
      "loss": 0.37,
      "step": 30690
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5777729749679565,
      "learning_rate": 6.405618947888715e-08,
      "loss": 0.4269,
      "step": 30691
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4594919681549072,
      "learning_rate": 6.394115518634448e-08,
      "loss": 0.38,
      "step": 30692
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6641706228256226,
      "learning_rate": 6.38262239466203e-08,
      "loss": 0.4784,
      "step": 30693
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.515191912651062,
      "learning_rate": 6.371139576090702e-08,
      "loss": 0.3775,
      "step": 30694
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7201670408248901,
      "learning_rate": 6.359667063039587e-08,
      "loss": 0.4193,
      "step": 30695
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4665708541870117,
      "learning_rate": 6.348204855627593e-08,
      "loss": 0.3475,
      "step": 30696
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4980125427246094,
      "learning_rate": 6.336752953973624e-08,
      "loss": 0.3837,
      "step": 30697
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.607492446899414,
      "learning_rate": 6.325311358196584e-08,
      "loss": 0.394,
      "step": 30698
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.643651008605957,
      "learning_rate": 6.313880068414935e-08,
      "loss": 0.4167,
      "step": 30699
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.476438045501709,
      "learning_rate": 6.302459084747358e-08,
      "loss": 0.3411,
      "step": 30700
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5562827587127686,
      "learning_rate": 6.291048407312317e-08,
      "loss": 0.4402,
      "step": 30701
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4894294738769531,
      "learning_rate": 6.279648036228047e-08,
      "loss": 0.3528,
      "step": 30702
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.560257911682129,
      "learning_rate": 6.268257971612901e-08,
      "loss": 0.4275,
      "step": 30703
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.487507700920105,
      "learning_rate": 6.256878213585004e-08,
      "loss": 0.3694,
      "step": 30704
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5391855239868164,
      "learning_rate": 6.245508762262375e-08,
      "loss": 0.4399,
      "step": 30705
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.511276125907898,
      "learning_rate": 6.234149617762808e-08,
      "loss": 0.3699,
      "step": 30706
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5593185424804688,
      "learning_rate": 6.22280078020432e-08,
      "loss": 0.4117,
      "step": 30707
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5169095993041992,
      "learning_rate": 6.211462249704481e-08,
      "loss": 0.3606,
      "step": 30708
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.502321720123291,
      "learning_rate": 6.200134026380867e-08,
      "loss": 0.4249,
      "step": 30709
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5286872386932373,
      "learning_rate": 6.188816110351159e-08,
      "loss": 0.3677,
      "step": 30710
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4490196704864502,
      "learning_rate": 6.177508501732599e-08,
      "loss": 0.4072,
      "step": 30711
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5181137323379517,
      "learning_rate": 6.166211200642314e-08,
      "loss": 0.3811,
      "step": 30712
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.505115270614624,
      "learning_rate": 6.154924207197876e-08,
      "loss": 0.3992,
      "step": 30713
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5463355779647827,
      "learning_rate": 6.143647521515972e-08,
      "loss": 0.359,
      "step": 30714
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5633915662765503,
      "learning_rate": 6.132381143713728e-08,
      "loss": 0.4366,
      "step": 30715
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5217068195343018,
      "learning_rate": 6.12112507390794e-08,
      "loss": 0.3903,
      "step": 30716
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.541999340057373,
      "learning_rate": 6.109879312215405e-08,
      "loss": 0.3906,
      "step": 30717
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5378843545913696,
      "learning_rate": 6.098643858752695e-08,
      "loss": 0.3736,
      "step": 30718
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5871412754058838,
      "learning_rate": 6.087418713636384e-08,
      "loss": 0.4304,
      "step": 30719
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4948525428771973,
      "learning_rate": 6.076203876982934e-08,
      "loss": 0.3588,
      "step": 30720
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.54078209400177,
      "learning_rate": 6.064999348908584e-08,
      "loss": 0.4442,
      "step": 30721
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5399082899093628,
      "learning_rate": 6.053805129529577e-08,
      "loss": 0.3989,
      "step": 30722
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5373380184173584,
      "learning_rate": 6.042621218962041e-08,
      "loss": 0.4218,
      "step": 30723
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5159419775009155,
      "learning_rate": 6.031447617321883e-08,
      "loss": 0.3763,
      "step": 30724
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.03655743598938,
      "learning_rate": 6.020284324725123e-08,
      "loss": 0.4047,
      "step": 30725
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4482367038726807,
      "learning_rate": 6.009131341287445e-08,
      "loss": 0.3646,
      "step": 30726
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5494908094406128,
      "learning_rate": 5.997988667124421e-08,
      "loss": 0.4446,
      "step": 30727
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.462161898612976,
      "learning_rate": 5.986856302351851e-08,
      "loss": 0.3452,
      "step": 30728
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5007468461990356,
      "learning_rate": 5.975734247085196e-08,
      "loss": 0.4381,
      "step": 30729
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5594396591186523,
      "learning_rate": 5.964622501439588e-08,
      "loss": 0.3712,
      "step": 30730
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5365703105926514,
      "learning_rate": 5.953521065530377e-08,
      "loss": 0.3767,
      "step": 30731
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5466187000274658,
      "learning_rate": 5.942429939472694e-08,
      "loss": 0.4087,
      "step": 30732
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.540321707725525,
      "learning_rate": 5.9313491233816686e-08,
      "loss": 0.4332,
      "step": 30733
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4827052354812622,
      "learning_rate": 5.9202786173720994e-08,
      "loss": 0.3477,
      "step": 30734
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.502943515777588,
      "learning_rate": 5.909218421558893e-08,
      "loss": 0.3793,
      "step": 30735
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5345244407653809,
      "learning_rate": 5.8981685360566256e-08,
      "loss": 0.3769,
      "step": 30736
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6238569021224976,
      "learning_rate": 5.887128960980093e-08,
      "loss": 0.4472,
      "step": 30737
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5539205074310303,
      "learning_rate": 5.876099696443649e-08,
      "loss": 0.3793,
      "step": 30738
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5565146207809448,
      "learning_rate": 5.8650807425617574e-08,
      "loss": 0.427,
      "step": 30739
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4980199337005615,
      "learning_rate": 5.854072099448771e-08,
      "loss": 0.3665,
      "step": 30740
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5232776403427124,
      "learning_rate": 5.84307376721871e-08,
      "loss": 0.4245,
      "step": 30741
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.509840965270996,
      "learning_rate": 5.832085745985594e-08,
      "loss": 0.3761,
      "step": 30742
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5364372730255127,
      "learning_rate": 5.821108035863665e-08,
      "loss": 0.393,
      "step": 30743
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5269966125488281,
      "learning_rate": 5.810140636966499e-08,
      "loss": 0.3877,
      "step": 30744
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6204142570495605,
      "learning_rate": 5.799183549408005e-08,
      "loss": 0.4208,
      "step": 30745
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5142313241958618,
      "learning_rate": 5.7882367733018696e-08,
      "loss": 0.3542,
      "step": 30746
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5166823863983154,
      "learning_rate": 5.777300308761446e-08,
      "loss": 0.3926,
      "step": 30747
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5251699686050415,
      "learning_rate": 5.766374155900312e-08,
      "loss": 0.3658,
      "step": 30748
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6677144765853882,
      "learning_rate": 5.7554583148317075e-08,
      "loss": 0.9096,
      "step": 30749
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4240680932998657,
      "learning_rate": 5.7445527856688774e-08,
      "loss": 0.9278,
      "step": 30750
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5961836576461792,
      "learning_rate": 5.733657568524953e-08,
      "loss": 0.5228,
      "step": 30751
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6036622524261475,
      "learning_rate": 5.7227726635129544e-08,
      "loss": 0.3758,
      "step": 30752
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4960087537765503,
      "learning_rate": 5.7118980707456804e-08,
      "loss": 0.3798,
      "step": 30753
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6403107643127441,
      "learning_rate": 5.7010337903359305e-08,
      "loss": 0.4031,
      "step": 30754
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5697270631790161,
      "learning_rate": 5.690179822396391e-08,
      "loss": 0.4456,
      "step": 30755
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4541242122650146,
      "learning_rate": 5.6793361670398616e-08,
      "loss": 0.3595,
      "step": 30756
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.544527292251587,
      "learning_rate": 5.668502824378475e-08,
      "loss": 0.4333,
      "step": 30757
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5207058191299438,
      "learning_rate": 5.6576797945246956e-08,
      "loss": 0.3646,
      "step": 30758
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6456170082092285,
      "learning_rate": 5.6468670775908783e-08,
      "loss": 0.4505,
      "step": 30759
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5161926746368408,
      "learning_rate": 5.636064673689046e-08,
      "loss": 0.3773,
      "step": 30760
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6027954816818237,
      "learning_rate": 5.6252725829314404e-08,
      "loss": 0.4156,
      "step": 30761
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.576893925666809,
      "learning_rate": 5.6144908054297506e-08,
      "loss": 0.3733,
      "step": 30762
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.659620761871338,
      "learning_rate": 5.603719341295777e-08,
      "loss": 0.4367,
      "step": 30763
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4554983377456665,
      "learning_rate": 5.59295819064154e-08,
      "loss": 0.3575,
      "step": 30764
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5925207138061523,
      "learning_rate": 5.582207353578395e-08,
      "loss": 0.4743,
      "step": 30765
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5053510665893555,
      "learning_rate": 5.5714668302179195e-08,
      "loss": 0.3537,
      "step": 30766
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4700756072998047,
      "learning_rate": 5.56073662067147e-08,
      "loss": 0.4055,
      "step": 30767
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4696378707885742,
      "learning_rate": 5.550016725050289e-08,
      "loss": 0.3789,
      "step": 30768
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5684559345245361,
      "learning_rate": 5.539307143465844e-08,
      "loss": 0.4208,
      "step": 30769
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5391377210617065,
      "learning_rate": 5.5286078760288245e-08,
      "loss": 0.3877,
      "step": 30770
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5641430616378784,
      "learning_rate": 5.517918922850474e-08,
      "loss": 0.3878,
      "step": 30771
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5147291421890259,
      "learning_rate": 5.507240284041593e-08,
      "loss": 0.418,
      "step": 30772
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5597273111343384,
      "learning_rate": 5.496571959712871e-08,
      "loss": 0.451,
      "step": 30773
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5200214385986328,
      "learning_rate": 5.485913949974886e-08,
      "loss": 0.3857,
      "step": 30774
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.870661735534668,
      "learning_rate": 5.4752662549384384e-08,
      "loss": 0.4118,
      "step": 30775
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5343958139419556,
      "learning_rate": 5.464628874713773e-08,
      "loss": 0.4264,
      "step": 30776
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5244680643081665,
      "learning_rate": 5.454001809411247e-08,
      "loss": 0.4132,
      "step": 30777
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5628031492233276,
      "learning_rate": 5.443385059140993e-08,
      "loss": 0.3768,
      "step": 30778
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5457313060760498,
      "learning_rate": 5.4327786240132576e-08,
      "loss": 0.4065,
      "step": 30779
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6459636688232422,
      "learning_rate": 5.422182504137952e-08,
      "loss": 0.3755,
      "step": 30780
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5205222368240356,
      "learning_rate": 5.411596699625099e-08,
      "loss": 0.4342,
      "step": 30781
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4718091487884521,
      "learning_rate": 5.4010212105845005e-08,
      "loss": 0.3789,
      "step": 30782
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5966588258743286,
      "learning_rate": 5.3904560371256245e-08,
      "loss": 0.3834,
      "step": 30783
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4908801317214966,
      "learning_rate": 5.37990117935816e-08,
      "loss": 0.3516,
      "step": 30784
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.523085355758667,
      "learning_rate": 5.369356637391687e-08,
      "loss": 0.4035,
      "step": 30785
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5389822721481323,
      "learning_rate": 5.358822411335452e-08,
      "loss": 0.3716,
      "step": 30786
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6304465532302856,
      "learning_rate": 5.34829850129881e-08,
      "loss": 0.4125,
      "step": 30787
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4575988054275513,
      "learning_rate": 5.337784907390675e-08,
      "loss": 0.3438,
      "step": 30788
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5905177593231201,
      "learning_rate": 5.3272816297204046e-08,
      "loss": 0.4631,
      "step": 30789
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.543036937713623,
      "learning_rate": 5.316788668396688e-08,
      "loss": 0.3953,
      "step": 30790
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5468488931655884,
      "learning_rate": 5.30630602352844e-08,
      "loss": 0.4274,
      "step": 30791
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5976587533950806,
      "learning_rate": 5.295833695224573e-08,
      "loss": 0.386,
      "step": 30792
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6529548168182373,
      "learning_rate": 5.285371683593332e-08,
      "loss": 0.9998,
      "step": 30793
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4450435638427734,
      "learning_rate": 5.2749199887435206e-08,
      "loss": 0.9904,
      "step": 30794
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.529720664024353,
      "learning_rate": 5.2644786107833856e-08,
      "loss": 0.4057,
      "step": 30795
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5384914875030518,
      "learning_rate": 5.254047549821284e-08,
      "loss": 0.3786,
      "step": 30796
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5588300228118896,
      "learning_rate": 5.243626805965352e-08,
      "loss": 0.4,
      "step": 30797
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4926451444625854,
      "learning_rate": 5.233216379323725e-08,
      "loss": 0.362,
      "step": 30798
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.651349425315857,
      "learning_rate": 5.222816270004427e-08,
      "loss": 0.438,
      "step": 30799
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4670101404190063,
      "learning_rate": 5.2124264781151514e-08,
      "loss": 0.3659,
      "step": 30800
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5495678186416626,
      "learning_rate": 5.2020470037638104e-08,
      "loss": 0.3916,
      "step": 30801
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5278511047363281,
      "learning_rate": 5.191677847058096e-08,
      "loss": 0.3795,
      "step": 30802
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5922268629074097,
      "learning_rate": 5.1813190081052566e-08,
      "loss": 0.4375,
      "step": 30803
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5068790912628174,
      "learning_rate": 5.1709704870130936e-08,
      "loss": 0.3492,
      "step": 30804
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6302465200424194,
      "learning_rate": 5.1606322838887444e-08,
      "loss": 0.4168,
      "step": 30805
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4993335008621216,
      "learning_rate": 5.1503043988394565e-08,
      "loss": 0.3808,
      "step": 30806
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5631840229034424,
      "learning_rate": 5.139986831972366e-08,
      "loss": 0.3756,
      "step": 30807
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5349249839782715,
      "learning_rate": 5.129679583394387e-08,
      "loss": 0.4044,
      "step": 30808
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.492904782295227,
      "learning_rate": 5.1193826532126564e-08,
      "loss": 0.4061,
      "step": 30809
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5551830530166626,
      "learning_rate": 5.109096041533757e-08,
      "loss": 0.4095,
      "step": 30810
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6047849655151367,
      "learning_rate": 5.0988197484643784e-08,
      "loss": 0.3707,
      "step": 30811
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5093467235565186,
      "learning_rate": 5.0885537741112157e-08,
      "loss": 0.3867,
      "step": 30812
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5465037822723389,
      "learning_rate": 5.078298118580738e-08,
      "loss": 0.4533,
      "step": 30813
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5637047290802002,
      "learning_rate": 5.068052781979194e-08,
      "loss": 0.3777,
      "step": 30814
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5687830448150635,
      "learning_rate": 5.0578177644129426e-08,
      "loss": 0.4119,
      "step": 30815
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.470882534980774,
      "learning_rate": 5.047593065988121e-08,
      "loss": 0.3554,
      "step": 30816
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6033304929733276,
      "learning_rate": 5.0373786868107567e-08,
      "loss": 0.4273,
      "step": 30817
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5262490510940552,
      "learning_rate": 5.027174626986875e-08,
      "loss": 0.3525,
      "step": 30818
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5545367002487183,
      "learning_rate": 5.0169808866221695e-08,
      "loss": 0.3952,
      "step": 30819
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4677304029464722,
      "learning_rate": 5.006797465822444e-08,
      "loss": 0.3551,
      "step": 30820
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5491092205047607,
      "learning_rate": 4.996624364693281e-08,
      "loss": 0.3889,
      "step": 30821
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4698165655136108,
      "learning_rate": 4.986461583340263e-08,
      "loss": 0.3528,
      "step": 30822
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5696450471878052,
      "learning_rate": 4.9763091218687495e-08,
      "loss": 0.4652,
      "step": 30823
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5190099477767944,
      "learning_rate": 4.966166980383991e-08,
      "loss": 0.3918,
      "step": 30824
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6246881484985352,
      "learning_rate": 4.956035158991235e-08,
      "loss": 0.414,
      "step": 30825
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5268230438232422,
      "learning_rate": 4.9459136577955094e-08,
      "loss": 0.3508,
      "step": 30826
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5593945980072021,
      "learning_rate": 4.9358024769018406e-08,
      "loss": 0.4106,
      "step": 30827
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4775344133377075,
      "learning_rate": 4.925701616415146e-08,
      "loss": 0.3558,
      "step": 30828
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5473169088363647,
      "learning_rate": 4.9156110764401186e-08,
      "loss": 0.39,
      "step": 30829
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5332454442977905,
      "learning_rate": 4.905530857081453e-08,
      "loss": 0.3703,
      "step": 30830
    },
    {
      "epoch": 0.97,
      "grad_norm": 12.319212913513184,
      "learning_rate": 4.8954609584436205e-08,
      "loss": 0.4771,
      "step": 30831
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5651098489761353,
      "learning_rate": 4.885401380631094e-08,
      "loss": 0.3534,
      "step": 30832
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.558588981628418,
      "learning_rate": 4.875352123748234e-08,
      "loss": 0.3675,
      "step": 30833
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5007113218307495,
      "learning_rate": 4.86531318789929e-08,
      "loss": 0.3859,
      "step": 30834
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.73867666721344,
      "learning_rate": 4.8552845731882906e-08,
      "loss": 0.3936,
      "step": 30835
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.450129747390747,
      "learning_rate": 4.845266279719374e-08,
      "loss": 0.3633,
      "step": 30836
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5163143873214722,
      "learning_rate": 4.835258307596347e-08,
      "loss": 0.4318,
      "step": 30837
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4657524824142456,
      "learning_rate": 4.8252606569230144e-08,
      "loss": 0.3621,
      "step": 30838
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.529237151145935,
      "learning_rate": 4.815273327803183e-08,
      "loss": 0.4076,
      "step": 30839
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4710326194763184,
      "learning_rate": 4.8052963203402136e-08,
      "loss": 0.3342,
      "step": 30840
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5112344026565552,
      "learning_rate": 4.795329634637802e-08,
      "loss": 0.3794,
      "step": 30841
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5108532905578613,
      "learning_rate": 4.7853732707991986e-08,
      "loss": 0.3575,
      "step": 30842
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5001139640808105,
      "learning_rate": 4.7754272289277646e-08,
      "loss": 0.4279,
      "step": 30843
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5419143438339233,
      "learning_rate": 4.7654915091265294e-08,
      "loss": 0.3863,
      "step": 30844
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6238105297088623,
      "learning_rate": 4.755566111498633e-08,
      "loss": 0.3738,
      "step": 30845
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4776216745376587,
      "learning_rate": 4.745651036146992e-08,
      "loss": 0.379,
      "step": 30846
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.525638222694397,
      "learning_rate": 4.735746283174525e-08,
      "loss": 0.3694,
      "step": 30847
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4927053451538086,
      "learning_rate": 4.725851852683816e-08,
      "loss": 0.3791,
      "step": 30848
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.566583275794983,
      "learning_rate": 4.71596774477745e-08,
      "loss": 0.3768,
      "step": 30849
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6115041971206665,
      "learning_rate": 4.706093959558122e-08,
      "loss": 0.3711,
      "step": 30850
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7141509056091309,
      "learning_rate": 4.696230497128196e-08,
      "loss": 0.9628,
      "step": 30851
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4562571048736572,
      "learning_rate": 4.686377357589811e-08,
      "loss": 1.0005,
      "step": 30852
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5192723274230957,
      "learning_rate": 4.676534541045441e-08,
      "loss": 0.4373,
      "step": 30853
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4996200799942017,
      "learning_rate": 4.666702047596894e-08,
      "loss": 0.3544,
      "step": 30854
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4974743127822876,
      "learning_rate": 4.65687987734631e-08,
      "loss": 0.4042,
      "step": 30855
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4602774381637573,
      "learning_rate": 4.6470680303954964e-08,
      "loss": 0.386,
      "step": 30856
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4633419513702393,
      "learning_rate": 4.6372665068462606e-08,
      "loss": 0.3637,
      "step": 30857
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5081537961959839,
      "learning_rate": 4.627475306800189e-08,
      "loss": 0.338,
      "step": 30858
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5783910751342773,
      "learning_rate": 4.617694430358976e-08,
      "loss": 0.4148,
      "step": 30859
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5271391868591309,
      "learning_rate": 4.607923877623877e-08,
      "loss": 0.3689,
      "step": 30860
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.573203682899475,
      "learning_rate": 4.598163648696252e-08,
      "loss": 0.4215,
      "step": 30861
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4962726831436157,
      "learning_rate": 4.588413743677578e-08,
      "loss": 0.3386,
      "step": 30862
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5676450729370117,
      "learning_rate": 4.5786741626686615e-08,
      "loss": 0.4152,
      "step": 30863
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5057454109191895,
      "learning_rate": 4.5689449057707555e-08,
      "loss": 0.3812,
      "step": 30864
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.545142412185669,
      "learning_rate": 4.559225973084558e-08,
      "loss": 0.4094,
      "step": 30865
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5185855627059937,
      "learning_rate": 4.549517364711098e-08,
      "loss": 0.3679,
      "step": 30866
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6445108652114868,
      "learning_rate": 4.539819080750851e-08,
      "loss": 0.4348,
      "step": 30867
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4674365520477295,
      "learning_rate": 4.530131121304515e-08,
      "loss": 0.3673,
      "step": 30868
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4690839052200317,
      "learning_rate": 4.520453486472565e-08,
      "loss": 0.4508,
      "step": 30869
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4102764129638672,
      "learning_rate": 4.510786176355475e-08,
      "loss": 0.3429,
      "step": 30870
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.495468258857727,
      "learning_rate": 4.501129191053277e-08,
      "loss": 0.3898,
      "step": 30871
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4339256286621094,
      "learning_rate": 4.4914825306662246e-08,
      "loss": 0.3515,
      "step": 30872
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5982484817504883,
      "learning_rate": 4.481846195294459e-08,
      "loss": 0.423,
      "step": 30873
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7457185983657837,
      "learning_rate": 4.472220185037901e-08,
      "loss": 0.4011,
      "step": 30874
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5404040813446045,
      "learning_rate": 4.462604499996248e-08,
      "loss": 0.3946,
      "step": 30875
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4970327615737915,
      "learning_rate": 4.45299914026942e-08,
      "loss": 0.413,
      "step": 30876
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5989480018615723,
      "learning_rate": 4.4434041059568944e-08,
      "loss": 0.3752,
      "step": 30877
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6346319913864136,
      "learning_rate": 4.4338193971582566e-08,
      "loss": 0.3774,
      "step": 30878
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5783549547195435,
      "learning_rate": 4.424245013972983e-08,
      "loss": 0.4047,
      "step": 30879
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5355405807495117,
      "learning_rate": 4.414680956500217e-08,
      "loss": 0.3678,
      "step": 30880
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4726319313049316,
      "learning_rate": 4.4051272248392117e-08,
      "loss": 0.3949,
      "step": 30881
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.521986722946167,
      "learning_rate": 4.395583819089111e-08,
      "loss": 0.378,
      "step": 30882
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5751608610153198,
      "learning_rate": 4.386050739348835e-08,
      "loss": 0.4541,
      "step": 30883
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6248224973678589,
      "learning_rate": 4.376527985717305e-08,
      "loss": 0.3795,
      "step": 30884
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.2787585258483887,
      "learning_rate": 4.36701555829322e-08,
      "loss": 0.4604,
      "step": 30885
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.516170859336853,
      "learning_rate": 4.3575134571753884e-08,
      "loss": 0.3918,
      "step": 30886
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6344469785690308,
      "learning_rate": 4.3480216824620666e-08,
      "loss": 0.4517,
      "step": 30887
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.586963176727295,
      "learning_rate": 4.3385402342520646e-08,
      "loss": 0.3758,
      "step": 30888
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5636557340621948,
      "learning_rate": 4.329069112643414e-08,
      "loss": 0.3981,
      "step": 30889
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5888866186141968,
      "learning_rate": 4.319608317734592e-08,
      "loss": 0.3771,
      "step": 30890
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5951480865478516,
      "learning_rate": 4.3101578496235194e-08,
      "loss": 0.4326,
      "step": 30891
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5945764780044556,
      "learning_rate": 4.3007177084083415e-08,
      "loss": 0.39,
      "step": 30892
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5856642723083496,
      "learning_rate": 4.2912878941868684e-08,
      "loss": 0.4592,
      "step": 30893
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4945549964904785,
      "learning_rate": 4.281868407057022e-08,
      "loss": 0.3685,
      "step": 30894
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6597793102264404,
      "learning_rate": 4.2724592471165005e-08,
      "loss": 0.4427,
      "step": 30895
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6563447713851929,
      "learning_rate": 4.263060414462672e-08,
      "loss": 0.3644,
      "step": 30896
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.544937252998352,
      "learning_rate": 4.2536719091932356e-08,
      "loss": 0.3858,
      "step": 30897
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5793037414550781,
      "learning_rate": 4.244293731405558e-08,
      "loss": 0.3669,
      "step": 30898
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5319466590881348,
      "learning_rate": 4.234925881196783e-08,
      "loss": 0.4427,
      "step": 30899
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5823516845703125,
      "learning_rate": 4.225568358664167e-08,
      "loss": 0.3785,
      "step": 30900
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6700018644332886,
      "learning_rate": 4.2162211639047437e-08,
      "loss": 0.4473,
      "step": 30901
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5570062398910522,
      "learning_rate": 4.206884297015434e-08,
      "loss": 0.4044,
      "step": 30902
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.679733395576477,
      "learning_rate": 4.197557758093052e-08,
      "loss": 0.4654,
      "step": 30903
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4617587327957153,
      "learning_rate": 4.1882415472342954e-08,
      "loss": 0.3676,
      "step": 30904
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.572084903717041,
      "learning_rate": 4.178935664535977e-08,
      "loss": 0.4269,
      "step": 30905
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4476912021636963,
      "learning_rate": 4.169640110094575e-08,
      "loss": 0.3821,
      "step": 30906
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7115906476974487,
      "learning_rate": 4.1603548840062344e-08,
      "loss": 0.3928,
      "step": 30907
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4968475103378296,
      "learning_rate": 4.1510799863675456e-08,
      "loss": 0.3803,
      "step": 30908
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5625096559524536,
      "learning_rate": 4.141815417274542e-08,
      "loss": 0.4638,
      "step": 30909
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6250149011611938,
      "learning_rate": 4.1325611768233685e-08,
      "loss": 0.3668,
      "step": 30910
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8064382076263428,
      "learning_rate": 4.1233172651100606e-08,
      "loss": 0.9522,
      "step": 30911
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.460679054260254,
      "learning_rate": 4.1140836822304295e-08,
      "loss": 1.0332,
      "step": 30912
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5712916851043701,
      "learning_rate": 4.1048604282801774e-08,
      "loss": 0.4161,
      "step": 30913
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4680066108703613,
      "learning_rate": 4.0956475033551156e-08,
      "loss": 0.3568,
      "step": 30914
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5749375820159912,
      "learning_rate": 4.086444907550724e-08,
      "loss": 0.4314,
      "step": 30915
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5763742923736572,
      "learning_rate": 4.0772526409624814e-08,
      "loss": 0.3789,
      "step": 30916
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.556989073753357,
      "learning_rate": 4.068070703685645e-08,
      "loss": 0.4589,
      "step": 30917
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4453295469284058,
      "learning_rate": 4.058899095815472e-08,
      "loss": 0.357,
      "step": 30918
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1931605339050293,
      "learning_rate": 4.049737817447108e-08,
      "loss": 0.4193,
      "step": 30919
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4805933237075806,
      "learning_rate": 4.0405868686755886e-08,
      "loss": 0.357,
      "step": 30920
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5625642538070679,
      "learning_rate": 4.0314462495958385e-08,
      "loss": 0.4068,
      "step": 30921
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5478373765945435,
      "learning_rate": 4.022315960302559e-08,
      "loss": 0.3862,
      "step": 30922
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5897849798202515,
      "learning_rate": 4.013196000890451e-08,
      "loss": 0.4443,
      "step": 30923
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4906749725341797,
      "learning_rate": 4.0040863714543297e-08,
      "loss": 0.35,
      "step": 30924
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6830167770385742,
      "learning_rate": 3.994987072088452e-08,
      "loss": 0.4301,
      "step": 30925
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5099983215332031,
      "learning_rate": 3.985898102887187e-08,
      "loss": 0.3555,
      "step": 30926
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.649074673652649,
      "learning_rate": 3.976819463944903e-08,
      "loss": 0.4367,
      "step": 30927
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.492551326751709,
      "learning_rate": 3.967751155355748e-08,
      "loss": 0.3537,
      "step": 30928
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5301390886306763,
      "learning_rate": 3.958693177213646e-08,
      "loss": 0.4056,
      "step": 30929
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.561610460281372,
      "learning_rate": 3.9496455296127444e-08,
      "loss": 0.3828,
      "step": 30930
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5462753772735596,
      "learning_rate": 3.940608212646857e-08,
      "loss": 0.428,
      "step": 30931
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4927538633346558,
      "learning_rate": 3.931581226409464e-08,
      "loss": 0.3696,
      "step": 30932
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8547968864440918,
      "learning_rate": 3.922564570994491e-08,
      "loss": 0.9697,
      "step": 30933
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5219147205352783,
      "learning_rate": 3.913558246495308e-08,
      "loss": 1.061,
      "step": 30934
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.612657904624939,
      "learning_rate": 3.904562253005284e-08,
      "loss": 0.4349,
      "step": 30935
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4945584535598755,
      "learning_rate": 3.8955765906179e-08,
      "loss": 0.3684,
      "step": 30936
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5474700927734375,
      "learning_rate": 3.886601259426193e-08,
      "loss": 0.383,
      "step": 30937
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5099011659622192,
      "learning_rate": 3.87763625952331e-08,
      "loss": 0.3657,
      "step": 30938
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4805601835250854,
      "learning_rate": 3.868681591002177e-08,
      "loss": 0.4355,
      "step": 30939
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4807121753692627,
      "learning_rate": 3.859737253955609e-08,
      "loss": 0.3529,
      "step": 30940
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.499359369277954,
      "learning_rate": 3.8508032484766425e-08,
      "loss": 0.394,
      "step": 30941
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6295019388198853,
      "learning_rate": 3.8418795746577585e-08,
      "loss": 0.3581,
      "step": 30942
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6399563550949097,
      "learning_rate": 3.8329662325914394e-08,
      "loss": 0.4718,
      "step": 30943
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5244840383529663,
      "learning_rate": 3.824063222370278e-08,
      "loss": 0.3732,
      "step": 30944
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5844314098358154,
      "learning_rate": 3.815170544086533e-08,
      "loss": 0.4752,
      "step": 30945
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4980610609054565,
      "learning_rate": 3.806288197832464e-08,
      "loss": 0.3629,
      "step": 30946
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.569919466972351,
      "learning_rate": 3.797416183700109e-08,
      "loss": 0.4413,
      "step": 30947
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5117238759994507,
      "learning_rate": 3.788554501781616e-08,
      "loss": 0.37,
      "step": 30948
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5329012870788574,
      "learning_rate": 3.779703152168801e-08,
      "loss": 0.45,
      "step": 30949
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.486342191696167,
      "learning_rate": 3.77086213495359e-08,
      "loss": 0.3627,
      "step": 30950
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5497550964355469,
      "learning_rate": 3.762031450227466e-08,
      "loss": 0.4167,
      "step": 30951
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5506055355072021,
      "learning_rate": 3.7532110980822433e-08,
      "loss": 0.3753,
      "step": 30952
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4847172498703003,
      "learning_rate": 3.7444010786091836e-08,
      "loss": 0.4301,
      "step": 30953
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5037903785705566,
      "learning_rate": 3.7356013918998793e-08,
      "loss": 0.3298,
      "step": 30954
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.586164951324463,
      "learning_rate": 3.72681203804548e-08,
      "loss": 0.4241,
      "step": 30955
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5090652704238892,
      "learning_rate": 3.718033017137135e-08,
      "loss": 0.3686,
      "step": 30956
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5588935613632202,
      "learning_rate": 3.709264329265994e-08,
      "loss": 0.397,
      "step": 30957
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4770079851150513,
      "learning_rate": 3.700505974522872e-08,
      "loss": 0.3827,
      "step": 30958
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5290477275848389,
      "learning_rate": 3.691757952998698e-08,
      "loss": 0.4189,
      "step": 30959
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5432227849960327,
      "learning_rate": 3.6830202647840654e-08,
      "loss": 0.3824,
      "step": 30960
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5351964235305786,
      "learning_rate": 3.6742929099697896e-08,
      "loss": 0.4244,
      "step": 30961
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4972954988479614,
      "learning_rate": 3.665575888646356e-08,
      "loss": 0.3605,
      "step": 30962
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5458563566207886,
      "learning_rate": 3.656869200904023e-08,
      "loss": 0.3914,
      "step": 30963
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5185210704803467,
      "learning_rate": 3.6481728468331645e-08,
      "loss": 0.374,
      "step": 30964
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5446922779083252,
      "learning_rate": 3.6394868265241525e-08,
      "loss": 0.422,
      "step": 30965
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4896739721298218,
      "learning_rate": 3.6308111400668033e-08,
      "loss": 0.3831,
      "step": 30966
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6254879236221313,
      "learning_rate": 3.622145787551268e-08,
      "loss": 0.4743,
      "step": 30967
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4946545362472534,
      "learning_rate": 3.613490769067363e-08,
      "loss": 0.3585,
      "step": 30968
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5475841760635376,
      "learning_rate": 3.6048460847047936e-08,
      "loss": 0.4532,
      "step": 30969
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5582302808761597,
      "learning_rate": 3.5962117345533785e-08,
      "loss": 0.3923,
      "step": 30970
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5524241924285889,
      "learning_rate": 3.58758771870249e-08,
      "loss": 0.4362,
      "step": 30971
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5326789617538452,
      "learning_rate": 3.578974037241723e-08,
      "loss": 0.3686,
      "step": 30972
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.6584105491638184,
      "learning_rate": 3.570370690260338e-08,
      "loss": 0.4609,
      "step": 30973
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5332220792770386,
      "learning_rate": 3.5617776778475996e-08,
      "loss": 0.3824,
      "step": 30974
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.512385368347168,
      "learning_rate": 3.5531950000926575e-08,
      "loss": 0.4137,
      "step": 30975
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6120706796646118,
      "learning_rate": 3.544622657084329e-08,
      "loss": 0.384,
      "step": 30976
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7628042697906494,
      "learning_rate": 3.5360606489117655e-08,
      "loss": 0.43,
      "step": 30977
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5545012950897217,
      "learning_rate": 3.527508975663785e-08,
      "loss": 0.347,
      "step": 30978
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6996771097183228,
      "learning_rate": 3.5189676374288714e-08,
      "loss": 0.9374,
      "step": 30979
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4558435678482056,
      "learning_rate": 3.510436634295622e-08,
      "loss": 0.9877,
      "step": 30980
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6021567583084106,
      "learning_rate": 3.501915966352743e-08,
      "loss": 0.4507,
      "step": 30981
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5307323932647705,
      "learning_rate": 3.493405633688496e-08,
      "loss": 0.3974,
      "step": 30982
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5843045711517334,
      "learning_rate": 3.4849056363910336e-08,
      "loss": 0.3925,
      "step": 30983
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5379981994628906,
      "learning_rate": 3.476415974548619e-08,
      "loss": 0.3748,
      "step": 30984
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.731542706489563,
      "learning_rate": 3.467936648249293e-08,
      "loss": 0.4556,
      "step": 30985
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5909193754196167,
      "learning_rate": 3.459467657581095e-08,
      "loss": 0.4056,
      "step": 30986
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.668660044670105,
      "learning_rate": 3.451009002631733e-08,
      "loss": 0.4397,
      "step": 30987
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5019153356552124,
      "learning_rate": 3.4425606834889155e-08,
      "loss": 0.3565,
      "step": 30988
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5171384811401367,
      "learning_rate": 3.43412270024035e-08,
      "loss": 0.4098,
      "step": 30989
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5243678092956543,
      "learning_rate": 3.425695052973521e-08,
      "loss": 0.3539,
      "step": 30990
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.612686038017273,
      "learning_rate": 3.417277741775804e-08,
      "loss": 0.445,
      "step": 30991
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4548170566558838,
      "learning_rate": 3.408870766734573e-08,
      "loss": 0.3542,
      "step": 30992
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8517177104949951,
      "learning_rate": 3.4004741279369815e-08,
      "loss": 0.4571,
      "step": 30993
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5056207180023193,
      "learning_rate": 3.3920878254700694e-08,
      "loss": 0.3717,
      "step": 30994
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6127244234085083,
      "learning_rate": 3.38371185942099e-08,
      "loss": 0.4379,
      "step": 30995
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4980995655059814,
      "learning_rate": 3.3753462298763416e-08,
      "loss": 0.3947,
      "step": 30996
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5095620155334473,
      "learning_rate": 3.366990936923053e-08,
      "loss": 0.4141,
      "step": 30997
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.48152494430542,
      "learning_rate": 3.3586459806478344e-08,
      "loss": 0.3714,
      "step": 30998
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.669228196144104,
      "learning_rate": 3.3503113611371704e-08,
      "loss": 0.428,
      "step": 30999
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5528150796890259,
      "learning_rate": 3.341987078477438e-08,
      "loss": 0.4009,
      "step": 31000
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5556927919387817,
      "learning_rate": 3.333673132755011e-08,
      "loss": 0.4285,
      "step": 31001
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5068901777267456,
      "learning_rate": 3.325369524056266e-08,
      "loss": 0.3577,
      "step": 31002
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5412555932998657,
      "learning_rate": 3.317076252467133e-08,
      "loss": 0.4534,
      "step": 31003
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.566426396369934,
      "learning_rate": 3.308793318073767e-08,
      "loss": 0.3704,
      "step": 31004
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5274310111999512,
      "learning_rate": 3.300520720961875e-08,
      "loss": 0.4118,
      "step": 31005
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.504982590675354,
      "learning_rate": 3.292258461217501e-08,
      "loss": 0.3786,
      "step": 31006
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.672227382659912,
      "learning_rate": 3.284006538926132e-08,
      "loss": 0.4404,
      "step": 31007
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.502361536026001,
      "learning_rate": 3.275764954173477e-08,
      "loss": 0.3422,
      "step": 31008
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5608129501342773,
      "learning_rate": 3.267533707045023e-08,
      "loss": 0.43,
      "step": 31009
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4964064359664917,
      "learning_rate": 3.2593127976260354e-08,
      "loss": 0.389,
      "step": 31010
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4998868703842163,
      "learning_rate": 3.25110222600189e-08,
      "loss": 0.401,
      "step": 31011
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4851230382919312,
      "learning_rate": 3.24290199225763e-08,
      "loss": 0.3681,
      "step": 31012
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.555220127105713,
      "learning_rate": 3.2347120964784096e-08,
      "loss": 0.4581,
      "step": 31013
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4964200258255005,
      "learning_rate": 3.2265325387491606e-08,
      "loss": 0.3759,
      "step": 31014
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4808030128479004,
      "learning_rate": 3.218363319154705e-08,
      "loss": 0.3817,
      "step": 31015
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5292067527770996,
      "learning_rate": 3.210204437779751e-08,
      "loss": 0.377,
      "step": 31016
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5281506776809692,
      "learning_rate": 3.2020558947088996e-08,
      "loss": 0.4583,
      "step": 31017
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5641342401504517,
      "learning_rate": 3.1939176900266376e-08,
      "loss": 0.4076,
      "step": 31018
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5558115243911743,
      "learning_rate": 3.1857898238175646e-08,
      "loss": 0.4115,
      "step": 31019
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5030701160430908,
      "learning_rate": 3.1776722961657235e-08,
      "loss": 0.342,
      "step": 31020
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5283621549606323,
      "learning_rate": 3.1695651071553816e-08,
      "loss": 0.4126,
      "step": 31021
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4815590381622314,
      "learning_rate": 3.161468256870803e-08,
      "loss": 0.369,
      "step": 31022
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5441761016845703,
      "learning_rate": 3.1533817453957006e-08,
      "loss": 0.4254,
      "step": 31023
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6625195741653442,
      "learning_rate": 3.145305572814117e-08,
      "loss": 0.3451,
      "step": 31024
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5062437057495117,
      "learning_rate": 3.137239739209763e-08,
      "loss": 0.4013,
      "step": 31025
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5375317335128784,
      "learning_rate": 3.12918424466635e-08,
      "loss": 0.3957,
      "step": 31026
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6587320566177368,
      "learning_rate": 3.121139089267256e-08,
      "loss": 0.4097,
      "step": 31027
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.524489164352417,
      "learning_rate": 3.113104273096079e-08,
      "loss": 0.3602,
      "step": 31028
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5682684183120728,
      "learning_rate": 3.105079796236199e-08,
      "loss": 0.4267,
      "step": 31029
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4255633354187012,
      "learning_rate": 3.0970656587707705e-08,
      "loss": 0.3341,
      "step": 31030
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5540715456008911,
      "learning_rate": 3.089061860782727e-08,
      "loss": 0.4087,
      "step": 31031
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5174287557601929,
      "learning_rate": 3.081068402355336e-08,
      "loss": 0.3647,
      "step": 31032
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.55918288230896,
      "learning_rate": 3.0730852835714196e-08,
      "loss": 0.4485,
      "step": 31033
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4867136478424072,
      "learning_rate": 3.065112504513912e-08,
      "loss": 0.3576,
      "step": 31034
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5624794960021973,
      "learning_rate": 3.0571500652651906e-08,
      "loss": 0.4044,
      "step": 31035
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5649160146713257,
      "learning_rate": 3.049197965908079e-08,
      "loss": 0.3977,
      "step": 31036
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5765807628631592,
      "learning_rate": 3.0412562065249564e-08,
      "loss": 0.416,
      "step": 31037
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5145361423492432,
      "learning_rate": 3.0333247871983104e-08,
      "loss": 0.3616,
      "step": 31038
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5356159210205078,
      "learning_rate": 3.0254037080102995e-08,
      "loss": 0.4402,
      "step": 31039
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.683219313621521,
      "learning_rate": 3.0174929690429675e-08,
      "loss": 0.3857,
      "step": 31040
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5122309923171997,
      "learning_rate": 3.009592570378583e-08,
      "loss": 0.4769,
      "step": 31041
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5455912351608276,
      "learning_rate": 3.00170251209897e-08,
      "loss": 0.3641,
      "step": 31042
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4625191688537598,
      "learning_rate": 2.9938227942860606e-08,
      "loss": 0.3918,
      "step": 31043
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4558814764022827,
      "learning_rate": 2.985953417021459e-08,
      "loss": 0.3523,
      "step": 31044
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5102777481079102,
      "learning_rate": 2.9780943803867644e-08,
      "loss": 0.426,
      "step": 31045
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.482403039932251,
      "learning_rate": 2.9702456844635798e-08,
      "loss": 0.3699,
      "step": 31046
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5524259805679321,
      "learning_rate": 2.9624073293332835e-08,
      "loss": 0.4039,
      "step": 31047
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5097992420196533,
      "learning_rate": 2.9545793150772552e-08,
      "loss": 0.3529,
      "step": 31048
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5733070373535156,
      "learning_rate": 2.946761641776541e-08,
      "loss": 0.4108,
      "step": 31049
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5206358432769775,
      "learning_rate": 2.9389543095121876e-08,
      "loss": 0.372,
      "step": 31050
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5721526145935059,
      "learning_rate": 2.931157318365352e-08,
      "loss": 0.4242,
      "step": 31051
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4651249647140503,
      "learning_rate": 2.923370668416747e-08,
      "loss": 0.3531,
      "step": 31052
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6543033123016357,
      "learning_rate": 2.9155943597471982e-08,
      "loss": 0.3987,
      "step": 31053
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4698076248168945,
      "learning_rate": 2.9078283924374174e-08,
      "loss": 0.3536,
      "step": 31054
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5876117944717407,
      "learning_rate": 2.9000727665678963e-08,
      "loss": 0.4291,
      "step": 31055
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5987111330032349,
      "learning_rate": 2.892327482219015e-08,
      "loss": 0.3881,
      "step": 31056
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5203746557235718,
      "learning_rate": 2.8845925394711538e-08,
      "loss": 0.4032,
      "step": 31057
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4588626623153687,
      "learning_rate": 2.8768679384045818e-08,
      "loss": 0.3452,
      "step": 31058
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5436503887176514,
      "learning_rate": 2.8691536790993457e-08,
      "loss": 0.4214,
      "step": 31059
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5491302013397217,
      "learning_rate": 2.861449761635382e-08,
      "loss": 0.3737,
      "step": 31060
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6436823606491089,
      "learning_rate": 2.8537561860928486e-08,
      "loss": 0.4236,
      "step": 31061
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5915759801864624,
      "learning_rate": 2.8460729525513487e-08,
      "loss": 0.4024,
      "step": 31062
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.581099510192871,
      "learning_rate": 2.838400061090485e-08,
      "loss": 0.3892,
      "step": 31063
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5072591304779053,
      "learning_rate": 2.8307375117899717e-08,
      "loss": 0.3809,
      "step": 31064
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5606834888458252,
      "learning_rate": 2.8230853047294117e-08,
      "loss": 0.4508,
      "step": 31065
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5175212621688843,
      "learning_rate": 2.8154434399879637e-08,
      "loss": 0.3647,
      "step": 31066
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5537899732589722,
      "learning_rate": 2.807811917644898e-08,
      "loss": 0.4612,
      "step": 31067
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5123276710510254,
      "learning_rate": 2.8001907377793736e-08,
      "loss": 0.3728,
      "step": 31068
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6426796913146973,
      "learning_rate": 2.792579900470549e-08,
      "loss": 0.437,
      "step": 31069
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.461003303527832,
      "learning_rate": 2.784979405797361e-08,
      "loss": 0.3463,
      "step": 31070
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5240137577056885,
      "learning_rate": 2.777389253838414e-08,
      "loss": 0.3828,
      "step": 31071
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6106042861938477,
      "learning_rate": 2.769809444672644e-08,
      "loss": 0.4308,
      "step": 31072
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.556908130645752,
      "learning_rate": 2.7622399783785446e-08,
      "loss": 0.4173,
      "step": 31073
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5566442012786865,
      "learning_rate": 2.7546808550348302e-08,
      "loss": 0.3585,
      "step": 31074
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5776292085647583,
      "learning_rate": 2.7471320747196606e-08,
      "loss": 0.4438,
      "step": 31075
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4977186918258667,
      "learning_rate": 2.7395936375114174e-08,
      "loss": 0.3505,
      "step": 31076
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4630954265594482,
      "learning_rate": 2.7320655434882603e-08,
      "loss": 0.3948,
      "step": 31077
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5426112413406372,
      "learning_rate": 2.7245477927283492e-08,
      "loss": 0.3537,
      "step": 31078
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6241403818130493,
      "learning_rate": 2.7170403853095106e-08,
      "loss": 0.4128,
      "step": 31079
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4867479801177979,
      "learning_rate": 2.7095433213097933e-08,
      "loss": 0.3681,
      "step": 31080
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5211948156356812,
      "learning_rate": 2.7020566008068017e-08,
      "loss": 0.3885,
      "step": 31081
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.824447751045227,
      "learning_rate": 2.6945802238782514e-08,
      "loss": 0.3755,
      "step": 31082
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5714625120162964,
      "learning_rate": 2.6871141906017474e-08,
      "loss": 0.4125,
      "step": 31083
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.519004225730896,
      "learning_rate": 2.6796585010545605e-08,
      "loss": 0.3633,
      "step": 31084
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5823863744735718,
      "learning_rate": 2.672213155314185e-08,
      "loss": 0.9812,
      "step": 31085
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3948633670806885,
      "learning_rate": 2.664778153457781e-08,
      "loss": 0.9552,
      "step": 31086
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6456328630447388,
      "learning_rate": 2.657353495562398e-08,
      "loss": 0.4124,
      "step": 31087
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4862172603607178,
      "learning_rate": 2.6499391817050856e-08,
      "loss": 0.3646,
      "step": 31088
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5845001935958862,
      "learning_rate": 2.6425352119627823e-08,
      "loss": 0.4958,
      "step": 31089
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5052909851074219,
      "learning_rate": 2.6351415864122044e-08,
      "loss": 0.3511,
      "step": 31090
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6154232025146484,
      "learning_rate": 2.627758305130068e-08,
      "loss": 0.4845,
      "step": 31091
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5125854015350342,
      "learning_rate": 2.6203853681929792e-08,
      "loss": 0.3594,
      "step": 31092
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5362907648086548,
      "learning_rate": 2.6130227756774317e-08,
      "loss": 0.4162,
      "step": 31093
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.524375319480896,
      "learning_rate": 2.6056705276595874e-08,
      "loss": 0.386,
      "step": 31094
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6011581420898438,
      "learning_rate": 2.598328624216051e-08,
      "loss": 0.4225,
      "step": 31095
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5766358375549316,
      "learning_rate": 2.5909970654226512e-08,
      "loss": 0.3704,
      "step": 31096
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.664088249206543,
      "learning_rate": 2.5836758513555494e-08,
      "loss": 0.4311,
      "step": 31097
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5392320156097412,
      "learning_rate": 2.5763649820906844e-08,
      "loss": 0.4039,
      "step": 31098
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5680691003799438,
      "learning_rate": 2.569064457703996e-08,
      "loss": 0.4279,
      "step": 31099
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.583051085472107,
      "learning_rate": 2.5617742782709787e-08,
      "loss": 0.4135,
      "step": 31100
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4823851585388184,
      "learning_rate": 2.5544944438673503e-08,
      "loss": 0.4012,
      "step": 31101
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5324400663375854,
      "learning_rate": 2.5472249545686055e-08,
      "loss": 0.3707,
      "step": 31102
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.518906593322754,
      "learning_rate": 2.539965810450129e-08,
      "loss": 0.4029,
      "step": 31103
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5087345838546753,
      "learning_rate": 2.5327170115873045e-08,
      "loss": 0.385,
      "step": 31104
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.520830750465393,
      "learning_rate": 2.5254785580551834e-08,
      "loss": 0.3776,
      "step": 31105
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5194003582000732,
      "learning_rate": 2.5182504499288162e-08,
      "loss": 0.3877,
      "step": 31106
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.520765781402588,
      "learning_rate": 2.5110326872833656e-08,
      "loss": 0.4071,
      "step": 31107
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.457030177116394,
      "learning_rate": 2.5038252701935495e-08,
      "loss": 0.3705,
      "step": 31108
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6427803039550781,
      "learning_rate": 2.4966281987340858e-08,
      "loss": 0.4005,
      "step": 31109
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.486373782157898,
      "learning_rate": 2.4894414729796924e-08,
      "loss": 0.366,
      "step": 31110
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5316762924194336,
      "learning_rate": 2.4822650930049764e-08,
      "loss": 0.4278,
      "step": 31111
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5135136842727661,
      "learning_rate": 2.4750990588842116e-08,
      "loss": 0.3746,
      "step": 31112
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5430150032043457,
      "learning_rate": 2.4679433706917834e-08,
      "loss": 0.4058,
      "step": 31113
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6035659313201904,
      "learning_rate": 2.460798028501965e-08,
      "loss": 0.3734,
      "step": 31114
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.0212714672088623,
      "learning_rate": 2.453663032388809e-08,
      "loss": 0.417,
      "step": 31115
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5334100723266602,
      "learning_rate": 2.4465383824262555e-08,
      "loss": 0.3642,
      "step": 31116
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4877510070800781,
      "learning_rate": 2.4394240786883573e-08,
      "loss": 0.4052,
      "step": 31117
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5412472486495972,
      "learning_rate": 2.4323201212487215e-08,
      "loss": 0.379,
      "step": 31118
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6367664337158203,
      "learning_rate": 2.4252265101811778e-08,
      "loss": 0.4925,
      "step": 31119
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.470571517944336,
      "learning_rate": 2.4181432455592237e-08,
      "loss": 0.372,
      "step": 31120
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6708464622497559,
      "learning_rate": 2.411070327456355e-08,
      "loss": 0.406,
      "step": 31121
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4836517572402954,
      "learning_rate": 2.4040077559458476e-08,
      "loss": 0.3659,
      "step": 31122
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6059809923171997,
      "learning_rate": 2.396955531101086e-08,
      "loss": 0.9223,
      "step": 31123
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4261466264724731,
      "learning_rate": 2.389913652995124e-08,
      "loss": 1.007,
      "step": 31124
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5544573068618774,
      "learning_rate": 2.3828821217010133e-08,
      "loss": 0.4469,
      "step": 31125
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.473127007484436,
      "learning_rate": 2.3758609372916964e-08,
      "loss": 0.3755,
      "step": 31126
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5484713315963745,
      "learning_rate": 2.3688500998400032e-08,
      "loss": 0.4398,
      "step": 31127
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4890013933181763,
      "learning_rate": 2.3618496094185428e-08,
      "loss": 0.3809,
      "step": 31128
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.557759404182434,
      "learning_rate": 2.3548594661001454e-08,
      "loss": 0.4125,
      "step": 31129
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5178993940353394,
      "learning_rate": 2.347879669956976e-08,
      "loss": 0.3853,
      "step": 31130
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5130449533462524,
      "learning_rate": 2.340910221061754e-08,
      "loss": 0.4028,
      "step": 31131
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5594635009765625,
      "learning_rate": 2.3339511194866437e-08,
      "loss": 0.3752,
      "step": 31132
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.511767864227295,
      "learning_rate": 2.32700236530381e-08,
      "loss": 0.4007,
      "step": 31133
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5533535480499268,
      "learning_rate": 2.320063958585306e-08,
      "loss": 0.3779,
      "step": 31134
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6449192762374878,
      "learning_rate": 2.313135899403074e-08,
      "loss": 0.3863,
      "step": 31135
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5382379293441772,
      "learning_rate": 2.306218187828946e-08,
      "loss": 0.3727,
      "step": 31136
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5144542455673218,
      "learning_rate": 2.299310823934864e-08,
      "loss": 0.4718,
      "step": 31137
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4652339220046997,
      "learning_rate": 2.2924138077922152e-08,
      "loss": 0.3651,
      "step": 31138
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4835549592971802,
      "learning_rate": 2.285527139472721e-08,
      "loss": 0.3997,
      "step": 31139
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6115820407867432,
      "learning_rate": 2.2786508190477675e-08,
      "loss": 0.3849,
      "step": 31140
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.516883134841919,
      "learning_rate": 2.2717848465886316e-08,
      "loss": 0.4063,
      "step": 31141
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6408445835113525,
      "learning_rate": 2.2649292221664787e-08,
      "loss": 0.3967,
      "step": 31142
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.498750925064087,
      "learning_rate": 2.258083945852474e-08,
      "loss": 0.4131,
      "step": 31143
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5502690076828003,
      "learning_rate": 2.2512490177176715e-08,
      "loss": 0.3722,
      "step": 31144
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5717076063156128,
      "learning_rate": 2.2444244378327928e-08,
      "loss": 0.4523,
      "step": 31145
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5688731670379639,
      "learning_rate": 2.237610206268781e-08,
      "loss": 0.3785,
      "step": 31146
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5985416173934937,
      "learning_rate": 2.230806323096357e-08,
      "loss": 0.4094,
      "step": 31147
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5068742036819458,
      "learning_rate": 2.2240127883857986e-08,
      "loss": 0.3624,
      "step": 31148
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7677550315856934,
      "learning_rate": 2.2172296022078265e-08,
      "loss": 1.0292,
      "step": 31149
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3865864276885986,
      "learning_rate": 2.210456764632718e-08,
      "loss": 0.9648,
      "step": 31150
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5354098081588745,
      "learning_rate": 2.20369427573075e-08,
      "loss": 0.4238,
      "step": 31151
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5420305728912354,
      "learning_rate": 2.1969421355720888e-08,
      "loss": 0.3602,
      "step": 31152
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7031441926956177,
      "learning_rate": 2.1902003442265674e-08,
      "loss": 0.9951,
      "step": 31153
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3913917541503906,
      "learning_rate": 2.1834689017643518e-08,
      "loss": 0.9329,
      "step": 31154
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5349440574645996,
      "learning_rate": 2.176747808255164e-08,
      "loss": 0.381,
      "step": 31155
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5780541896820068,
      "learning_rate": 2.170037063768615e-08,
      "loss": 0.3855,
      "step": 31156
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.515586256980896,
      "learning_rate": 2.1633366683744272e-08,
      "loss": 0.458,
      "step": 31157
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4771146774291992,
      "learning_rate": 2.156646622141989e-08,
      "loss": 0.3605,
      "step": 31158
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5211602449417114,
      "learning_rate": 2.1499669251409117e-08,
      "loss": 0.3945,
      "step": 31159
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4994449615478516,
      "learning_rate": 2.143297577440251e-08,
      "loss": 0.3425,
      "step": 31160
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5976896286010742,
      "learning_rate": 2.1366385791091736e-08,
      "loss": 0.4477,
      "step": 31161
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5414859056472778,
      "learning_rate": 2.129989930216958e-08,
      "loss": 0.343,
      "step": 31162
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5228095054626465,
      "learning_rate": 2.1233516308323266e-08,
      "loss": 0.4132,
      "step": 31163
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5304436683654785,
      "learning_rate": 2.1167236810242243e-08,
      "loss": 0.371,
      "step": 31164
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5753661394119263,
      "learning_rate": 2.1101060808613737e-08,
      "loss": 0.4075,
      "step": 31165
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5180299282073975,
      "learning_rate": 2.103498830412498e-08,
      "loss": 0.3667,
      "step": 31166
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6414252519607544,
      "learning_rate": 2.096901929745987e-08,
      "loss": 0.4564,
      "step": 31167
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5041483640670776,
      "learning_rate": 2.090315378930452e-08,
      "loss": 0.3924,
      "step": 31168
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.522899866104126,
      "learning_rate": 2.08373917803395e-08,
      "loss": 0.4121,
      "step": 31169
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4893982410430908,
      "learning_rate": 2.0771733271248706e-08,
      "loss": 0.3779,
      "step": 31170
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5085721015930176,
      "learning_rate": 2.0706178262712706e-08,
      "loss": 0.3727,
      "step": 31171
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4637324810028076,
      "learning_rate": 2.0640726755410956e-08,
      "loss": 0.3817,
      "step": 31172
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.543915867805481,
      "learning_rate": 2.057537875002291e-08,
      "loss": 0.4617,
      "step": 31173
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5205621719360352,
      "learning_rate": 2.051013424722692e-08,
      "loss": 0.384,
      "step": 31174
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5494294166564941,
      "learning_rate": 2.0444993247697997e-08,
      "loss": 0.4396,
      "step": 31175
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4697140455245972,
      "learning_rate": 2.0379955752112268e-08,
      "loss": 0.3702,
      "step": 31176
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.495991587638855,
      "learning_rate": 2.0315021761145858e-08,
      "loss": 0.4377,
      "step": 31177
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5337153673171997,
      "learning_rate": 2.0250191275470454e-08,
      "loss": 0.3681,
      "step": 31178
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5545536279678345,
      "learning_rate": 2.018546429575885e-08,
      "loss": 0.3922,
      "step": 31179
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4593477249145508,
      "learning_rate": 2.012084082268162e-08,
      "loss": 0.3613,
      "step": 31180
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5767351388931274,
      "learning_rate": 2.005632085691045e-08,
      "loss": 0.4607,
      "step": 31181
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5442885160446167,
      "learning_rate": 1.9991904399114805e-08,
      "loss": 0.3683,
      "step": 31182
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5983010530471802,
      "learning_rate": 1.9927591449960816e-08,
      "loss": 0.3954,
      "step": 31183
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.472886323928833,
      "learning_rate": 1.9863382010117947e-08,
      "loss": 0.3578,
      "step": 31184
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5741956233978271,
      "learning_rate": 1.9799276080250118e-08,
      "loss": 0.4179,
      "step": 31185
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5761057138442993,
      "learning_rate": 1.9735273661022347e-08,
      "loss": 0.4186,
      "step": 31186
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6150870323181152,
      "learning_rate": 1.9671374753099658e-08,
      "loss": 0.4453,
      "step": 31187
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5141185522079468,
      "learning_rate": 1.9607579357144855e-08,
      "loss": 0.3746,
      "step": 31188
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5594643354415894,
      "learning_rate": 1.9543887473818524e-08,
      "loss": 0.417,
      "step": 31189
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5163782835006714,
      "learning_rate": 1.9480299103781242e-08,
      "loss": 0.3647,
      "step": 31190
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4918212890625,
      "learning_rate": 1.9416814247692483e-08,
      "loss": 0.399,
      "step": 31191
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5503427982330322,
      "learning_rate": 1.935343290621172e-08,
      "loss": 0.3592,
      "step": 31192
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5217540264129639,
      "learning_rate": 1.929015507999621e-08,
      "loss": 0.4259,
      "step": 31193
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5209673643112183,
      "learning_rate": 1.9226980769700974e-08,
      "loss": 0.3647,
      "step": 31194
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5790833234786987,
      "learning_rate": 1.9163909975982166e-08,
      "loss": 0.3998,
      "step": 31195
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.548351526260376,
      "learning_rate": 1.9100942699493696e-08,
      "loss": 0.3799,
      "step": 31196
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7547543048858643,
      "learning_rate": 1.9038078940888383e-08,
      "loss": 1.0388,
      "step": 31197
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4469355344772339,
      "learning_rate": 1.8975318700819035e-08,
      "loss": 0.9572,
      "step": 31198
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.593781590461731,
      "learning_rate": 1.8912661979937353e-08,
      "loss": 0.3926,
      "step": 31199
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5252970457077026,
      "learning_rate": 1.8850108778889486e-08,
      "loss": 0.3858,
      "step": 31200
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6153390407562256,
      "learning_rate": 1.878765909832825e-08,
      "loss": 0.4384,
      "step": 31201
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5467933416366577,
      "learning_rate": 1.8725312938899786e-08,
      "loss": 0.42,
      "step": 31202
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5213643312454224,
      "learning_rate": 1.866307030125025e-08,
      "loss": 0.3927,
      "step": 31203
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4860897064208984,
      "learning_rate": 1.8600931186025796e-08,
      "loss": 0.3758,
      "step": 31204
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5598196983337402,
      "learning_rate": 1.853889559387034e-08,
      "loss": 0.4045,
      "step": 31205
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5215535163879395,
      "learning_rate": 1.8476963525427827e-08,
      "loss": 0.3995,
      "step": 31206
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6028814315795898,
      "learning_rate": 1.8415134981341065e-08,
      "loss": 0.4712,
      "step": 31207
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.552101969718933,
      "learning_rate": 1.8353409962249545e-08,
      "loss": 0.3758,
      "step": 31208
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.482011079788208,
      "learning_rate": 1.829178846879609e-08,
      "loss": 0.3954,
      "step": 31209
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.506805658340454,
      "learning_rate": 1.8230270501616854e-08,
      "loss": 0.3759,
      "step": 31210
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5676469802856445,
      "learning_rate": 1.8168856061352435e-08,
      "loss": 0.4134,
      "step": 31211
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4199186563491821,
      "learning_rate": 1.8107545148638996e-08,
      "loss": 0.3561,
      "step": 31212
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5402519702911377,
      "learning_rate": 1.8046337764110466e-08,
      "loss": 0.4122,
      "step": 31213
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6343920230865479,
      "learning_rate": 1.7985233908405232e-08,
      "loss": 0.374,
      "step": 31214
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5105259418487549,
      "learning_rate": 1.7924233582155005e-08,
      "loss": 0.3845,
      "step": 31215
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5443955659866333,
      "learning_rate": 1.7863336785991502e-08,
      "loss": 0.354,
      "step": 31216
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5402066707611084,
      "learning_rate": 1.7802543520547556e-08,
      "loss": 0.406,
      "step": 31217
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5029844045639038,
      "learning_rate": 1.7741853786453767e-08,
      "loss": 0.4054,
      "step": 31218
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6814154386520386,
      "learning_rate": 1.7681267584340744e-08,
      "loss": 0.4311,
      "step": 31219
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5150727033615112,
      "learning_rate": 1.7620784914834656e-08,
      "loss": 0.3709,
      "step": 31220
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4926772117614746,
      "learning_rate": 1.756040577856277e-08,
      "loss": 0.3944,
      "step": 31221
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.56100594997406,
      "learning_rate": 1.7500130176152376e-08,
      "loss": 0.3872,
      "step": 31222
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5371136665344238,
      "learning_rate": 1.7439958108229627e-08,
      "loss": 0.4464,
      "step": 31223
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4637972116470337,
      "learning_rate": 1.7379889575416254e-08,
      "loss": 0.3618,
      "step": 31224
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6228504180908203,
      "learning_rate": 1.731992457833731e-08,
      "loss": 0.4132,
      "step": 31225
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5009098052978516,
      "learning_rate": 1.7260063117613414e-08,
      "loss": 0.3747,
      "step": 31226
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.616321086883545,
      "learning_rate": 1.7200305193866284e-08,
      "loss": 0.8543,
      "step": 31227
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3621398210525513,
      "learning_rate": 1.7140650807714322e-08,
      "loss": 0.962,
      "step": 31228
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4689018726348877,
      "learning_rate": 1.7081099959778136e-08,
      "loss": 0.4413,
      "step": 31229
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5540474653244019,
      "learning_rate": 1.7021652650673902e-08,
      "loss": 0.4051,
      "step": 31230
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.636938452720642,
      "learning_rate": 1.6962308881020017e-08,
      "loss": 0.9495,
      "step": 31231
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4457619190216064,
      "learning_rate": 1.690306865142932e-08,
      "loss": 1.0764,
      "step": 31232
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5661810636520386,
      "learning_rate": 1.684393196251799e-08,
      "loss": 0.4299,
      "step": 31233
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.484209656715393,
      "learning_rate": 1.6784898814898865e-08,
      "loss": 0.3905,
      "step": 31234
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5971750020980835,
      "learning_rate": 1.672596920918368e-08,
      "loss": 0.4625,
      "step": 31235
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5393753051757812,
      "learning_rate": 1.6667143145985277e-08,
      "loss": 0.3662,
      "step": 31236
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5616466999053955,
      "learning_rate": 1.660842062591206e-08,
      "loss": 0.4186,
      "step": 31237
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4099318981170654,
      "learning_rate": 1.6549801649574647e-08,
      "loss": 0.3655,
      "step": 31238
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.61398184299469,
      "learning_rate": 1.6491286217579226e-08,
      "loss": 0.4173,
      "step": 31239
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5395762920379639,
      "learning_rate": 1.6432874330534198e-08,
      "loss": 0.3585,
      "step": 31240
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.546265959739685,
      "learning_rate": 1.637456598904463e-08,
      "loss": 0.4356,
      "step": 31241
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.48150634765625,
      "learning_rate": 1.631636119371449e-08,
      "loss": 0.3728,
      "step": 31242
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5137825012207031,
      "learning_rate": 1.6258259945149957e-08,
      "loss": 0.4272,
      "step": 31243
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5035669803619385,
      "learning_rate": 1.620026224395166e-08,
      "loss": 0.3544,
      "step": 31244
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5478816032409668,
      "learning_rate": 1.614236809072134e-08,
      "loss": 0.4091,
      "step": 31245
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6750755310058594,
      "learning_rate": 1.608457748605963e-08,
      "loss": 0.3661,
      "step": 31246
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.581191062927246,
      "learning_rate": 1.6026890430566044e-08,
      "loss": 0.4634,
      "step": 31247
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.631353735923767,
      "learning_rate": 1.596930692483789e-08,
      "loss": 0.3939,
      "step": 31248
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5249031782150269,
      "learning_rate": 1.5911826969474688e-08,
      "loss": 0.3952,
      "step": 31249
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6068572998046875,
      "learning_rate": 1.5854450565070402e-08,
      "loss": 0.3712,
      "step": 31250
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7583004236221313,
      "learning_rate": 1.5797177712221224e-08,
      "loss": 0.3808,
      "step": 31251
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4830416440963745,
      "learning_rate": 1.5740008411520012e-08,
      "loss": 0.3537,
      "step": 31252
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5111916065216064,
      "learning_rate": 1.5682942663561852e-08,
      "loss": 0.3874,
      "step": 31253
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4914039373397827,
      "learning_rate": 1.5625980468936264e-08,
      "loss": 0.3646,
      "step": 31254
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.558431625366211,
      "learning_rate": 1.5569121828236112e-08,
      "loss": 0.4271,
      "step": 31255
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5505083799362183,
      "learning_rate": 1.5512366742049812e-08,
      "loss": 0.379,
      "step": 31256
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5705739259719849,
      "learning_rate": 1.5455715210965783e-08,
      "loss": 0.4656,
      "step": 31257
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.491608738899231,
      "learning_rate": 1.5399167235572444e-08,
      "loss": 0.3782,
      "step": 31258
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.653989315032959,
      "learning_rate": 1.534272281645488e-08,
      "loss": 0.3954,
      "step": 31259
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.535323143005371,
      "learning_rate": 1.52863819542004e-08,
      "loss": 0.3698,
      "step": 31260
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5675599575042725,
      "learning_rate": 1.5230144649392987e-08,
      "loss": 0.4282,
      "step": 31261
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4990887641906738,
      "learning_rate": 1.5174010902615498e-08,
      "loss": 0.3903,
      "step": 31262
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8555320501327515,
      "learning_rate": 1.51179807144497e-08,
      "loss": 0.5361,
      "step": 31263
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5363370180130005,
      "learning_rate": 1.506205408547623e-08,
      "loss": 0.3622,
      "step": 31264
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.634777545928955,
      "learning_rate": 1.500623101627685e-08,
      "loss": 0.4218,
      "step": 31265
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5639904737472534,
      "learning_rate": 1.495051150742999e-08,
      "loss": 0.4034,
      "step": 31266
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7012299299240112,
      "learning_rate": 1.4894895559512957e-08,
      "loss": 0.4401,
      "step": 31267
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4663174152374268,
      "learning_rate": 1.4839383173101962e-08,
      "loss": 0.3488,
      "step": 31268
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6199829578399658,
      "learning_rate": 1.4783974348774321e-08,
      "loss": 0.4098,
      "step": 31269
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4936259984970093,
      "learning_rate": 1.4728669087104019e-08,
      "loss": 0.3761,
      "step": 31270
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4618926048278809,
      "learning_rate": 1.4673467388665042e-08,
      "loss": 0.402,
      "step": 31271
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4743496179580688,
      "learning_rate": 1.4618369254029152e-08,
      "loss": 0.3442,
      "step": 31272
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5341832637786865,
      "learning_rate": 1.4563374683768117e-08,
      "loss": 0.3808,
      "step": 31273
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4901366233825684,
      "learning_rate": 1.4508483678452589e-08,
      "loss": 0.3559,
      "step": 31274
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5050644874572754,
      "learning_rate": 1.4453696238652116e-08,
      "loss": 0.4579,
      "step": 31275
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4774173498153687,
      "learning_rate": 1.4399012364934018e-08,
      "loss": 0.3541,
      "step": 31276
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5154132843017578,
      "learning_rate": 1.4344432057865621e-08,
      "loss": 0.4052,
      "step": 31277
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5844197273254395,
      "learning_rate": 1.4289955318013138e-08,
      "loss": 0.3885,
      "step": 31278
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5803571939468384,
      "learning_rate": 1.4235582145942784e-08,
      "loss": 0.4704,
      "step": 31279
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5487631559371948,
      "learning_rate": 1.418131254221744e-08,
      "loss": 0.3827,
      "step": 31280
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4911816120147705,
      "learning_rate": 1.4127146507399991e-08,
      "loss": 0.4373,
      "step": 31281
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4610861539840698,
      "learning_rate": 1.40730840420511e-08,
      "loss": 0.3661,
      "step": 31282
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.504129409790039,
      "learning_rate": 1.4019125146733648e-08,
      "loss": 0.3985,
      "step": 31283
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5567725896835327,
      "learning_rate": 1.3965269822006078e-08,
      "loss": 0.3838,
      "step": 31284
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.558307409286499,
      "learning_rate": 1.3911518068426833e-08,
      "loss": 0.4007,
      "step": 31285
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5580962896347046,
      "learning_rate": 1.3857869886554353e-08,
      "loss": 0.3718,
      "step": 31286
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6996508836746216,
      "learning_rate": 1.3804325276943753e-08,
      "loss": 0.4092,
      "step": 31287
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5421100854873657,
      "learning_rate": 1.3750884240151252e-08,
      "loss": 0.3559,
      "step": 31288
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.828211784362793,
      "learning_rate": 1.3697546776730853e-08,
      "loss": 0.4619,
      "step": 31289
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5507159233093262,
      "learning_rate": 1.3644312887236555e-08,
      "loss": 0.3581,
      "step": 31290
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6869710683822632,
      "learning_rate": 1.3591182572219031e-08,
      "loss": 0.4664,
      "step": 31291
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5293203592300415,
      "learning_rate": 1.353815583223006e-08,
      "loss": 0.384,
      "step": 31292
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5760549306869507,
      "learning_rate": 1.3485232667819204e-08,
      "loss": 0.4139,
      "step": 31293
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5260932445526123,
      "learning_rate": 1.3432413079536022e-08,
      "loss": 0.3562,
      "step": 31294
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5524102449417114,
      "learning_rate": 1.3379697067927854e-08,
      "loss": 0.4056,
      "step": 31295
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5762276649475098,
      "learning_rate": 1.3327084633542043e-08,
      "loss": 0.3735,
      "step": 31296
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.524571180343628,
      "learning_rate": 1.3274575776923704e-08,
      "loss": 0.4272,
      "step": 31297
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4667993783950806,
      "learning_rate": 1.3222170498617959e-08,
      "loss": 0.3631,
      "step": 31298
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5220764875411987,
      "learning_rate": 1.3169868799167707e-08,
      "loss": 0.4274,
      "step": 31299
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4880298376083374,
      "learning_rate": 1.3117670679115846e-08,
      "loss": 0.4039,
      "step": 31300
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.612687349319458,
      "learning_rate": 1.3065576139003056e-08,
      "loss": 0.4366,
      "step": 31301
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5188294649124146,
      "learning_rate": 1.3013585179371125e-08,
      "loss": 0.3856,
      "step": 31302
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.601650357246399,
      "learning_rate": 1.29616978007574e-08,
      "loss": 0.4552,
      "step": 31303
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5121506452560425,
      "learning_rate": 1.2909914003701451e-08,
      "loss": 0.3735,
      "step": 31304
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5699728727340698,
      "learning_rate": 1.2858233788739516e-08,
      "loss": 0.4327,
      "step": 31305
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5165692567825317,
      "learning_rate": 1.2806657156407832e-08,
      "loss": 0.3925,
      "step": 31306
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6046583652496338,
      "learning_rate": 1.2755184107241524e-08,
      "loss": 0.4937,
      "step": 31307
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6151154041290283,
      "learning_rate": 1.2703814641774614e-08,
      "loss": 0.3924,
      "step": 31308
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5112354755401611,
      "learning_rate": 1.2652548760538897e-08,
      "loss": 0.3879,
      "step": 31309
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6589336395263672,
      "learning_rate": 1.2601386464067278e-08,
      "loss": 0.379,
      "step": 31310
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.536552906036377,
      "learning_rate": 1.2550327752890445e-08,
      "loss": 0.4354,
      "step": 31311
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5281028747558594,
      "learning_rate": 1.2499372627537975e-08,
      "loss": 0.3779,
      "step": 31312
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5142707824707031,
      "learning_rate": 1.2448521088536115e-08,
      "loss": 0.4204,
      "step": 31313
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.509090900421143,
      "learning_rate": 1.2397773136415548e-08,
      "loss": 0.4024,
      "step": 31314
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5299229621887207,
      "learning_rate": 1.2347128771700301e-08,
      "loss": 0.4403,
      "step": 31315
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4858509302139282,
      "learning_rate": 1.229658799491662e-08,
      "loss": 0.3671,
      "step": 31316
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7256165742874146,
      "learning_rate": 1.2246150806587421e-08,
      "loss": 0.4539,
      "step": 31317
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6181143522262573,
      "learning_rate": 1.2195817207237837e-08,
      "loss": 0.3636,
      "step": 31318
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6356197595596313,
      "learning_rate": 1.2145587197389675e-08,
      "loss": 0.4119,
      "step": 31319
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5497887134552002,
      "learning_rate": 1.2095460777561407e-08,
      "loss": 0.3672,
      "step": 31320
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5294476747512817,
      "learning_rate": 1.2045437948275952e-08,
      "loss": 0.4314,
      "step": 31321
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5308157205581665,
      "learning_rate": 1.1995518710050669e-08,
      "loss": 0.3661,
      "step": 31322
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5196055173873901,
      "learning_rate": 1.1945703063402925e-08,
      "loss": 0.3838,
      "step": 31323
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5534411668777466,
      "learning_rate": 1.1895991008851193e-08,
      "loss": 0.3745,
      "step": 31324
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7463804483413696,
      "learning_rate": 1.1846382546908397e-08,
      "loss": 0.3861,
      "step": 31325
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.522391676902771,
      "learning_rate": 1.179687767809079e-08,
      "loss": 0.3824,
      "step": 31326
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6565728187561035,
      "learning_rate": 1.1747476402912406e-08,
      "loss": 0.4156,
      "step": 31327
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.515352487564087,
      "learning_rate": 1.1698178721883946e-08,
      "loss": 0.3749,
      "step": 31328
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.529706358909607,
      "learning_rate": 1.1648984635517224e-08,
      "loss": 0.3739,
      "step": 31329
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6318714618682861,
      "learning_rate": 1.1599894144324054e-08,
      "loss": 0.4033,
      "step": 31330
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6464238166809082,
      "learning_rate": 1.1550907248810694e-08,
      "loss": 0.4631,
      "step": 31331
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4615861177444458,
      "learning_rate": 1.150202394948785e-08,
      "loss": 0.3541,
      "step": 31332
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5498921871185303,
      "learning_rate": 1.1453244246860674e-08,
      "loss": 0.4476,
      "step": 31333
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4702407121658325,
      "learning_rate": 1.1404568141436534e-08,
      "loss": 0.3824,
      "step": 31334
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5733487606048584,
      "learning_rate": 1.1355995633718364e-08,
      "loss": 0.4284,
      "step": 31335
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5154083967208862,
      "learning_rate": 1.1307526724212426e-08,
      "loss": 0.3562,
      "step": 31336
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5195966958999634,
      "learning_rate": 1.1259161413419427e-08,
      "loss": 0.4255,
      "step": 31337
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4974501132965088,
      "learning_rate": 1.1210899701842303e-08,
      "loss": 0.3729,
      "step": 31338
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4659934043884277,
      "learning_rate": 1.1162741589980653e-08,
      "loss": 0.4062,
      "step": 31339
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5603010654449463,
      "learning_rate": 1.1114687078335185e-08,
      "loss": 0.3876,
      "step": 31340
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4850313663482666,
      "learning_rate": 1.106673616740217e-08,
      "loss": 0.3576,
      "step": 31341
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5900402069091797,
      "learning_rate": 1.1018888857681209e-08,
      "loss": 0.4191,
      "step": 31342
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.504003882408142,
      "learning_rate": 1.0971145149666352e-08,
      "loss": 0.4112,
      "step": 31343
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.496581792831421,
      "learning_rate": 1.0923505043854975e-08,
      "loss": 0.3636,
      "step": 31344
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.615105152130127,
      "learning_rate": 1.0875968540740022e-08,
      "loss": 0.4388,
      "step": 31345
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.519713282585144,
      "learning_rate": 1.0828535640814431e-08,
      "loss": 0.3852,
      "step": 31346
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.510995626449585,
      "learning_rate": 1.0781206344571139e-08,
      "loss": 0.3895,
      "step": 31347
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4788206815719604,
      "learning_rate": 1.0733980652499753e-08,
      "loss": 0.3431,
      "step": 31348
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7563921213150024,
      "learning_rate": 1.0686858565090996e-08,
      "loss": 0.4762,
      "step": 31349
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4683953523635864,
      "learning_rate": 1.0639840082832253e-08,
      "loss": 0.3643,
      "step": 31350
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5880622863769531,
      "learning_rate": 1.0592925206213134e-08,
      "loss": 0.4204,
      "step": 31351
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5775028467178345,
      "learning_rate": 1.0546113935718804e-08,
      "loss": 0.3718,
      "step": 31352
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.2918386459350586,
      "learning_rate": 1.0499406271835543e-08,
      "loss": 0.39,
      "step": 31353
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5485154390335083,
      "learning_rate": 1.0452802215047409e-08,
      "loss": 0.3738,
      "step": 31354
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4134080410003662,
      "learning_rate": 1.0406301765837346e-08,
      "loss": 0.3473,
      "step": 31355
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5460485219955444,
      "learning_rate": 1.0359904924688301e-08,
      "loss": 0.3755,
      "step": 31356
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6089012622833252,
      "learning_rate": 1.0313611692082116e-08,
      "loss": 0.4267,
      "step": 31357
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5070778131484985,
      "learning_rate": 1.026742206849729e-08,
      "loss": 0.3739,
      "step": 31358
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5534995794296265,
      "learning_rate": 1.0221336054413445e-08,
      "loss": 0.4038,
      "step": 31359
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5255188941955566,
      "learning_rate": 1.0175353650309084e-08,
      "loss": 0.3874,
      "step": 31360
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9265165328979492,
      "learning_rate": 1.0129474856661602e-08,
      "loss": 0.4182,
      "step": 31361
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4752976894378662,
      "learning_rate": 1.008369967394507e-08,
      "loss": 0.381,
      "step": 31362
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6739405393600464,
      "learning_rate": 1.0038028102635766e-08,
      "loss": 0.4293,
      "step": 31363
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5034065246582031,
      "learning_rate": 9.99246014320665e-09,
      "loss": 0.3769,
      "step": 31364
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6745402812957764,
      "learning_rate": 9.946995796130676e-09,
      "loss": 0.8848,
      "step": 31365
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4206067323684692,
      "learning_rate": 9.901635061878579e-09,
      "loss": 0.9843,
      "step": 31366
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5287362337112427,
      "learning_rate": 9.856377940922202e-09,
      "loss": 0.4072,
      "step": 31367
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.577647089958191,
      "learning_rate": 9.81122443373117e-09,
      "loss": 0.3848,
      "step": 31368
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6614211797714233,
      "learning_rate": 9.766174540771778e-09,
      "loss": 0.4644,
      "step": 31369
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4912225008010864,
      "learning_rate": 9.721228262512538e-09,
      "loss": 0.3544,
      "step": 31370
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6375041007995605,
      "learning_rate": 9.676385599418637e-09,
      "loss": 0.425,
      "step": 31371
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7290477752685547,
      "learning_rate": 9.631646551956364e-09,
      "loss": 0.3943,
      "step": 31372
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.516396403312683,
      "learning_rate": 9.587011120589795e-09,
      "loss": 0.4082,
      "step": 31373
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.541675329208374,
      "learning_rate": 9.542479305781892e-09,
      "loss": 0.3831,
      "step": 31374
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4875212907791138,
      "learning_rate": 9.49805110799229e-09,
      "loss": 0.4161,
      "step": 31375
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5728402137756348,
      "learning_rate": 9.453726527685059e-09,
      "loss": 0.3901,
      "step": 31376
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.788222312927246,
      "learning_rate": 9.409505565317611e-09,
      "loss": 0.4381,
      "step": 31377
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.521618366241455,
      "learning_rate": 9.36538822134958e-09,
      "loss": 0.3572,
      "step": 31378
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.619846224784851,
      "learning_rate": 9.321374496238378e-09,
      "loss": 0.439,
      "step": 31379
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.532706379890442,
      "learning_rate": 9.277464390440305e-09,
      "loss": 0.373,
      "step": 31380
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.510991096496582,
      "learning_rate": 9.233657904411664e-09,
      "loss": 0.3906,
      "step": 31381
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.526936411857605,
      "learning_rate": 9.189955038605425e-09,
      "loss": 0.3672,
      "step": 31382
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5375880002975464,
      "learning_rate": 9.14635579347567e-09,
      "loss": 0.4359,
      "step": 31383
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.666709065437317,
      "learning_rate": 9.102860169474258e-09,
      "loss": 0.3968,
      "step": 31384
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6503404378890991,
      "learning_rate": 9.059468167051944e-09,
      "loss": 0.4144,
      "step": 31385
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5419325828552246,
      "learning_rate": 9.016179786659473e-09,
      "loss": 0.3786,
      "step": 31386
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4924217462539673,
      "learning_rate": 8.972995028745379e-09,
      "loss": 0.3912,
      "step": 31387
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5013270378112793,
      "learning_rate": 8.929913893758191e-09,
      "loss": 0.3876,
      "step": 31388
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5390046834945679,
      "learning_rate": 8.886936382144218e-09,
      "loss": 0.4315,
      "step": 31389
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.490934133529663,
      "learning_rate": 8.84406249434977e-09,
      "loss": 0.3422,
      "step": 31390
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5159744024276733,
      "learning_rate": 8.801292230818936e-09,
      "loss": 0.39,
      "step": 31391
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.585918664932251,
      "learning_rate": 8.758625591994696e-09,
      "loss": 0.3979,
      "step": 31392
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.548753023147583,
      "learning_rate": 8.716062578321138e-09,
      "loss": 0.421,
      "step": 31393
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.528609037399292,
      "learning_rate": 8.673603190239022e-09,
      "loss": 0.3611,
      "step": 31394
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5336798429489136,
      "learning_rate": 8.631247428187995e-09,
      "loss": 0.414,
      "step": 31395
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5692805051803589,
      "learning_rate": 8.588995292607705e-09,
      "loss": 0.3734,
      "step": 31396
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5388312339782715,
      "learning_rate": 8.546846783936691e-09,
      "loss": 0.423,
      "step": 31397
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6449075937271118,
      "learning_rate": 8.50480190261238e-09,
      "loss": 0.3419,
      "step": 31398
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7936885356903076,
      "learning_rate": 8.462860649071092e-09,
      "loss": 0.4226,
      "step": 31399
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5350967645645142,
      "learning_rate": 8.421023023745811e-09,
      "loss": 0.3611,
      "step": 31400
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.623463749885559,
      "learning_rate": 8.379289027072857e-09,
      "loss": 0.4135,
      "step": 31401
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.614373803138733,
      "learning_rate": 8.337658659484105e-09,
      "loss": 0.3646,
      "step": 31402
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5648317337036133,
      "learning_rate": 8.296131921410322e-09,
      "loss": 0.3949,
      "step": 31403
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5786495208740234,
      "learning_rate": 8.254708813284496e-09,
      "loss": 0.4011,
      "step": 31404
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5745068788528442,
      "learning_rate": 8.213389335534062e-09,
      "loss": 0.4405,
      "step": 31405
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.49189031124115,
      "learning_rate": 8.172173488588676e-09,
      "loss": 0.3606,
      "step": 31406
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5189881324768066,
      "learning_rate": 8.131061272875773e-09,
      "loss": 0.3919,
      "step": 31407
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5022664070129395,
      "learning_rate": 8.090052688821682e-09,
      "loss": 0.3647,
      "step": 31408
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5631431341171265,
      "learning_rate": 8.049147736851615e-09,
      "loss": 0.4356,
      "step": 31409
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.508266568183899,
      "learning_rate": 8.008346417389678e-09,
      "loss": 0.3543,
      "step": 31410
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.507601261138916,
      "learning_rate": 7.967648730858867e-09,
      "loss": 0.4256,
      "step": 31411
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5162216424942017,
      "learning_rate": 7.927054677682177e-09,
      "loss": 0.3821,
      "step": 31412
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6286559104919434,
      "learning_rate": 7.886564258279273e-09,
      "loss": 0.5177,
      "step": 31413
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5558385848999023,
      "learning_rate": 7.846177473070927e-09,
      "loss": 0.3622,
      "step": 31414
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6117045879364014,
      "learning_rate": 7.805894322476803e-09,
      "loss": 0.474,
      "step": 31415
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.468806266784668,
      "learning_rate": 7.765714806913238e-09,
      "loss": 0.3661,
      "step": 31416
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5249944925308228,
      "learning_rate": 7.725638926798784e-09,
      "loss": 0.4315,
      "step": 31417
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.540108323097229,
      "learning_rate": 7.685666682546444e-09,
      "loss": 0.3599,
      "step": 31418
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.520655870437622,
      "learning_rate": 7.645798074572552e-09,
      "loss": 0.4154,
      "step": 31419
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5252293348312378,
      "learning_rate": 7.606033103291222e-09,
      "loss": 0.373,
      "step": 31420
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5339689254760742,
      "learning_rate": 7.566371769113234e-09,
      "loss": 0.4296,
      "step": 31421
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5336670875549316,
      "learning_rate": 7.526814072450484e-09,
      "loss": 0.3869,
      "step": 31422
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6658304929733276,
      "learning_rate": 7.487360013713751e-09,
      "loss": 0.4616,
      "step": 31423
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4683382511138916,
      "learning_rate": 7.4480095933116e-09,
      "loss": 0.342,
      "step": 31424
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4184068441390991,
      "learning_rate": 7.408762811653703e-09,
      "loss": 0.3787,
      "step": 31425
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4412986040115356,
      "learning_rate": 7.369619669144179e-09,
      "loss": 0.3719,
      "step": 31426
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5413042306900024,
      "learning_rate": 7.330580166191592e-09,
      "loss": 0.4816,
      "step": 31427
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5565955638885498,
      "learning_rate": 7.291644303200063e-09,
      "loss": 0.3942,
      "step": 31428
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5667535066604614,
      "learning_rate": 7.252812080573712e-09,
      "loss": 0.4553,
      "step": 31429
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5049132108688354,
      "learning_rate": 7.214083498714442e-09,
      "loss": 0.36,
      "step": 31430
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5392200946807861,
      "learning_rate": 7.175458558024151e-09,
      "loss": 0.4489,
      "step": 31431
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.603198766708374,
      "learning_rate": 7.136937258903631e-09,
      "loss": 0.4091,
      "step": 31432
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5492066144943237,
      "learning_rate": 7.098519601752563e-09,
      "loss": 0.425,
      "step": 31433
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.56679368019104,
      "learning_rate": 7.060205586969515e-09,
      "loss": 0.3664,
      "step": 31434
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5110119581222534,
      "learning_rate": 7.021995214951949e-09,
      "loss": 0.4013,
      "step": 31435
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5081210136413574,
      "learning_rate": 6.983888486096213e-09,
      "loss": 0.3611,
      "step": 31436
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5576633214950562,
      "learning_rate": 6.945885400796437e-09,
      "loss": 0.4044,
      "step": 31437
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4771637916564941,
      "learning_rate": 6.90798595944675e-09,
      "loss": 0.3352,
      "step": 31438
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.508493185043335,
      "learning_rate": 6.870190162442392e-09,
      "loss": 0.4161,
      "step": 31439
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5080130100250244,
      "learning_rate": 6.83249801017305e-09,
      "loss": 0.3638,
      "step": 31440
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.594340443611145,
      "learning_rate": 6.794909503031743e-09,
      "loss": 0.4199,
      "step": 31441
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4959053993225098,
      "learning_rate": 6.757424641405941e-09,
      "loss": 0.3804,
      "step": 31442
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6253074407577515,
      "learning_rate": 6.72004342568644e-09,
      "loss": 0.4069,
      "step": 31443
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5188446044921875,
      "learning_rate": 6.682765856259599e-09,
      "loss": 0.3787,
      "step": 31444
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.540662407875061,
      "learning_rate": 6.645591933512885e-09,
      "loss": 0.4023,
      "step": 31445
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5225094556808472,
      "learning_rate": 6.608521657831546e-09,
      "loss": 0.3911,
      "step": 31446
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4641773700714111,
      "learning_rate": 6.571555029600829e-09,
      "loss": 0.4027,
      "step": 31447
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4837639331817627,
      "learning_rate": 6.534692049202651e-09,
      "loss": 0.3618,
      "step": 31448
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6426188945770264,
      "learning_rate": 6.497932717020039e-09,
      "loss": 0.4364,
      "step": 31449
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.569406509399414,
      "learning_rate": 6.4612770334338e-09,
      "loss": 0.3816,
      "step": 31450
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5438076257705688,
      "learning_rate": 6.424724998825849e-09,
      "loss": 0.4074,
      "step": 31451
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5393421649932861,
      "learning_rate": 6.388276613572553e-09,
      "loss": 0.3878,
      "step": 31452
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5221166610717773,
      "learning_rate": 6.351931878054718e-09,
      "loss": 0.3987,
      "step": 31453
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5546884536743164,
      "learning_rate": 6.315690792647599e-09,
      "loss": 0.3785,
      "step": 31454
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5323352813720703,
      "learning_rate": 6.279553357726453e-09,
      "loss": 0.4346,
      "step": 31455
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5322822332382202,
      "learning_rate": 6.243519573668755e-09,
      "loss": 0.3639,
      "step": 31456
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5698440074920654,
      "learning_rate": 6.207589440845318e-09,
      "loss": 0.3957,
      "step": 31457
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4483113288879395,
      "learning_rate": 6.1717629596314e-09,
      "loss": 0.3831,
      "step": 31458
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8468823432922363,
      "learning_rate": 6.136040130395593e-09,
      "loss": 0.422,
      "step": 31459
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4765093326568604,
      "learning_rate": 6.100420953510933e-09,
      "loss": 0.3595,
      "step": 31460
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5597169399261475,
      "learning_rate": 6.064905429346013e-09,
      "loss": 0.4341,
      "step": 31461
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4260581731796265,
      "learning_rate": 6.029493558269428e-09,
      "loss": 0.3573,
      "step": 31462
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5527790784835815,
      "learning_rate": 5.994185340647551e-09,
      "loss": 0.407,
      "step": 31463
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5089151859283447,
      "learning_rate": 5.958980776847867e-09,
      "loss": 0.3939,
      "step": 31464
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5172661542892456,
      "learning_rate": 5.923879867234528e-09,
      "loss": 0.4321,
      "step": 31465
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5009735822677612,
      "learning_rate": 5.888882612171687e-09,
      "loss": 0.3437,
      "step": 31466
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5470682382583618,
      "learning_rate": 5.8539890120223874e-09,
      "loss": 0.4209,
      "step": 31467
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5418665409088135,
      "learning_rate": 5.819199067148562e-09,
      "loss": 0.3785,
      "step": 31468
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5327156782150269,
      "learning_rate": 5.784512777911033e-09,
      "loss": 0.3991,
      "step": 31469
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5427167415618896,
      "learning_rate": 5.749930144670624e-09,
      "loss": 0.3873,
      "step": 31470
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5151097774505615,
      "learning_rate": 5.715451167784824e-09,
      "loss": 0.4087,
      "step": 31471
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6149541139602661,
      "learning_rate": 5.681075847611128e-09,
      "loss": 0.4218,
      "step": 31472
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5887441635131836,
      "learning_rate": 5.646804184505917e-09,
      "loss": 0.4418,
      "step": 31473
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4313031435012817,
      "learning_rate": 5.61263617882557e-09,
      "loss": 0.3747,
      "step": 31474
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5069597959518433,
      "learning_rate": 5.578571830924251e-09,
      "loss": 0.4259,
      "step": 31475
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5160919427871704,
      "learning_rate": 5.544611141155009e-09,
      "loss": 0.3562,
      "step": 31476
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7238085269927979,
      "learning_rate": 5.510754109869787e-09,
      "loss": 0.3931,
      "step": 31477
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.46175217628479,
      "learning_rate": 5.477000737420524e-09,
      "loss": 0.3767,
      "step": 31478
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5683962106704712,
      "learning_rate": 5.443351024155829e-09,
      "loss": 0.4264,
      "step": 31479
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.516359806060791,
      "learning_rate": 5.409804970426535e-09,
      "loss": 0.4299,
      "step": 31480
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5351475477218628,
      "learning_rate": 5.376362576580141e-09,
      "loss": 0.4306,
      "step": 31481
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4525607824325562,
      "learning_rate": 5.343023842961925e-09,
      "loss": 0.3497,
      "step": 31482
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.539928674697876,
      "learning_rate": 5.309788769919388e-09,
      "loss": 0.4285,
      "step": 31483
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5462455749511719,
      "learning_rate": 5.2766573577967e-09,
      "loss": 0.3907,
      "step": 31484
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8574409484863281,
      "learning_rate": 5.243629606938028e-09,
      "loss": 0.4597,
      "step": 31485
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5701760053634644,
      "learning_rate": 5.2107055176842115e-09,
      "loss": 0.3823,
      "step": 31486
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5569589138031006,
      "learning_rate": 5.1778850903783094e-09,
      "loss": 0.3756,
      "step": 31487
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5356837511062622,
      "learning_rate": 5.1451683253611605e-09,
      "loss": 0.3601,
      "step": 31488
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5800042152404785,
      "learning_rate": 5.112555222970273e-09,
      "loss": 0.4383,
      "step": 31489
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5209330320358276,
      "learning_rate": 5.080045783544263e-09,
      "loss": 0.3852,
      "step": 31490
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4357357025146484,
      "learning_rate": 5.04764000742175e-09,
      "loss": 0.3928,
      "step": 31491
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4900827407836914,
      "learning_rate": 5.015337894938022e-09,
      "loss": 0.3828,
      "step": 31492
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5952861309051514,
      "learning_rate": 4.983139446427254e-09,
      "loss": 0.4263,
      "step": 31493
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5501176118850708,
      "learning_rate": 4.951044662224735e-09,
      "loss": 0.3635,
      "step": 31494
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5232107639312744,
      "learning_rate": 4.919053542662422e-09,
      "loss": 0.4075,
      "step": 31495
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.479161262512207,
      "learning_rate": 4.887166088073381e-09,
      "loss": 0.3674,
      "step": 31496
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9228429794311523,
      "learning_rate": 4.855382298786238e-09,
      "loss": 0.4299,
      "step": 31497
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4714608192443848,
      "learning_rate": 4.823702175132949e-09,
      "loss": 0.3814,
      "step": 31498
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.608683705329895,
      "learning_rate": 4.7921257174399215e-09,
      "loss": 0.4306,
      "step": 31499
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.537256121635437,
      "learning_rate": 4.760652926036891e-09,
      "loss": 0.3984,
      "step": 31500
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4527708292007446,
      "learning_rate": 4.729283801248041e-09,
      "loss": 0.4157,
      "step": 31501
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4862558841705322,
      "learning_rate": 4.69801834339978e-09,
      "loss": 0.3746,
      "step": 31502
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4972234964370728,
      "learning_rate": 4.66685655281629e-09,
      "loss": 0.3864,
      "step": 31503
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5332624912261963,
      "learning_rate": 4.635798429821758e-09,
      "loss": 0.3897,
      "step": 31504
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5336790084838867,
      "learning_rate": 4.6048439747370385e-09,
      "loss": 0.4524,
      "step": 31505
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5088603496551514,
      "learning_rate": 4.573993187882986e-09,
      "loss": 0.4158,
      "step": 31506
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.588602900505066,
      "learning_rate": 4.543246069580454e-09,
      "loss": 0.4028,
      "step": 31507
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.558599591255188,
      "learning_rate": 4.512602620148077e-09,
      "loss": 0.3827,
      "step": 31508
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4948883056640625,
      "learning_rate": 4.482062839903378e-09,
      "loss": 0.397,
      "step": 31509
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5310757160186768,
      "learning_rate": 4.451626729163883e-09,
      "loss": 0.3739,
      "step": 31510
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5132949352264404,
      "learning_rate": 4.421294288243782e-09,
      "loss": 0.4113,
      "step": 31511
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4625589847564697,
      "learning_rate": 4.391065517459492e-09,
      "loss": 0.3431,
      "step": 31512
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.698622226715088,
      "learning_rate": 4.360940417122983e-09,
      "loss": 0.421,
      "step": 31513
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5781413316726685,
      "learning_rate": 4.330918987548449e-09,
      "loss": 0.382,
      "step": 31514
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4798647165298462,
      "learning_rate": 4.301001229044533e-09,
      "loss": 0.3933,
      "step": 31515
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5679287910461426,
      "learning_rate": 4.271187141924316e-09,
      "loss": 0.3814,
      "step": 31516
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5788880586624146,
      "learning_rate": 4.241476726496441e-09,
      "loss": 0.4535,
      "step": 31517
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5062716007232666,
      "learning_rate": 4.211869983067329e-09,
      "loss": 0.3956,
      "step": 31518
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.68053138256073,
      "learning_rate": 4.182366911945623e-09,
      "loss": 0.4825,
      "step": 31519
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5673089027404785,
      "learning_rate": 4.152967513436635e-09,
      "loss": 0.3463,
      "step": 31520
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5715739727020264,
      "learning_rate": 4.1236717878456734e-09,
      "loss": 0.4569,
      "step": 31521
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4583975076675415,
      "learning_rate": 4.094479735475832e-09,
      "loss": 0.3618,
      "step": 31522
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.57304847240448,
      "learning_rate": 4.0653913566313095e-09,
      "loss": 0.3914,
      "step": 31523
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6351655721664429,
      "learning_rate": 4.0364066516129785e-09,
      "loss": 0.4163,
      "step": 31524
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8776147365570068,
      "learning_rate": 4.0075256207217085e-09,
      "loss": 0.4443,
      "step": 31525
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.458476185798645,
      "learning_rate": 3.978748264256149e-09,
      "loss": 0.3662,
      "step": 31526
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5344158411026,
      "learning_rate": 3.950074582516062e-09,
      "loss": 0.4073,
      "step": 31527
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6509718894958496,
      "learning_rate": 3.921504575797874e-09,
      "loss": 0.4107,
      "step": 31528
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.608176350593567,
      "learning_rate": 3.893038244398017e-09,
      "loss": 0.3932,
      "step": 31529
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4728021621704102,
      "learning_rate": 3.86467558861181e-09,
      "loss": 0.3501,
      "step": 31530
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5917514562606812,
      "learning_rate": 3.836416608733462e-09,
      "loss": 0.3698,
      "step": 31531
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4703497886657715,
      "learning_rate": 3.808261305057181e-09,
      "loss": 0.3735,
      "step": 31532
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5865681171417236,
      "learning_rate": 3.780209677872737e-09,
      "loss": 0.4189,
      "step": 31533
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5203466415405273,
      "learning_rate": 3.752261727473228e-09,
      "loss": 0.3688,
      "step": 31534
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5324591398239136,
      "learning_rate": 3.724417454146201e-09,
      "loss": 0.4335,
      "step": 31535
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.495403528213501,
      "learning_rate": 3.6966768581836456e-09,
      "loss": 0.3787,
      "step": 31536
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5899134874343872,
      "learning_rate": 3.6690399398697783e-09,
      "loss": 0.4334,
      "step": 31537
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5341136455535889,
      "learning_rate": 3.641506699494368e-09,
      "loss": 0.3695,
      "step": 31538
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.659135103225708,
      "learning_rate": 3.6140771373405213e-09,
      "loss": 0.4483,
      "step": 31539
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.4347941875457764,
      "learning_rate": 3.5867512536946757e-09,
      "loss": 0.3451,
      "step": 31540
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5525709390640259,
      "learning_rate": 3.5595290488377175e-09,
      "loss": 0.4341,
      "step": 31541
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.567391276359558,
      "learning_rate": 3.5324105230549745e-09,
      "loss": 0.3988,
      "step": 31542
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.602117657661438,
      "learning_rate": 3.505395676626222e-09,
      "loss": 0.4413,
      "step": 31543
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.472746729850769,
      "learning_rate": 3.478484509831237e-09,
      "loss": 0.3958,
      "step": 31544
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.531950831413269,
      "learning_rate": 3.4516770229497953e-09,
      "loss": 0.4251,
      "step": 31545
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.593173623085022,
      "learning_rate": 3.4249732162594528e-09,
      "loss": 0.3808,
      "step": 31546
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.634688138961792,
      "learning_rate": 3.3983730900377654e-09,
      "loss": 0.9903,
      "step": 31547
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4083915948867798,
      "learning_rate": 3.371876644560068e-09,
      "loss": 1.0063,
      "step": 31548
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4680180549621582,
      "learning_rate": 3.3454838801016963e-09,
      "loss": 0.4032,
      "step": 31549
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.476705551147461,
      "learning_rate": 3.319194796935765e-09,
      "loss": 0.3697,
      "step": 31550
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5999780893325806,
      "learning_rate": 3.2930093953353894e-09,
      "loss": 0.461,
      "step": 31551
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5102691650390625,
      "learning_rate": 3.2669276755725733e-09,
      "loss": 0.3642,
      "step": 31552
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.681237816810608,
      "learning_rate": 3.2409496379171014e-09,
      "loss": 0.449,
      "step": 31553
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5817229747772217,
      "learning_rate": 3.2150752826387575e-09,
      "loss": 0.3898,
      "step": 31554
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4788256883621216,
      "learning_rate": 3.189304610005106e-09,
      "loss": 0.4092,
      "step": 31555
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4364725351333618,
      "learning_rate": 3.163637620284821e-09,
      "loss": 0.342,
      "step": 31556
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5907392501831055,
      "learning_rate": 3.138074313742134e-09,
      "loss": 0.4407,
      "step": 31557
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.491263508796692,
      "learning_rate": 3.1126146906446108e-09,
      "loss": 0.3661,
      "step": 31558
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5201107263565063,
      "learning_rate": 3.087258751255373e-09,
      "loss": 0.4067,
      "step": 31559
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.524878740310669,
      "learning_rate": 3.062006495836434e-09,
      "loss": 0.3795,
      "step": 31560
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5276819467544556,
      "learning_rate": 3.036857924649805e-09,
      "loss": 0.4069,
      "step": 31561
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.609097957611084,
      "learning_rate": 3.0118130379575005e-09,
      "loss": 0.3661,
      "step": 31562
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8201764822006226,
      "learning_rate": 2.9868718360193116e-09,
      "loss": 1.0068,
      "step": 31563
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.464910864830017,
      "learning_rate": 2.96203431909281e-09,
      "loss": 1.0523,
      "step": 31564
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.519343376159668,
      "learning_rate": 2.937300487435568e-09,
      "loss": 0.4542,
      "step": 31565
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5433712005615234,
      "learning_rate": 2.9126703413051573e-09,
      "loss": 0.3849,
      "step": 31566
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5054303407669067,
      "learning_rate": 2.8881438809569283e-09,
      "loss": 0.4751,
      "step": 31567
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5450105667114258,
      "learning_rate": 2.8637211066440128e-09,
      "loss": 0.3794,
      "step": 31568
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4931666851043701,
      "learning_rate": 2.8394020186206518e-09,
      "loss": 0.3816,
      "step": 31569
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5845245122909546,
      "learning_rate": 2.8151866171399754e-09,
      "loss": 0.4087,
      "step": 31570
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.500980257987976,
      "learning_rate": 2.7910749024506746e-09,
      "loss": 0.4125,
      "step": 31571
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5293656587600708,
      "learning_rate": 2.767066874804769e-09,
      "loss": 0.356,
      "step": 31572
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5514355897903442,
      "learning_rate": 2.7431625344509494e-09,
      "loss": 0.4092,
      "step": 31573
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7110844850540161,
      "learning_rate": 2.719361881636795e-09,
      "loss": 0.3876,
      "step": 31574
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.586734414100647,
      "learning_rate": 2.6956649166098856e-09,
      "loss": 0.4184,
      "step": 31575
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5681735277175903,
      "learning_rate": 2.6720716396144707e-09,
      "loss": 0.3718,
      "step": 31576
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5480552911758423,
      "learning_rate": 2.6485820508970197e-09,
      "loss": 0.4111,
      "step": 31577
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6334595680236816,
      "learning_rate": 2.6251961507006708e-09,
      "loss": 0.372,
      "step": 31578
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6115446090698242,
      "learning_rate": 2.601913939266343e-09,
      "loss": 0.4255,
      "step": 31579
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5288124084472656,
      "learning_rate": 2.578735416838285e-09,
      "loss": 0.3802,
      "step": 31580
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5367112159729004,
      "learning_rate": 2.5556605836540848e-09,
      "loss": 0.4127,
      "step": 31581
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.632341742515564,
      "learning_rate": 2.532689439955771e-09,
      "loss": 0.3935,
      "step": 31582
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.520290732383728,
      "learning_rate": 2.509821985978711e-09,
      "loss": 0.43,
      "step": 31583
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.540549635887146,
      "learning_rate": 2.4870582219627125e-09,
      "loss": 0.3575,
      "step": 31584
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.533506989479065,
      "learning_rate": 2.4643981481420334e-09,
      "loss": 0.4261,
      "step": 31585
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4818410873413086,
      "learning_rate": 2.44184176475315e-09,
      "loss": 0.3479,
      "step": 31586
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6256299018859863,
      "learning_rate": 2.41938907202921e-09,
      "loss": 0.4196,
      "step": 31587
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4415597915649414,
      "learning_rate": 2.39704007020225e-09,
      "loss": 0.3552,
      "step": 31588
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.59229576587677,
      "learning_rate": 2.3747947595054166e-09,
      "loss": 0.423,
      "step": 31589
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5648486614227295,
      "learning_rate": 2.352653140169636e-09,
      "loss": 0.3791,
      "step": 31590
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6204633712768555,
      "learning_rate": 2.3306152124236147e-09,
      "loss": 0.4816,
      "step": 31591
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5474649667739868,
      "learning_rate": 2.308680976494948e-09,
      "loss": 0.3779,
      "step": 31592
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.506718397140503,
      "learning_rate": 2.286850432613452e-09,
      "loss": 0.3925,
      "step": 31593
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.553318977355957,
      "learning_rate": 2.2651235810045024e-09,
      "loss": 0.3811,
      "step": 31594
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6026147603988647,
      "learning_rate": 2.243500421893474e-09,
      "loss": 0.3968,
      "step": 31595
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4610872268676758,
      "learning_rate": 2.2219809555046323e-09,
      "loss": 0.3529,
      "step": 31596
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5695112943649292,
      "learning_rate": 2.200565182061132e-09,
      "loss": 0.4262,
      "step": 31597
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4431260824203491,
      "learning_rate": 2.179253101783907e-09,
      "loss": 0.3593,
      "step": 31598
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5414834022521973,
      "learning_rate": 2.1580447148961125e-09,
      "loss": 0.4189,
      "step": 31599
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4972760677337646,
      "learning_rate": 2.1369400216175727e-09,
      "loss": 0.3713,
      "step": 31600
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5832209587097168,
      "learning_rate": 2.1159390221647815e-09,
      "loss": 0.4501,
      "step": 31601
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4858256578445435,
      "learning_rate": 2.095041716757562e-09,
      "loss": 0.3737,
      "step": 31602
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.79228675365448,
      "learning_rate": 2.0742481056124086e-09,
      "loss": 0.4374,
      "step": 31603
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4801310300827026,
      "learning_rate": 2.053558188944704e-09,
      "loss": 0.3634,
      "step": 31604
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5236743688583374,
      "learning_rate": 2.0329719669698324e-09,
      "loss": 0.4549,
      "step": 31605
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.515960693359375,
      "learning_rate": 2.0124894398998453e-09,
      "loss": 0.3906,
      "step": 31606
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4968448877334595,
      "learning_rate": 1.9921106079479056e-09,
      "loss": 0.4173,
      "step": 31607
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.480483055114746,
      "learning_rate": 1.971835471326067e-09,
      "loss": 0.3616,
      "step": 31608
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5829659700393677,
      "learning_rate": 1.95166403024305e-09,
      "loss": 0.3913,
      "step": 31609
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.466172218322754,
      "learning_rate": 1.931596284909798e-09,
      "loss": 0.359,
      "step": 31610
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5570862293243408,
      "learning_rate": 1.9116322355339224e-09,
      "loss": 0.4487,
      "step": 31611
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5432103872299194,
      "learning_rate": 1.891771882320814e-09,
      "loss": 0.3847,
      "step": 31612
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5463463068008423,
      "learning_rate": 1.8720152254791957e-09,
      "loss": 0.4058,
      "step": 31613
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5852913856506348,
      "learning_rate": 1.8523622652111274e-09,
      "loss": 0.3935,
      "step": 31614
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5363420248031616,
      "learning_rate": 1.8328130017231104e-09,
      "loss": 0.4186,
      "step": 31615
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5458085536956787,
      "learning_rate": 1.8133674352160957e-09,
      "loss": 0.3587,
      "step": 31616
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5103869438171387,
      "learning_rate": 1.7940255658921435e-09,
      "loss": 0.3595,
      "step": 31617
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5165667533874512,
      "learning_rate": 1.7747873939533145e-09,
      "loss": 0.4149,
      "step": 31618
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6156368255615234,
      "learning_rate": 1.7556529195972282e-09,
      "loss": 0.4079,
      "step": 31619
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.4576802253723145,
      "learning_rate": 1.736622143022615e-09,
      "loss": 0.3913,
      "step": 31620
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7865442037582397,
      "learning_rate": 1.7176950644270939e-09,
      "loss": 0.9827,
      "step": 31621
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.48711097240448,
      "learning_rate": 1.6988716840082852e-09,
      "loss": 1.0314,
      "step": 31622
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5110549926757812,
      "learning_rate": 1.6801520019593676e-09,
      "loss": 0.3871,
      "step": 31623
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4756847620010376,
      "learning_rate": 1.6615360184757401e-09,
      "loss": 0.3805,
      "step": 31624
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5210901498794556,
      "learning_rate": 1.6430237337494714e-09,
      "loss": 0.4067,
      "step": 31625
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5678527355194092,
      "learning_rate": 1.6246151479737405e-09,
      "loss": 0.3769,
      "step": 31626
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.521147608757019,
      "learning_rate": 1.6063102613395054e-09,
      "loss": 0.4176,
      "step": 31627
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5184663534164429,
      "learning_rate": 1.5881090740355042e-09,
      "loss": 0.3913,
      "step": 31628
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.118081569671631,
      "learning_rate": 1.570011586250475e-09,
      "loss": 0.3983,
      "step": 31629
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5456891059875488,
      "learning_rate": 1.5520177981742657e-09,
      "loss": 0.4151,
      "step": 31630
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.588888168334961,
      "learning_rate": 1.5341277099900632e-09,
      "loss": 0.4305,
      "step": 31631
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5726083517074585,
      "learning_rate": 1.5163413218866053e-09,
      "loss": 0.3723,
      "step": 31632
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.53556227684021,
      "learning_rate": 1.4986586340470787e-09,
      "loss": 0.4569,
      "step": 31633
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.474399209022522,
      "learning_rate": 1.4810796466546706e-09,
      "loss": 0.367,
      "step": 31634
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5000265836715698,
      "learning_rate": 1.4636043598914573e-09,
      "loss": 0.3876,
      "step": 31635
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.489538311958313,
      "learning_rate": 1.4462327739395155e-09,
      "loss": 0.3636,
      "step": 31636
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.52679443359375,
      "learning_rate": 1.428964888978701e-09,
      "loss": 0.4756,
      "step": 31637
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4660778045654297,
      "learning_rate": 1.4118007051877602e-09,
      "loss": 0.3701,
      "step": 31638
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5506927967071533,
      "learning_rate": 1.3947402227454387e-09,
      "loss": 0.4385,
      "step": 31639
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6188470125198364,
      "learning_rate": 1.377783441828262e-09,
      "loss": 0.3718,
      "step": 31640
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5488578081130981,
      "learning_rate": 1.3609303626116455e-09,
      "loss": 0.4456,
      "step": 31641
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5108892917633057,
      "learning_rate": 1.3441809852710042e-09,
      "loss": 0.3962,
      "step": 31642
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.534452199935913,
      "learning_rate": 1.327535309979533e-09,
      "loss": 0.4124,
      "step": 31643
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4999712705612183,
      "learning_rate": 1.3109933369104267e-09,
      "loss": 0.3631,
      "step": 31644
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6469414234161377,
      "learning_rate": 1.2945550662357697e-09,
      "loss": 0.4376,
      "step": 31645
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5081430673599243,
      "learning_rate": 1.2782204981243162e-09,
      "loss": 0.3968,
      "step": 31646
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5871272087097168,
      "learning_rate": 1.2619896327470405e-09,
      "loss": 0.4556,
      "step": 31647
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5968143939971924,
      "learning_rate": 1.245862470271586e-09,
      "loss": 0.3822,
      "step": 31648
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5587249994277954,
      "learning_rate": 1.2298390108655966e-09,
      "loss": 0.4084,
      "step": 31649
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5570181608200073,
      "learning_rate": 1.2139192546944955e-09,
      "loss": 0.3922,
      "step": 31650
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4981470108032227,
      "learning_rate": 1.198103201923706e-09,
      "loss": 0.3956,
      "step": 31651
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4676270484924316,
      "learning_rate": 1.1823908527186512e-09,
      "loss": 0.3547,
      "step": 31652
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.60590660572052,
      "learning_rate": 1.1667822072403134e-09,
      "loss": 0.4131,
      "step": 31653
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6507679224014282,
      "learning_rate": 1.1512772656507853e-09,
      "loss": 0.3878,
      "step": 31654
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.558395266532898,
      "learning_rate": 1.1358760281121595e-09,
      "loss": 0.4373,
      "step": 31655
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5070210695266724,
      "learning_rate": 1.120578494784308e-09,
      "loss": 0.3609,
      "step": 31656
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4946521520614624,
      "learning_rate": 1.1053846658237722e-09,
      "loss": 0.3842,
      "step": 31657
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4548875093460083,
      "learning_rate": 1.0902945413904242e-09,
      "loss": 0.3642,
      "step": 31658
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.51717209815979,
      "learning_rate": 1.0753081216385852e-09,
      "loss": 0.4064,
      "step": 31659
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5747379064559937,
      "learning_rate": 1.0604254067259067e-09,
      "loss": 0.4229,
      "step": 31660
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.494615912437439,
      "learning_rate": 1.0456463968055996e-09,
      "loss": 0.4088,
      "step": 31661
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5892398357391357,
      "learning_rate": 1.0309710920308747e-09,
      "loss": 0.3959,
      "step": 31662
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5645109415054321,
      "learning_rate": 1.0163994925538324e-09,
      "loss": 0.4115,
      "step": 31663
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4327991008758545,
      "learning_rate": 1.0019315985265731e-09,
      "loss": 0.3589,
      "step": 31664
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5340145826339722,
      "learning_rate": 9.875674100978672e-10,
      "loss": 0.4058,
      "step": 31665
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5914808511734009,
      "learning_rate": 9.733069274175944e-10,
      "loss": 0.3946,
      "step": 31666
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5151153802871704,
      "learning_rate": 9.591501506345247e-10,
      "loss": 0.4547,
      "step": 31667
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5578656196594238,
      "learning_rate": 9.45097079892987e-10,
      "loss": 0.381,
      "step": 31668
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7049987316131592,
      "learning_rate": 9.311477153406412e-10,
      "loss": 0.4334,
      "step": 31669
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4946653842926025,
      "learning_rate": 9.173020571207059e-10,
      "loss": 0.3621,
      "step": 31670
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6446223258972168,
      "learning_rate": 9.035601053786203e-10,
      "loss": 0.4646,
      "step": 31671
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4717202186584473,
      "learning_rate": 8.899218602553828e-10,
      "loss": 0.3667,
      "step": 31672
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5481758117675781,
      "learning_rate": 8.763873218931018e-10,
      "loss": 0.4732,
      "step": 31673
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4561794996261597,
      "learning_rate": 8.629564904327759e-10,
      "loss": 0.367,
      "step": 31674
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6475321054458618,
      "learning_rate": 8.496293660120725e-10,
      "loss": 0.4452,
      "step": 31675
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4553152322769165,
      "learning_rate": 8.364059487708798e-10,
      "loss": 0.3698,
      "step": 31676
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5923293828964233,
      "learning_rate": 8.232862388457552e-10,
      "loss": 0.4239,
      "step": 31677
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.483397126197815,
      "learning_rate": 8.10270236372146e-10,
      "loss": 0.3563,
      "step": 31678
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.784462332725525,
      "learning_rate": 7.973579414854993e-10,
      "loss": 0.9456,
      "step": 31679
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6840646266937256,
      "learning_rate": 7.845493543201521e-10,
      "loss": 1.0244,
      "step": 31680
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5677931308746338,
      "learning_rate": 7.71844475008221e-10,
      "loss": 0.4578,
      "step": 31681
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4839909076690674,
      "learning_rate": 7.592433036818225e-10,
      "loss": 0.3631,
      "step": 31682
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.778876543045044,
      "learning_rate": 7.46745840471963e-10,
      "loss": 0.4473,
      "step": 31683
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5128122568130493,
      "learning_rate": 7.343520855085385e-10,
      "loss": 0.3741,
      "step": 31684
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4013913869857788,
      "learning_rate": 7.220620389181143e-10,
      "loss": 0.3577,
      "step": 31685
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5022144317626953,
      "learning_rate": 7.098757008305867e-10,
      "loss": 0.3641,
      "step": 31686
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5582497119903564,
      "learning_rate": 6.977930713714109e-10,
      "loss": 0.4564,
      "step": 31687
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4896125793457031,
      "learning_rate": 6.858141506649318e-10,
      "loss": 0.4007,
      "step": 31688
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6061255931854248,
      "learning_rate": 6.739389388377149e-10,
      "loss": 0.4189,
      "step": 31689
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5240107774734497,
      "learning_rate": 6.621674360096642e-10,
      "loss": 0.3811,
      "step": 31690
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.638273000717163,
      "learning_rate": 6.504996423062349e-10,
      "loss": 0.3945,
      "step": 31691
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5886926651000977,
      "learning_rate": 6.389355578462209e-10,
      "loss": 0.3817,
      "step": 31692
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7567237615585327,
      "learning_rate": 6.274751827495263e-10,
      "loss": 1.0557,
      "step": 31693
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4133009910583496,
      "learning_rate": 6.161185171371653e-10,
      "loss": 1.0057,
      "step": 31694
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5137906074523926,
      "learning_rate": 6.048655611246012e-10,
      "loss": 0.4007,
      "step": 31695
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.579949140548706,
      "learning_rate": 5.937163148295178e-10,
      "loss": 0.3984,
      "step": 31696
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4800622463226318,
      "learning_rate": 5.82670778367378e-10,
      "loss": 0.4102,
      "step": 31697
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.651091456413269,
      "learning_rate": 5.717289518536451e-10,
      "loss": 0.3973,
      "step": 31698
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5079209804534912,
      "learning_rate": 5.608908353993414e-10,
      "loss": 0.4132,
      "step": 31699
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5324738025665283,
      "learning_rate": 5.501564291199302e-10,
      "loss": 0.3725,
      "step": 31700
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.520821213722229,
      "learning_rate": 5.395257331253234e-10,
      "loss": 0.3874,
      "step": 31701
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5058727264404297,
      "learning_rate": 5.289987475254332e-10,
      "loss": 0.3869,
      "step": 31702
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5668567419052124,
      "learning_rate": 5.185754724290614e-10,
      "loss": 0.3843,
      "step": 31703
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5025005340576172,
      "learning_rate": 5.082559079461202e-10,
      "loss": 0.3918,
      "step": 31704
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.2047836780548096,
      "learning_rate": 4.980400541820807e-10,
      "loss": 0.4309,
      "step": 31705
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6092617511749268,
      "learning_rate": 4.879279112435243e-10,
      "loss": 0.3924,
      "step": 31706
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5542503595352173,
      "learning_rate": 4.779194792348118e-10,
      "loss": 0.3988,
      "step": 31707
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5348082780838013,
      "learning_rate": 4.680147582603045e-10,
      "loss": 0.3975,
      "step": 31708
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5516149997711182,
      "learning_rate": 4.5821374842325294e-10,
      "loss": 0.4282,
      "step": 31709
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.570896029472351,
      "learning_rate": 4.4851644982357724e-10,
      "loss": 0.3673,
      "step": 31710
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6734716892242432,
      "learning_rate": 4.3892286256341787e-10,
      "loss": 0.4446,
      "step": 31711
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5076006650924683,
      "learning_rate": 4.2943298674158473e-10,
      "loss": 0.376,
      "step": 31712
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.70716392993927,
      "learning_rate": 4.200468224568877e-10,
      "loss": 0.3989,
      "step": 31713
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5226346254348755,
      "learning_rate": 4.1076436980591604e-10,
      "loss": 0.381,
      "step": 31714
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.610721230506897,
      "learning_rate": 4.0158562888525933e-10,
      "loss": 0.4376,
      "step": 31715
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4484232664108276,
      "learning_rate": 3.9251059979150686e-10,
      "loss": 0.3705,
      "step": 31716
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7075984477996826,
      "learning_rate": 3.8353928261680717e-10,
      "loss": 0.4164,
      "step": 31717
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5050818920135498,
      "learning_rate": 3.7467167745552926e-10,
      "loss": 0.378,
      "step": 31718
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.509190559387207,
      "learning_rate": 3.6590778439871134e-10,
      "loss": 0.4368,
      "step": 31719
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.485556960105896,
      "learning_rate": 3.5724760353739177e-10,
      "loss": 0.3766,
      "step": 31720
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6164257526397705,
      "learning_rate": 3.486911349614985e-10,
      "loss": 0.4352,
      "step": 31721
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5381869077682495,
      "learning_rate": 3.402383787609598e-10,
      "loss": 0.3736,
      "step": 31722
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7389875650405884,
      "learning_rate": 3.3188933502126264e-10,
      "loss": 0.4635,
      "step": 31723
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5351860523223877,
      "learning_rate": 3.2364400383122495e-10,
      "loss": 0.3801,
      "step": 31724
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5258342027664185,
      "learning_rate": 3.1550238527411346e-10,
      "loss": 0.4111,
      "step": 31725
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4752904176712036,
      "learning_rate": 3.074644794365256e-10,
      "loss": 0.369,
      "step": 31726
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5213820934295654,
      "learning_rate": 2.9953028640061776e-10,
      "loss": 0.4369,
      "step": 31727
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4868923425674438,
      "learning_rate": 2.9169980624854657e-10,
      "loss": 0.3701,
      "step": 31728
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4676703214645386,
      "learning_rate": 2.8397303906246843e-10,
      "loss": 0.3914,
      "step": 31729
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.521607756614685,
      "learning_rate": 2.7634998492120926e-10,
      "loss": 0.4041,
      "step": 31730
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5273551940917969,
      "learning_rate": 2.688306439058153e-10,
      "loss": 0.3968,
      "step": 31731
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7298020124435425,
      "learning_rate": 2.6141501609178167e-10,
      "loss": 0.3421,
      "step": 31732
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4645792245864868,
      "learning_rate": 2.541031015579343e-10,
      "loss": 0.4029,
      "step": 31733
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.474767804145813,
      "learning_rate": 2.4689490037865803e-10,
      "loss": 0.3944,
      "step": 31734
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.739072561264038,
      "learning_rate": 2.3979041263055834e-10,
      "loss": 0.977,
      "step": 31735
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4184364080429077,
      "learning_rate": 2.3278963838579972e-10,
      "loss": 0.9169,
      "step": 31736
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5306330919265747,
      "learning_rate": 2.2589257771765682e-10,
      "loss": 0.4086,
      "step": 31737
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6373687982559204,
      "learning_rate": 2.1909923069718396e-10,
      "loss": 0.3772,
      "step": 31738
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5519675016403198,
      "learning_rate": 2.1240959739543544e-10,
      "loss": 0.4183,
      "step": 31739
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.518865704536438,
      "learning_rate": 2.0582367788124502e-10,
      "loss": 0.3474,
      "step": 31740
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5152627229690552,
      "learning_rate": 1.9934147222233635e-10,
      "loss": 0.4137,
      "step": 31741
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5391433238983154,
      "learning_rate": 1.9296298048754324e-10,
      "loss": 0.3734,
      "step": 31742
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.468520164489746,
      "learning_rate": 1.8668820274236887e-10,
      "loss": 0.4168,
      "step": 31743
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5085833072662354,
      "learning_rate": 1.8051713905120615e-10,
      "loss": 0.3561,
      "step": 31744
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5499366521835327,
      "learning_rate": 1.7444978947955827e-10,
      "loss": 0.4141,
      "step": 31745
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5938299894332886,
      "learning_rate": 1.6848615408848746e-10,
      "loss": 0.3801,
      "step": 31746
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5795024633407593,
      "learning_rate": 1.6262623294127643e-10,
      "loss": 0.442,
      "step": 31747
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8211274147033691,
      "learning_rate": 1.5687002609787728e-10,
      "loss": 0.3944,
      "step": 31748
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5426005125045776,
      "learning_rate": 1.51217533618242e-10,
      "loss": 0.406,
      "step": 31749
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5487165451049805,
      "learning_rate": 1.456687555612124e-10,
      "loss": 0.3517,
      "step": 31750
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5247623920440674,
      "learning_rate": 1.4022369198452014e-10,
      "loss": 0.3938,
      "step": 31751
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4565339088439941,
      "learning_rate": 1.348823429436763e-10,
      "loss": 0.3499,
      "step": 31752
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5889298915863037,
      "learning_rate": 1.2964470849530232e-10,
      "loss": 0.4101,
      "step": 31753
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5602350234985352,
      "learning_rate": 1.2451078869268883e-10,
      "loss": 0.3511,
      "step": 31754
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6577759981155396,
      "learning_rate": 1.1948058358912663e-10,
      "loss": 0.4254,
      "step": 31755
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5105541944503784,
      "learning_rate": 1.1455409323790633e-10,
      "loss": 0.3628,
      "step": 31756
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4759188890457153,
      "learning_rate": 1.0973131768898804e-10,
      "loss": 0.396,
      "step": 31757
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5635595321655273,
      "learning_rate": 1.0501225699233175e-10,
      "loss": 0.39,
      "step": 31758
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5620994567871094,
      "learning_rate": 1.0039691119789751e-10,
      "loss": 0.4076,
      "step": 31759
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.588240385055542,
      "learning_rate": 9.588528035342493e-11,
      "loss": 0.388,
      "step": 31760
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6440521478652954,
      "learning_rate": 9.147736450554334e-11,
      "loss": 0.3836,
      "step": 31761
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.51177978515625,
      "learning_rate": 8.71731636986617e-11,
      "loss": 0.3895,
      "step": 31762
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5635170936584473,
      "learning_rate": 8.29726779782991e-11,
      "loss": 0.4391,
      "step": 31763
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5000852346420288,
      "learning_rate": 7.887590738886452e-11,
      "loss": 0.3526,
      "step": 31764
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4710882902145386,
      "learning_rate": 7.488285197254641e-11,
      "loss": 0.4107,
      "step": 31765
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6186940670013428,
      "learning_rate": 7.09935117693128e-11,
      "loss": 0.3636,
      "step": 31766
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5772947072982788,
      "learning_rate": 6.720788682135216e-11,
      "loss": 0.4346,
      "step": 31767
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.535060167312622,
      "learning_rate": 6.352597716641206e-11,
      "loss": 0.3773,
      "step": 31768
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6094721555709839,
      "learning_rate": 5.994778284446057e-11,
      "loss": 0.4153,
      "step": 31769
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5274038314819336,
      "learning_rate": 5.647330389102479e-11,
      "loss": 0.3503,
      "step": 31770
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5604537725448608,
      "learning_rate": 5.3102540342742095e-11,
      "loss": 0.4037,
      "step": 31771
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5338009595870972,
      "learning_rate": 4.983549223402939e-11,
      "loss": 0.3875,
      "step": 31772
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4568357467651367,
      "learning_rate": 4.667215960041382e-11,
      "loss": 0.3953,
      "step": 31773
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4729539155960083,
      "learning_rate": 4.3612542472981634e-11,
      "loss": 0.3519,
      "step": 31774
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7263190746307373,
      "learning_rate": 4.065664088392929e-11,
      "loss": 0.4164,
      "step": 31775
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5099135637283325,
      "learning_rate": 3.780445486323281e-11,
      "loss": 0.4039,
      "step": 31776
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6754777431488037,
      "learning_rate": 3.505598444197844e-11,
      "loss": 0.4233,
      "step": 31777
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4923274517059326,
      "learning_rate": 3.241122964792176e-11,
      "loss": 0.3626,
      "step": 31778
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4823135137557983,
      "learning_rate": 2.987019050881834e-11,
      "loss": 0.4344,
      "step": 31779
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.525744915008545,
      "learning_rate": 2.7432867050203317e-11,
      "loss": 0.3618,
      "step": 31780
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5177152156829834,
      "learning_rate": 2.5099259298722033e-11,
      "loss": 0.3869,
      "step": 31781
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4911293983459473,
      "learning_rate": 2.2869367276578957e-11,
      "loss": 0.3518,
      "step": 31782
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.511959433555603,
      "learning_rate": 2.0743191008198993e-11,
      "loss": 0.3979,
      "step": 31783
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.492691159248352,
      "learning_rate": 1.8720730516896824e-11,
      "loss": 0.3734,
      "step": 31784
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4890005588531494,
      "learning_rate": 1.6801985820436016e-11,
      "loss": 0.3911,
      "step": 31785
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5756508111953735,
      "learning_rate": 1.4986956941021036e-11,
      "loss": 0.3844,
      "step": 31786
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4976223707199097,
      "learning_rate": 1.327564389752567e-11,
      "loss": 0.4008,
      "step": 31787
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6102675199508667,
      "learning_rate": 1.1668046707713487e-11,
      "loss": 0.3993,
      "step": 31788
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5813350677490234,
      "learning_rate": 1.0164165386017388e-11,
      "loss": 0.4163,
      "step": 31789
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.547007441520691,
      "learning_rate": 8.763999951311165e-12,
      "loss": 0.3738,
      "step": 31790
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.664136528968811,
      "learning_rate": 7.467550415807267e-12,
      "loss": 0.9757,
      "step": 31791
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4351414442062378,
      "learning_rate": 6.2748167928283755e-12,
      "loss": 1.0221,
      "step": 31792
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5422388315200806,
      "learning_rate": 5.185799096807387e-12,
      "loss": 0.401,
      "step": 31793
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4265910387039185,
      "learning_rate": 4.200497337736309e-12,
      "loss": 0.366,
      "step": 31794
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4722626209259033,
      "learning_rate": 3.318911525607149e-12,
      "loss": 0.4074,
      "step": 31795
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5388723611831665,
      "learning_rate": 2.5410416693016913e-12,
      "loss": 0.3567,
      "step": 31796
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5807336568832397,
      "learning_rate": 1.8668877777017204e-12,
      "loss": 0.4039,
      "step": 31797
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5612895488739014,
      "learning_rate": 1.2964498585787967e-12,
      "loss": 0.3836,
      "step": 31798
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5067535638809204,
      "learning_rate": 8.297279152635896e-13,
      "loss": 0.4417,
      "step": 31799
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5608527660369873,
      "learning_rate": 4.667219555276603e-13,
      "loss": 0.3827,
      "step": 31800
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4469521045684814,
      "learning_rate": 2.0743198159145493e-13,
      "loss": 0.3717,
      "step": 31801
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4815075397491455,
      "learning_rate": 5.1857995675419495e-14,
      "loss": 0.4467,
      "step": 31802
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5450313091278076,
      "learning_rate": 0.0,
      "loss": 0.987,
      "step": 31803
    },
    {
      "epoch": 1.0,
      "step": 31803,
      "total_flos": 5.636373687662805e+18,
      "train_loss": 0.01061394915255991,
      "train_runtime": 2408.0285,
      "train_samples_per_second": 1690.511,
      "train_steps_per_second": 13.207
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 31803,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 5.636373687662805e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}